The new study, published in the journal Machine Learning by Chandreyee Bhowmick and Xenofon Koutsoukos of Vanderbilt’s Institute of Software Integrated Systems, tackles a problem that has long haunted peer-to-peer machine learning: what happens when some of the workers in a decentralized network are adversaries bent on poisoning the shared model? In a peer-to-peer setting, there is no central coordinator to vet incoming updates. Each device, or worker, exchanges model parameters only with its immediate neighbors on a communication graph. If even a handful of those neighbors are compromised, they can inject corrupted parameters that drag everyone’s model toward garbage, a scenario known in the field as a Byzantine attack, named after the Byzantine Generals Problem in distributed computing.

The Vanderbilt team’s answer is a resilient adaptive aggregation scheme built around a deceptively simple idea: encourage similarity among honest workers. Rather than treating all neighbor contributions equally, each worker solves an optimization problem that assigns weights to its neighbors’ model parameters, favoring those whose learning behavior resembles its own. The weights emerge from a principled formulation rather than hand-tuned heuristics, and the optimization is designed so that no worker ever needs to see another worker’s private data. Instead, each worker evaluates its neighbors’ models against its own local dataset, computing losses that reflect how well a neighbor’s model performs on data it was never trained on. This preserves privacy while still giving the aggregation step the information it needs to distinguish helpful peers from hostile ones.

The technical machinery matters here. In each round of training, a worker receives the current model parameters of its neighbors and blends them into a weighted sum, with the weights determined by solving a constrained optimization that balances fitting the local objective against staying close to the collective consensus. The formulation effectively learns, on the fly, which neighbors are pulling in the same direction and which are outliers. Adversarial workers, whose parameters are crafted to mislead rather than to learn, tend to produce models whose behavior diverges sharply from that of honest peers, and the weighting scheme naturally down-weights them. Because the weights are recomputed adaptively as training proceeds, the method can track changing conditions, including the non-convex loss landscapes that arise in deep learning, where standard convergence arguments often break down.

What sets this work apart from earlier Byzantine-resilient approaches is the combination of three hard conditions at once: non-convex loss functions, non-iid data distributions, and a fully decentralized topology. Most real-world deployments face all three. Data on different devices is rarely identically distributed; a hospital’s patient records, a phone’s photo library, and a factory’s sensor logs all look wildly different. Non-iid data makes it harder to tell a malicious outlier from an honest worker that simply has unusual data, since both may produce parameters that deviate from the crowd. Non-convex losses, characteristic of neural networks, mean the loss surface is riddled with local minima and saddle points, complicating both the algorithm design and the mathematical analysis of whether the method actually works.

And the authors do provide such analysis. Their theoretical results establish two key guarantees for honest workers. First, the workers’ model parameters reach consensus, meaning that despite the presence of adversaries and the heterogeneity of their data, the honest devices converge to agreement on a shared model. Second, the gap between the honest workers’ parameters and their respective optimal values remains bounded, and crucially, that bound is expressed as a function of a small number of hyperparameters and the variance of the non-iid data distribution across the network. In plain terms, the more heterogeneous the data, the looser the guarantee, which is an honest and interpretable characterization rather than an idealized claim that assumes away the messiness of real deployments.

The empirical side of the study puts those guarantees to the test across three classification tasks, drawing on widely used benchmark datasets including human activity recognition from smartphone sensors, the MNIST handwritten digit collection, the Spambase email dataset, and CIFAR image data. The experiments span multiple adversarial scenarios and attack models, simulating networks in which a fraction of workers behave maliciously in different ways. Across these settings, the proposed adaptive aggregation method consistently improved the test accuracy achieved by honest workers compared with state-of-the-art resilient aggregation techniques. The improvement is meaningful in practice: in adversarial distributed learning, the difference between a defense that merely limits damage and one that preserves high accuracy can determine whether a system is deployable at all.

The implications reach well beyond benchmark datasets. Decentralized, peer-to-peer learning is attractive for settings where a central server is impractical, untrusted, or simply absent: fleets of autonomous vehicles coordinating in real time, swarms of drones, industrial IoT networks, smart city infrastructure, and healthcare consortia where no single institution can legally pool patient data. In such environments, resilience is not optional. A connected vehicle network in which one compromised node can poison the collective perception model is a safety hazard, not just a security nuisance. By removing the single point of failure that plagues federated learning and simultaneously hardening the network against Byzantine participants, the new approach sketches a blueprint for collaborative AI that is both decentralized and trustworthy.

The privacy dimension deserves equal emphasis. The scheme’s design ensures that workers never share raw data; the only information exchanged is model parameters, and even the loss evaluations that guide the aggregation weights are computed locally, with each worker testing neighbor models against its own private dataset. This stands in contrast to approaches that require sharing gradients or statistics that can leak information about training data. Combined with the elimination of a central aggregation server, the method reduces the number of parties that must be trusted, a shift that security researchers often describe as moving from trusting a single authority to trusting a protocol.

Challenges remain before such systems see widespread adoption. The optimization required to compute aggregation weights adds computational overhead on each device, and the theoretical bounds, while informative, depend on hyperparameters that practitioners must tune. The convergence guarantees also assume a certain network structure and adversary budget, and real deployments may face adversaries that adapt their strategies over time. Still, the work represents a notable step forward in a research area that sits at the intersection of machine learning, distributed systems, and cybersecurity. As AI models increasingly live at the edge, on phones, vehicles, sensors, and medical devices, the question is no longer whether decentralized learning will matter, but whether it can be made safe. This study offers a rigorous, empirically validated answer to that question, showing that a network of peers, even one infiltrated by adversaries, can still learn well, provided its members know how to weigh each other’s advice.

Subject of Research: Byzantine-resilient adaptive aggregation for peer-to-peer distributed machine learning under non-convex losses and non-iid data

Article Title: Similarity-Promoting Resilient Adaptive Aggregation in Peer-to-Peer Machine Learning

Article References: Bhowmick, C., & Koutsoukos, X. (2026). Similarity-Promoting Resilient Adaptive Aggregation in Peer-to-Peer Machine Learning. Machine Learning, 115(10), Article 223. https://doi.org/10.1007/s10994-026-07162-3

Image Credits: AI Generated

DOI: 10.1007/s10994-026-07162-3

Keywords: peer-to-peer machine learning, decentralized learning, Byzantine resilience, adaptive aggregation, federated learning, non-iid data, non-convex optimization, distributed machine learning, adversarial attacks, data privacy, consensus, resilient learning

Secure multi-party computation is a branch of cryptography in which several parties jointly compute a function over their private inputs without revealing those inputs to one another. In the quantum version of this problem, quantum states and quantum operations supplement or replace classical techniques to provide security guarantees rooted in the laws of physics rather than in assumptions about computational hardness. Multiplication is a particularly important target operation: once parties can multiply their secret values securely, they can build a wide range of more elaborate privacy-preserving applications, from secure auctions and electronic voting to privacy-preserving machine learning and distributed statistical analysis. The new protocol addresses multiplication directly, allowing multiple participants to combine their secret inputs into a shared product while keeping every individual input hidden.

The central technical innovation lies in the protocol’s use of the multiplicative homomorphic property of the RSA encryption algorithm. Homomorphic encryption allows arithmetic to be performed directly on encrypted data: a party can encrypt a value, and the structure of the encryption ensures that combining ciphertexts corresponds to combining the underlying plaintexts. In the multiplicative case, the product of two ciphertexts decrypts to the product of the two plaintexts. This property means participants can contribute encrypted shares of their secrets, the encrypted shares can be multiplied together, and only the final decrypted result reveals anything about the combined product—never the individual contributions. By anchoring the protocol in RSA’s well-understood multiplicative homomorphism, the authors inherit a mature cryptographic foundation while layering quantum techniques on top for input protection and verification.

What distinguishes this work from earlier quantum secure multiplication protocols is its verifiability. In many existing schemes, participants simply have to trust that the final result is correct. If a malicious participant injects a corrupted share, or if noise and error creep into the process, the output may be silently wrong. The new protocol allows every participant to check the integrity of the final computation result and to detect any error affecting correctness, whether that error was introduced intentionally by a cheater or unintentionally by some failure in the process. Verification transforms the protocol from a trust-based arrangement into an auditable one, which is essential for any realistic deployment involving parties who may have conflicting interests.

The dynamic aspect of the protocol is equally significant. Real-world collaborations are rarely static: organizations join consortiums, employees leave companies, and partners withdraw from joint ventures. A secure multi-party protocol that must be restarted from scratch every time the participant list changes would be impractically rigid. The new scheme supports dynamic updates, meaning participants can be added or removed while the computation’s security guarantees are preserved. This flexibility, however, creates a dangerous loophole that the authors explicitly confront: a participant who is being revoked has a clear incentive to cheat during the update process, corrupting the computation on the way out. The protocol’s verification mechanism is specifically designed to detect deceptive behavior by revoked participants during these dynamic updates, closing an attack window that earlier dynamic protocols left open.

The authors describe this threat as a dishonest revocation attack, and their security analysis demonstrates that the protocol resists it alongside a series of other typical external and internal attacks. External attacks, in the quantum setting, include eavesdropping attempts in which an outsider tries to extract information from the quantum states exchanged between honest participants; the protocol’s quantum components are designed so that such interference leaves detectable traces. Internal attacks are subtler and often more damaging, since they come from participants who hold legitimate credentials but choose to deviate from the protocol to learn others’ inputs or to bias the result. By combining homomorphic encryption with verification checks, the protocol ensures that neither class of adversary can compromise either the privacy of the inputs or the correctness of the product without being caught.

Efficiency matters as much as security in this domain, because quantum protocols can impose heavy computational and communication burdens. The authors report that their scheme achieves relatively low computational costs compared with existing multiplication protocols, making it a more practical candidate for real applications. The reliance on classical RSA homomorphic operations for the arithmetic core, rather than on expensive quantum computations for every step, helps keep the overhead manageable, while quantum resources are deployed where they add the most security value. The result, the authors argue, is a protocol that offers enhanced practicality and meaningful security guarantees at a computational price that realistic deployments could afford.

The work builds on a substantial body of prior research in quantum secure multi-party computation. Earlier protocols have addressed secure summation using single photons, quantum Fourier transforms, Grover’s search algorithm and mutually unbiased bases, and secure multiplication has been explored through secret sharing and hybrid quantum-classical approaches. Previous work by members of the same team introduced a (k, n)-threshold dynamic quantum secure multiparty multiplication protocol and a verifiable threshold quantum secure multiparty summation protocol, and the present study extends that lineage by adding verifiable result integrity to the dynamic multiplication setting through homomorphic encryption. The broader field also draws on foundational results in quantum state determination, secret sharing with d-level systems, and analyses of practical attacks such as Trojan-horse attacks on quantum communication systems, all of which inform the threat model the new protocol is designed to survive.

The implications extend beyond the immediate technical contribution. As quantum computers edge closer to threatening classical public-key cryptography, and as organizations increasingly need to compute jointly over sensitive data—financial positions, medical records, proprietary models—protocols that combine quantum security with classical efficiency are likely to attract growing attention. A verifiable, dynamic multiplication protocol offers a building block for such systems: it demonstrates that participant churn need not be a security liability, and that even a departing adversary with every reason to cheat can be prevented from corrupting a shared computation. The research was supported by the National Natural Science Foundation of China, and the authors declare no conflict of interest. While laboratory-scale quantum networks and practical homomorphic quantum deployments remain works in progress, this protocol represents a concrete step toward secure multi-party computation that is simultaneously quantum-resistant in its guarantees, flexible in its membership, and honest in its arithmetic.

Subject of Research: A verifiable dynamic quantum secure multi-party multiplication protocol based on RSA homomorphic encryption that resists cheating by revoked participants.

Article Title: Verifiable dynamic quantum secure multi-party multiplication protocol based on homomorphic encryption

Article References: Li, F., Li, R., Sun, Y., & Zhu, S. (2026). Verifiable dynamic quantum secure multi-party multiplication protocol based on homomorphic encryption. Quantum Information Processing, 25(10), Article 317. https://doi.org/10.1007/s11128-026-05347-2

Image Credits: AI Generated

DOI: 10.1007/s11128-026-05347-2

Keywords: quantum cryptography, secure multi-party computation, homomorphic encryption, RSA, verifiability, dynamic protocols, multiplication protocol, dishonest revocation attack, quantum information, privacy-preserving computation, Quantum Information Processing, Hefei University of Technology

The work, published in Nature Physics by Kaixin Ji, Yu Zhao, Ce Shen, Yidun Wan and Ling-Yan Hung, draws on some of the deepest ideas in modern condensed matter theory. The authors’ strategy does not start from spins or magnets at all. Instead, they engineer the boundary conditions of three-dimensional topological orders, exotic phases of matter whose excitations, called anyons, can carry quantum statistics that are neither bosonic nor fermionic. These topological orders are described concretely by string-net models, exactly soluble constructions introduced by Michael Levin and Xiao-Gang Wen in 2005, in which the vacuum is pictured as a tangle of fluctuating strings whose allowed patterns are dictated by algebraic data known as a fusion category.

The key innovation lies in how the critical points are created. In a topological phase, certain anyon types can undergo condensation, a process analogous to the condensation of a Bose-Einstein condensate, in which the anyon becomes part of the vacuum and other excitations are reorganized accordingly. When a single set of anyons condenses, the system typically flows from one gapped topological phase to another. The researchers instead arranged for non-commuting anyons to condense in a carefully balanced, commensurate fashion, meaning that two or more condensation channels that cannot coexist in an ordinary gapped phase are forced into competition. The tug-of-war between these incompatible orders prevents the system from settling into any gapped phase, and the resulting critical points flow in the infrared limit to conformal field theories. By tuning the relative weights of the competing condensates, the algorithm generates a lattice Hamiltonian whose low-energy behavior is precisely the desired conformal theory.

The machinery relies on a holographic device known as the strange correlator, a quantity computed as a three-dimensional path integral that maps the boundary lattice model onto the bulk topological order. In this picture, the two-dimensional critical model lives on the boundary of the three-dimensional string-net system, and the algebraic rules governing anyon fusion in the bulk translate directly into the interaction terms of the boundary model. The critical couplings, the parameter values at which the phase transitions occur, are encoded exactly in algebraic data associated with the string-net construction, specifically in the Frobenius algebras that specify which anyons condense. This means that instead of laboriously scanning parameter space numerically to hunt for critical points, physicists can read off where the transitions happen from the underlying category theory, a level of precision control that is rare in the study of strongly correlated systems.

The practical payoff is an infinite family of critical lattice models. The authors demonstrate that their procedure recovers known conformal field theories that preserve the so-called Haagerup symmetries, exotic non-invertible symmetries named after the mathematician Uffe Haagerup, whose fusion categories have intrigued both mathematicians and physicists since the 1990s. Haagerup-symmetric theories have become a testing ground for the emerging theory of categorical symmetry, in which ordinary symmetry groups are replaced by richer algebraic structures. Critical lattice models realizing these symmetries had been proposed only recently, and the new algorithm reproduces them as a special case of a much more general construction, providing independent confirmation of earlier numerical work that had reported evidence for Haagerup conformal field theories in tensor network calculations.

More strikingly, the factory does not merely recycle known results. Among the models it generates, the researchers identified three previously unknown candidate conformal field theories, critical points that had never been observed or catalogued before. These discoveries suggest that the space of two-dimensional conformal field theories is far more densely populated by accessible lattice realizations than the traditional, largely ad hoc methods of statistical mechanics had revealed. Historically, finding a lattice model for a given critical phenomenon was a matter of insight and luck, from Onsager’s solution of the Ising model to the Ashkin-Teller models studied in the early 1980s. The new algorithm replaces that serendipity with a recipe: choose a fusion category, select competing condensable algebras, and compute the resulting boundary model and its phase diagram.

The numerical verification of the construction is itself technically notable. The team developed symmetry-preserving tensor network algorithms to map out the phase diagrams of their models, coloring the parameter space by the numerically determined central charge, a fundamental invariant of a conformal field theory that measures the number of its degrees of freedom. In the phase diagrams, regions corresponding to different condensed anyon orders meet along critical lines and surfaces, and the interpolation between multiple competing condensates can be visualized in ternary diagrams representing three-condensate mixtures. The agreement between the predicted critical couplings extracted from the algebraic data and the numerical scans provides a stringent consistency check of the entire framework, and the MATLAB code and source data used to regenerate the phase diagrams have been made available with the paper.

The broader implications extend beyond two-dimensional statistical mechanics. Conformal field theories occupy a central role in high-energy theoretical physics as well, appearing as limits of quantum field theories, as building blocks of string theory, and through the AdS/CFT correspondence as dual descriptions of quantum gravity. A systematic method for discretizing conformal field theories onto lattices offers a potential route to studying them with the numerical tools of condensed matter, including tensor networks and quantum simulation. The authors and other researchers in the field have also drawn connections to topological holography and the idea that symmetries themselves can be understood as shadows of topological order, suggesting that the factory could illuminate how generalized, non-invertible symmetries emerge at quantum critical points.

The work also raises tantalizing prospects for classification. One of the great unsolved problems in theoretical physics is to classify all possible conformal field theories, a task that has proved formidable even in two dimensions where the machinery is most powerful. By establishing a structured scheme in which critical theories arise from combinatorial algebraic data, the conformal field theory factory provides a framework for discovering and potentially organizing these theories in families. If every entry in a suitable catalogue of fusion categories and condensable algebras yields a critical model, physicists may be able to enumerate, or at least systematically sample, far more of the landscape of critical behavior than ever before. For a field that has spent half a century stitching together critical phenomena one painstaking example at a time, the prospect of a factory that produces them by the dozen marks a genuine shift in method, and the three brand-new candidate theories that emerged from its first run hint at how much of that landscape still lies unexplored.

Subject of Research: An algorithm generating two-dimensional critical lattice models from competing anyon condensation in three-dimensional topological orders

Article Title: An algorithm to generate two-dimensional critical lattice models using competing anyon condensation

Article References: Ji, K., Zhao, Y., Shen, C., Wan, Y., & Hung, L.-Y. (2026). An algorithm to generate two-dimensional critical lattice models using competing anyon condensation. Nature Physics. https://doi.org/10.1038/s41567-026-03438-6

Image Credits: AI Generated

DOI: 10.1038/s41567-026-03438-6

Keywords: conformal field theory, anyon condensation, topological order, string-net models, critical phenomena, lattice models, Haagerup symmetry, phase transitions, fusion categories, tensor networks, categorical symmetry, theoretical physics

The research, published in Multimedia Tools and Applications by Melisa Alara Ozuberk and Ilkay Cinar, delivers one of the first head-to-head evaluations of three leading real-time open-vocabulary detectors: YOLO-World, its successor YOLO-World v2, and YOLOE. Rather than benchmarking accuracy alone, the authors designed their experiments around three scenarios that mirror how these systems are actually deployed: zero-shot inference on entirely new datasets, sensitivity to variations in the textual prompts that steer detection, and fine-tuning on domain-specific data followed by tests of whether open-vocabulary generalization survives. The evaluation spans three datasets with deliberately different characteristics—the classic VOC2012 segmentation subset, the HomeObjects-3K indoor detection dataset, and the demanding KITTI autonomous driving benchmark.

The zero-shot results reveal a striking dependence on domain. The highest performance was achieved on HomeObjects-3K, where YOLO-World v2 reached 0.443 mAP@0.5:0.95, a metric that rewards both accurate localization and correct classification across a range of overlap thresholds. KITTI, by contrast, produced the weakest results across all three models, a consequence of domain shift: the driving imagery, with its unusual viewpoints, small distant objects, and harsh lighting conditions, differs substantially from the data distributions these models encountered during pre-training. On the VOC dataset, YOLOE claimed the highest zero-shot accuracy at 0.310 mAP@0.5:0.95, outperforming both YOLO-World variants on that benchmark, although this advantage came with a caveat that emerged clearly in the timing analysis.

That caveat is speed. YOLO-World v2 proved to offer the best overall balance between accuracy and throughput, sustaining between 20 and 30 frames per second—comfortably real-time for many applications. YOLOE, despite its stronger zero-shot accuracy on VOC, exhibited lower inference speed in some configurations, a trade-off that could prove decisive in latency-sensitive settings such as autonomous navigation or live video analytics. The YOLO-World family also benefited from an embedding cache mechanism, which pre-computes text embeddings for the prompt vocabulary and reuses them across frames. Latency analysis with increasing prompt counts showed that this design keeps inference efficient even as the number of textual categories grows, an architectural advantage that becomes more valuable the richer the vocabulary deployed in production.

Perhaps the most practically important finding concerns prompt robustness—the question of how much detection quality degrades when the words fed to the model change. The authors constructed four categories of prompts for each dataset: base prompts, attribute prompts that add descriptive modifiers, longer descriptive prompts, and noisy prompts containing degraded or perturbed language. The results showed measurable performance drops under noisy prompt conditions for all models, confirming that open-vocabulary detectors are not immune to the fragility of language interfaces that has been documented across the broader vision-language literature. However, YOLO-World v2 maintained better stability across these variations than its competitors, suggesting that its training recipe or text-encoding pathway confers a degree of resilience that practitioners should weigh when deploying systems in the hands of non-expert users who cannot be relied upon to craft optimal prompts.

Fine-tuning delivered the expected gains but also exposed an uncomfortable truth about what adaptation costs. After fine-tuning on each dataset, mAP scores improved for all three models, demonstrating that standard transfer learning techniques remain effective when open-vocabulary detectors are specialized to a target domain. But the fine-tuned models showed zero performance on some unseen categories—classes that were never part of the fine-tuning data. This is precisely the failure mode that open-vocabulary detection is supposed to prevent, and its appearance after adaptation indicates that the boundary between open and closed vocabulary is thinner than the field often assumes.

The divergence between the model families was especially pronounced here. The YOLO-World family retained some of its open-vocabulary generalization ability after fine-tuning, continuing to respond to textual prompts for categories outside the training set. YOLOE, under the fine-tuning protocol adopted in the study, exhibited closed-set-like behavior: its predictions remained insensitive to the evaluated prompt variations, effectively behaving as if the text interface had been switched off and the model had reverted to a conventional fixed-category detector. For teams choosing between these architectures, the implication is significant—fine-tuning YOLOE may buy accuracy on known classes at the price of the very flexibility that motivated choosing an open-vocabulary model in the first place.

The study’s methodology reflects a growing recognition that evaluation practices in this field have been too narrow. The authors note that existing research has focused mainly on accuracy metrics while prompt robustness, unseen class generalization, and computational costs are rarely assessed together. By combining confusion-matrix-based analysis, standard detection metrics such as mAP at multiple intersection-over-union thresholds, and latency profiling under varying prompt counts, the work offers a template for more honest benchmarking. The datasets themselves are all publicly available—VOC2012 from the PASCAL repository, KITTI from the KITTI Vision Benchmark Suite, and HomeObjects-3K through its original repository—making the evaluation pipeline reproducible by other groups.

The broader context makes these findings timely. Open-vocabulary detection builds on a lineage that runs from the original YOLO real-time detector through open-set recognition and open-world detection to caption-supervised methods and the CLIP-style vision-language models that supply the text-image alignment these detectors depend on. YOLO-World, introduced in 2024, brought this capability to real-time speeds, and YOLOE pushed the concept further with its “see anything” design. Applications documented in the literature now span automatic image annotation, number plate recognition, wildlife monitoring, medical imaging, underwater fish counting, robotic navigation, and anomaly detection in surveillance—domains where the ability to name new categories without retraining is transformative.

For practitioners, the study’s bottom line is that model selection should be a multi-dimensional decision. Accuracy, inference speed, prompt robustness, and unseen class generalization form a set of trade-offs that no single model dominates. YOLO-World v2 emerges as the most balanced option, combining competitive accuracy, real-time throughput, an efficient embedding cache, and the strongest prompt stability. YOLOE offers the best zero-shot accuracy in some settings but pays in speed and, critically, appears to surrender its open-vocabulary character when fine-tuned. As these systems move from research demos into safety-relevant deployments—self-driving perception, medical triage, industrial inspection—the lesson of this comparative study is that the questions worth asking about a detector extend well beyond its leaderboard score, reaching into how it behaves when the words change, the domain shifts, and the training data runs out.

Subject of Research: Comparative evaluation of prompt robustness, fine-tuning, and generalization in open-vocabulary object detection models

Article Title: Prompt robustness, fine-tuning, and Generalization in open-vocabulary object detection: a comparative study of YOLO-World, YOLO-World v2 and YOLOE

Article References: Ozuberk, M. A., & Cinar, I. (2026). Prompt robustness, fine-tuning, and Generalization in open-vocabulary object detection: a comparative study of YOLO-World, YOLO-World v2 and YOLOE. Multimedia Tools and Applications, 85(10), Article 767. https://doi.org/10.1007/s11042-026-21928-w

Image Credits: AI Generated

DOI: 10.1007/s11042-026-21928-w

Keywords: open-vocabulary detection, YOLO-World, YOLOE, zero-shot object detection, prompt sensitivity, fine-tuning, computer vision, mAP, inference latency, domain shift, vision-language models, object recognition

The research, published in the journal Cybersecurity by a team from the Institute of Information Engineering at the Chinese Academy of Sciences and the University of Chinese Academy of Sciences, addresses a fundamental dilemma in provenance-based security analysis. Provenance graphs record causal interactions among system entities—processes, files, and network sockets—allowing analysts to reconstruct how an intrusion propagated from initial compromise through privilege escalation, lateral movement, and data exfiltration. But the sheer scale of these graphs makes analysis slow and expensive, and existing compression techniques often destroy the very evidence they are meant to protect.

The core insight behind CSCProv is that compression decisions must respect two dimensions at once. Structure-only methods identify repetitive topological patterns and merge them, but this can bury a critical anomaly inside a haystack of benign activity. The researchers illustrate the problem with a network reconnaissance scenario: an attacker scanning roughly 250 hosts with a tool like Nmap creates a large fan-out structure in which nearly all destination sockets look topologically identical. If one host is actually compromised, its socket spawns a distinct causal chain—receiving data and spawning a reverse shell—that a structure-driven compressor is likely to collapse into the aggregate, obscuring the attack lineage entirely.

Semantic-only compression suffers from the opposite weakness. In a Living-off-the-Land style attack against an Nginx web server drawn from the DARPA E3 CADETS dataset, a malicious backdoor script shares the same /var/www/html/ path prefix as thousands of benign web resources. A semantics-based compressor that merges nodes with similar attributes would homogenize the backdoor with ordinary file accesses, irrecoverably blending the malicious event into benign noise. Attackers deliberately exploit both blind spots, camouflaging malicious actions within legitimate binaries and repetitive system behavior.

CSCProv resolves this tension with a novel metric the authors call Causal-Semantic Consistency, or CSC. The framework computes two complementary representations for every node in a provenance graph. A structural embedding, produced by a topology-dominant graph convolutional network, captures each entity’s dependency patterns—its degree statistics, local connectivity, and node type—while deliberately excluding behavioral attributes. A semantic embedding, generated by a domain-adapted SecBERT language model, encodes the textual content of audit events, including command-line arguments, file paths, network metadata, and event types, which are then aggregated into node-level behavioral representations.

The CSC score itself is computed as the cosine similarity between joint embeddings that fuse both dimensions over each node’s localized k-hop provenance context. Two nodes are merged only when their combined score exceeds a configurable threshold, meaning they must be consistent in both dependency structure and behavioral semantics. Candidate pairs are processed in descending order of similarity, with merged nodes replacing their constituents and duplicate edges consolidated while preserving event counts. The researchers analyze the computational cost of the pipeline and note that restricting candidate generation by entity type and temporal context substantially reduces the number of pairwise comparisons in practice.

Evaluation on the DARPA Transparent Computing Engagement 3 dataset, spanning the CADETS, THEIA, CLEARSCOPE, FIVEDIRECTIONS, and TRACE subsets across FreeBSD, Linux, Android, and Windows platforms, demonstrates striking results. CSCProv achieves approximately a thirtyfold reduction in graph size on CADETS, outperforming baselines such as ProTracer at 3.5 times and TAPAS at 9.0 times, and exceeds thirtyfold average reduction across the other datasets. On CLEARSCOPE, a grouping-based competitor achieved slightly higher compression, but the authors argue this reflects CSCProv’s deliberately stricter merge criterion, which refuses to aggregate entities that are structurally similar yet semantically different.

Crucially, compression does not come at the cost of security fidelity. Using metrics including Attack Information Loss, Causal Information Loss, Attack Node Preservation, and Attack Path Completeness, the team shows that CSCProv consistently achieves the lowest information loss and preserves over 96 percent of attack-related nodes across all datasets. End-to-end attack propagation paths remain traceable and distinguishable after compression, meaning forensic analysts can still reconstruct how an attack unfolded. In a case study on the THEIA dataset, the framework actually improved detection: repetitive memory and file entities that had caused a false positive were consolidated, while the attack-related file /home/admin/clean, previously missed, was correctly flagged.

Downstream testing with three established detectors—THREATRACE, MAGIC, and UNICORN—confirmed that compressed graphs maintain or modestly improve detection accuracy while cutting detection time by 32.7 to 48.3 percent across datasets and frameworks. Ablation studies reinforced the necessity of the dual-dimensional design: structure-only compression performed poorly on Living-off-the-Land attacks where semantics are camouflaged, while semantics-only compression failed on reconnaissance scenarios where structure is nearly uniform. The full CSC model achieved the best results on every metric, and sensitivity analysis showed the framework remains robust across a range of similarity thresholds.

The authors acknowledge limitations: all evaluated datasets follow the DARPA Common Data Model schema, so generalization to other audit logging systems requires further validation, and the current implementation performs offline compression of host-level graphs, leaving streaming and cross-host analysis to future work. Even so, CSCProv represents a significant step toward making provenance-based defense practical at enterprise scale, offering a way to tame the data explosion without blinding the defenders who depend on it.

Subject of Research: Causal-semantic consistent compression of system provenance graphs for advanced persistent threat detection

Article Title: CSCProv: causal-semantic consistent provenance graph compression for attack detection

Article References: An, N., Zhu, Y., Yan, W., Wu, X., Jiang, B., Liu, J., & Lu, Z. (2026). CSCProv: causal-semantic consistent provenance graph compression for attack detection. Cybersecurity, 9(1), Article 216. https://doi.org/10.1186/s42400-026-00648-6

Image Credits: AI Generated

DOI: 10.1186/s42400-026-00648-6

Keywords: provenance graph, graph compression, attack detection, advanced persistent threats, cybersecurity, system audit logs, Causal-Semantic Consistency, DARPA E3 dataset, graph neural networks, SecBERT, forensic analysis, threat detection

Diffractive neural networks, sometimes called diffractive deep neural networks, are built from a sequence of thin layers whose transmission or reflection coefficients are optimized by a computer. When a light wave passes through these layers, each point on one layer diffracts light toward many points on the next layer, and the pattern of connections between points behaves like the weights of an artificial neural network. Because the connection strength is set by how light spreads and interferes, the entire forward pass of the network happens at the speed of light. In prior demonstrations, however, the input was typically a static two-dimensional image or a single spatial pattern, and the layers were designed to transform that pattern from one spatial plane to the next. Such systems excel at tasks like image classification, but they treat every input as frozen in time.

The new approach recognizes that real-world signals, from communications waveforms to biological dynamics, are inherently temporal, and that a beam of light carries information in multiple degrees of freedom simultaneously: its spatial distribution, its wavelength, its polarization, and its temporal profile. The researchers show that by manipulating the spatiotemporal light field, meaning the way the field’s amplitude and phase evolve across space and time together, a diffractive network can encode, transform, and classify information that changes over time, all within the optical domain. The network layers are no longer merely spatial masks; they become spatiotemporal operators that shape how different temporal components of the input interfere and propagate.

Technically, the framework treats the optical field as a function of both position and time, and the diffractive layers are optimized so that the light field emerging from the final layer encodes the desired output, for example a classification decision or a transformed waveform. The design process draws on numerical modeling of wave propagation combined with training procedures familiar from machine learning, in which the layer parameters are iteratively adjusted to minimize an error function. Once training converges, the learned parameters are physically implemented in the optical layers, and the network performs inference passively, with no computation performed electronically during operation. This all-optical inference path is what distinguishes the architecture from hybrid optical-electronic schemes, where light performs some operations but electronic processors handle the rest.

The significance of adding the temporal dimension is substantial. A purely spatial diffractive network processes each snapshot independently, so it cannot natively recognize patterns that unfold over time, such as a spoken word, a sequence of pulses in a fiber, or the changing intensity of a dynamic scene. A spatiotemporal diffractive network, by contrast, can in principle integrate information across a temporal window as the light propagates, allowing the physics of diffraction and interference to perform temporal filtering, correlation, and sequence recognition. The authors present this capability as a route toward optical systems that can handle streaming data directly at the front end of a sensing or communication system, before any signal is converted to electronics.

The implications for energy efficiency are among the most compelling aspects of the research. Conventional artificial intelligence hardware consumes considerable power moving data between memory and processing units, and much of that cost is incurred performing the matrix multiplications that dominate neural network inference. Diffractive optical networks perform those multiplications passively, as light diffracts and interferes, so the energy cost of the forward pass is largely limited to the energy used to generate and detect the light. By extending the architecture to spatiotemporal operation, the new framework broadens the class of problems that can benefit from this efficiency, potentially including ultrafast signal processing in optical communications, where data streams already exist as modulated light and never need to be converted at all.

Speed is the other headline advantage. Because the computation is performed by propagating light, the latency of inference is set by the time it takes the wave to traverse the network, which can be on the order of picoseconds for compact devices. For temporal signals, this means the network can in principle keep pace with data rates that overwhelm electronic processors. The authors emphasize that the spatiotemporal manipulation of the light field is what unlocks this regime: by structuring the field in time as well as space, the network can perform operations on waveforms that would otherwise require high-speed sampling and digital signal processing chains.

The framework also connects to a broader research effort aimed at exploiting the full dimensionality of light for computing. Modern optical technologies can control wavelength, polarization, orbital angular momentum, and coherence, and each of these degrees of freedom can serve as a carrier of information or as a computational resource. Spatiotemporal light field manipulation, in which ultrafast pulses are shaped simultaneously in space and time, has matured rapidly in recent years, enabling phenomena such as space-time wave packets and light sheets with engineered group velocities. The new work harnesses this toolbox for neural computation, suggesting that the design space of optical neural networks is far larger than the spatial-only architectures explored to date.

As with any emerging technology, practical considerations will shape how quickly these systems move from laboratory demonstrations to deployed applications. Implementing spatiotemporal diffractive layers requires optical components that can impose carefully designed transformations on fast-varying fields, and the accuracy of the physical implementation relative to the trained model determines the network’s real-world performance. Alignment, fabrication tolerances, and detector bandwidth all matter. The authors frame their contribution as establishing the principles and design methodology for this new class of networks, providing a foundation on which experimental implementations across different spectral bands and platform technologies can be built.

The research arrives at a moment of intense global interest in unconventional computing substrates, driven by the growing energy and speed demands of artificial intelligence. Photonic approaches ranging from integrated silicon photonics to free-space diffractive optics promise orders-of-magnitude improvements in the energy efficiency of certain computations, and diffractive neural networks are among the simplest and most scalable of these approaches, since they can be fabricated as passive optical elements and require no active switching during inference. By showing that the same diffractive framework can be extended into the temporal domain, the study expands the reach of optical neural computation from static pattern recognition toward dynamic signal processing, a capability that could matter for applications as varied as ultrafast imaging, optical communications, lidar, and the analysis of fast biological processes. The work suggests a future in which the front end of an intelligent system is not a camera feeding a processor, but a shaped light field that has already done the thinking on its way to the detector.

Subject of Research: All-optical diffractive neural networks that process spatiotemporal light fields for temporal information processing

Article Title: Spatiotemporal all-optical diffractive neural networks empowered by spatiotemporal light field manipulation

Article References: Feng, F., Zhang, Z., Huo, D., Li, X., Lin, Q., Zhao, X., Hou, G., Dai, D., Somekh, M. G., & Yuan, X. (2026). Spatiotemporal all-optical diffractive neural networks empowered by spatiotemporal light field manipulation. Light: Science & Applications, 15(1), Article 382. https://doi.org/10.1038/s41377-026-02366-7

Image Credits: AI Generated

DOI: 10.1038/s41377-026-02366-7

Keywords: diffractive neural networks, all-optical computing, spatiotemporal light field manipulation, optical neural networks, photonic computing, wavefront shaping, ultrafast optics, machine learning hardware, energy-efficient computing, optical signal processing, Light Science and Applications, deep learning optics

The work, carried out by Hameed Ullah of King Fahd University of Petroleum and Minerals, Julien Mellet and Vincenzo Lippiello of the University of Naples Federico II, and Santos Miguel Orozco Soto of the Autonomous University of Mexico City, addresses a task roboticists call push-and-slide interaction. In this scenario, the robot must press its end-effector against a wall with a sustained, controlled normal force while simultaneously tracking a trajectory along the surface, such as a rectangle or a circle. The difficulty is that every Newton of force the robot exerts on the wall pushes back on the drone itself, threatening to destabilize a vehicle that must balance thrust, attitude, and position hundreds of times per second.

Conventional solutions to this problem lean heavily on force and torque sensors mounted at the wrist of the manipulator. These devices measure the interaction forces directly and feed them to the controller, but they come with punishing drawbacks for aerial platforms. They are fragile, adding risk in any operation involving unplanned contact. They are also relatively heavy, and on a flying robot every gram of payload translates directly into reduced flight time and reduced capacity for mission equipment. The research team set out to eliminate the external force sensor entirely, replacing its measurements with software estimation built on the physics of the vehicle itself.

The platform at the heart of the study is an actively tilting co-axial octarotor, a drone with eight rotors arranged in four coaxial pairs whose rotors can tilt in flight. This architecture is what gives the vehicle its omnidirectional capability: rather than relying solely on body tilt to direct thrust, as a conventional quadrotor must, the tilting rotors can generate forces and torques in arbitrary directions while the airframe stays level. For contact tasks this is a decisive advantage, because pushing sideways on a wall does not require the drone to bank, which would couple the contact force into the attitude dynamics and make the control problem far harder.

The second hardware innovation is at the tip of the manipulator. The team designed a specialized omni-wheel end-effector, a small roller-wheeled contact tool reminiscent of the wheels used on omni-directional ground robots. Its job is to decouple the two directions of interaction at the contact point. The normal direction carries the pressing force, which the wheels’ rollers allow to pass through with rolling contact in the tangential direction, meaning the lateral sliding motion along the surface encounters dramatically less friction. By stripping away most of the sliding friction that would otherwise fight the controller, the end-effector significantly reduces the disturbances injected into the drone during a slide, allowing a cleaner separation between the force control problem and the trajectory tracking problem.

With the sensor removed, the system still needs to know how hard it is pressing. The researchers’ answer is a momentum-based external wrench estimator, a well-established technique in robot control that has now been adapted for this tilting aerial platform. The estimator uses a dynamic model of the complete vehicle and manipulator, continuously comparing the momentum change predicted by the model with the momentum actually observed through the onboard state estimate. Any discrepancy that cannot be explained by the model’s own dynamics is attributed to generalized external forces and torques, which in this case are precisely the contact interaction wrenches. Because the method exploits the generalized momentum rather than raw accelerations, it is robust to the noise and unmodeled effects that plague aerial platforms, and it provides the controller with a smooth, real-time estimate of the force at the contact.

That force estimate then feeds a model-free hyperbolic position-based force controller. The controller is described as model-free because it does not require an explicit model of the contact environment, such as the wall’s stiffness, which is often unknown and can change from mission to mission. Instead, the error between the desired force setpoint and the estimated external force is mapped through a hyperbolic function into a position correction for the end-effector. The hyperbolic formulation, which the group had previously developed for robust flight control of tilting rotors, provides globally attractive convergence properties, meaning the force error is driven toward zero from a wide range of initial conditions rather than only when the system starts close to equilibrium. Combined with a position loop that handles the tangential direction, the result is a hybrid force and position scheme: force is regulated along the surface normal while position is controlled along the sliding direction, the classical decomposition that ground and arm robots have used for decades, finally executed stably by a free-flying vehicle.

The team validated the framework through an extensive simulation campaign in the widely used ROS and Gazebo robotics environment, chosen specifically for its high-fidelity physics. The experiments tasked the aerial manipulator with maintaining sustained contact force on vertical surfaces while tracking both rectangular and circular trajectories. Across these trials the system successfully followed the commanded paths while holding a stable force setpoint, demonstrating that the estimated wrenches were accurate enough to close the force loop reliably and that the tilting octarotor’s actuators had the authority to absorb the interaction loads without losing control. The authors report that the integration proved accurate, reliable, and fully sensorless for multi-dimensional aerial manipulation.

Beyond the specific results, the study claims to fill a recognized gap in the field: coordinated omnidirectional interaction, in which the full omnidirectional capability of an actively tilting coaxial platform is exploited to execute complex, nonlinear trajectories on vertical surfaces under stable force control. Push-and-slide inspection is a real industrial need, with prior projects in the literature targeting contact-based inspection of industrial plants, oil and gas infrastructure, and power lines, where a robot must both press and move. By showing that this can be done without a wrist force sensor, the framework lowers the weight, cost, and fragility barriers that have kept aerial physical interaction confined largely to laboratories.

The work also reflects a broader trend in aerial robotics: the migration of capabilities from traditional manipulators onto fully actuated flying vehicles. Momentum-based estimation, hybrid force and position control, and impedance-style interaction control were all pioneered on industrial arms, and adapting them to the fast, underdamped, aerodynamics-dominated world of multirotors has been an ongoing research program. This study contributes a piece of that program, demonstrating that a model-based observer paired with a mathematically grounded, environment-agnostic force controller can stand in for hardware sensing. The researchers caution that the validation to date is in simulation, and physical flight experiments with real contact uncertainty remain the next step, but the framework as presented offers a complete recipe, from omni-wheel tool design through wrench estimation to closed-loop force control, for any team building the next generation of touching, pushing, sliding drones.

Subject of Research: Hybrid force/position control of an omnidirectional aerial manipulator for sensorless two-dimensional push-and-slide interaction on vertical surfaces

Article Title: Hybrid force/position control for an omnidirectional aerial manipulator performing two-dimensional push-and-slide tasks

Article References: Ullah, H., Mellet, J., Orozco Soto, S. M., & Lippiello, V. (2026). Hybrid force/position control for an omnidirectional aerial manipulator performing two-dimensional push-and-slide tasks. Aerospace Systems. https://doi.org/10.1007/s42401-026-00543-5

Image Credits: AI Generated

DOI: 10.1007/s42401-026-00543-5

Keywords: aerial manipulation, hybrid force/position control, aerial physical interaction, omnidirectional drone, tilting co-axial octarotor, momentum-based estimator, push-and-slide, force estimation, ROS/Gazebo simulation, omni-wheel end-effector, sensorless control, vertical surface tracking

The research rests on a deceptively simple question: across the whole of Central Africa, is the forest gaining or losing woody carbon? Answering it has been notoriously difficult. Traditional approaches to estimating tropical carbon stocks rely either on sparse ground plots, which sample tiny fractions of the landscape, or on coarse-resolution satellite products that blur the fine-grained mosaic of intact forest, secondary growth, smallholder agriculture and logging gaps that characterizes the region. Averaged over tens or hundreds of kilometers, such coarse products can hide critical dynamics: a hectare of intensifying degradation next to a hectare of recovering vegetation may appear perfectly balanced in a low-resolution pixel, even as the actual carbon balance tips into deficit. The new maps, by contrast, resolve biomass change at a spatial grain fine enough to distinguish individual disturbance events, from industrial logging roads snaking into remote concessions to the slow expansion of farmland at forest edges.

To build these maps, the researchers combined multiple streams of satellite observation into a single, temporally consistent record of aboveground biomass. Spaceborne radar instruments are particularly valuable in the perpetually cloud-covered Congo Basin, where optical sensors are frequently blinded by persistent cloud cover. Radar signals penetrate clouds and, at longer wavelengths, interact directly with the woody structure of the forest, providing a measure of how much vegetation is standing on the ground. These radar observations were fused with data from spaceborne lidar, which samples vertical forest structure along satellite tracks, and with optical imagery that captures disturbance timing and vegetation recovery. Machine-learning models trained against forest inventory plots tie the satellite signals to actual quantities of carbon per hectare, allowing the mapping to extend calibrated, plot-level measurements continuously across tens of millions of hectares.

What distinguishes the new analysis is not merely the spatial detail but the accounting. Rather than snapshotting carbon stocks at two points in time and subtracting, which is vulnerable to errors in either map, the study tracks biomass change pixel by pixel through time, capturing both the losses caused by deforestation and degradation and the gains accumulated by growing forests. This dual bookkeeping matters because the two flows are of very different character. Losses are usually abrupt: a forest cleared for agriculture or hauled away as timber releases decades of stored carbon within months or years. Gains are slow: a recovering forest needs decades to rebuild what was lost. When the researchers tallied both sides of the ledger across Central Africa, the result was unambiguous: gains in growing biomass were insufficient to compensate for losses, yielding a net emission of carbon from the region’s forests rather than the net removal that many global models had assumed.

The geography of these losses is as informative as their magnitude. The study shows that the net sink strength varies enormously across the region, and that the declines are concentrated in specific zones rather than spread evenly across the basin. Forests in the western part of the Congo Basin, including areas of Cameroon, the Republic of Congo, Gabon and Equatorial Guinea, have historically exhibited among the highest biomass densities of any tropical forest on Earth, with some stands holding more carbon per hectare than lowland Amazonia. The new maps indicate that where these carbon-dense forests are disturbed, the resulting emissions are disproportionately large, because each hectare lost carries an exceptionally heavy carbon cargo. In other areas, long-term degradation from selective logging, fuelwood harvesting and shifting cultivation thins the forest canopy and erodes biomass gradually, a process that is largely invisible to conventional deforestation monitoring, which traditionally registers only complete forest clearance.

This distinction between deforestation and degradation is one of the study’s central contributions. International policy frameworks, including REDD+ programs that channel climate finance into forest conservation, have historically focused on monitoring deforestation, the visible and permanent conversion of forest to non-forest. But the high-resolution biomass change maps make clear that degradation, the partial and often reversible loss of carbon within standing forest, accounts for a large share of the region’s net carbon losses. Selective logging removes only the commercially valuable stems, yet each extracted tree leaves behind damaged neighbors, abandoned roads and a canopy gap through which the forest floor dries and decomposes faster. Fire, increasingly frequent at the humid forest’s dry margins, similarly kills trees without clearing them. Because degraded forest remains classified as forest, its carbon losses accumulate below the threshold of conventional monitoring, silently converting a regional sink into a source.

The findings arrive at a moment of genuine uncertainty about the future of tropical carbon. Global climate models generally assume that intact tropical forests will continue to absorb carbon, offsetting a meaningful fraction of fossil fuel emissions, but the empirical basis for that assumption is weakening. Long-term forest plots across the tropics have documented a slowdown in the rate at which undisturbed forest gains biomass, a pattern widely attributed to increasing drought, heat stress and atmospheric changes. If the Central African forests, previously the most resilient of the major tropical forest blocks, are now slipping into net carbon loss, the implications extend beyond the region itself. Carbon budgets consistent with the Paris Agreement already have little room for the world’s forests to flip from helping to hindering; a Central African reversal would consume a portion of that remaining room all on its own.

The study also underscores a regional irony with global resonance. Central Africa’s per capita emissions are among the lowest in the world, and its forests have been doing the planet a service for decades by storing carbon at exceptional densities. Yet the drivers of the emerging carbon losses are entangled with pressures that are partly global in origin: demand for timber and agricultural commodities, infrastructure corridors that open previously inaccessible forest, and climatic shifts driven by emissions generated far from the basin. Drought episodes that once receded without lasting damage now leave measurable scars in the biomass record. The high-resolution maps make it possible to see, for the first time with clarity at scale, how these pressures interact across the landscape, and where intact forest refugia still persist as anchors for conservation.

There are, however, constructive signals embedded in the data. The same maps that reveal net losses also identify the places where forests are reliably gaining carbon: regrowing secondary forests, abandoned agricultural land recovering toward maturity, and well-protected core areas where intact forests continue to accumulate biomass. This spatial intelligence is precisely what national forest monitoring systems and international climate finance mechanisms need in order to target interventions where they will matter most. Protecting the carbon-dense forests of the western basin, accelerating the recovery of degraded areas, and strengthening enforcement against illegal logging all emerge as evidence-backed priorities. The study’s methodology also offers a template that other forest nations can adopt, demonstrating that plot-calibrated, multi-sensor satellite mapping can now deliver wall-to-wall carbon accounting at a resolution fine enough to guide policy.

For decades, the Congo Basin forests have been the quiet heroes of the global carbon story, absorbing emissions without fanfare while deforestation focused global attention elsewhere. The new biomass change maps retire that comfortable assumption and replace it with a more demanding truth: these forests are not immune to the pressures reshaping tropical ecosystems worldwide, and their carbon balance has already tipped negative. Whether that tipping proves to be a temporary fluctuation, driven by drought and disturbance pulses that forests can still recover from, or the early stage of a durable transition from sink to source, is one of the most consequential open questions in climate science. What is no longer open to question is that the answer must be tracked in detail. With high-resolution biomass monitoring now demonstrated at regional scale, the world’s ability to see what Central African forests are doing, and to act before their decline accelerates, has taken a decisive step forward.

Subject of Research: Satellite-based high-resolution mapping of biomass and carbon changes in Central African tropical forests

Article Title: Net carbon losses in Central African forests revealed by high-resolution biomass change maps

Article References: Wan, L., Ciais, P., de Truchis, A., Xu, Y., Brandt, M., Chave, J., Bourgoin, C., Wigneron, J.-P., Bastin, J.-F., Li, W., Ryu, Y., Liu, S., Purnell, D., Fayad, I., Sagang, L. B., Vander Linden, A., Besisa, T., & Ploton, P. (2026). Net carbon losses in Central African forests revealed by high-resolution biomass change maps. Nature Communications. https://doi.org/10.1038/s41467-026-77531-y

Image Credits: AI Generated

DOI: 10.1038/s41467-026-77531-y

Keywords: Central Africa, tropical forests, carbon cycle, biomass mapping, remote sensing, climate change, carbon sink, deforestation, forest degradation, Nature Communications, net carbon loss, satellite imagery

The appeal of quantum batteries lies in exploiting resources that classical devices simply do not have. Entanglement, quantum coherence, and many-body correlations can be harnessed to accelerate charging beyond classical transport limits. Since Alicki and Fannes formalized the concept more than a decade ago, theorists have demonstrated that collective charging protocols can push charging power to scale superextensively with system size—approaching the fundamental Heisenberg bound, where power grows quadratically with the number of cells rather than linearly. Experiments using spin systems, superconducting circuits, and organic molecular aggregates have begun validating these predictions, turning quantum batteries into a genuine technological frontier for powering quantum computers, sensors, and communication networks at the nanoscale.

The Iranian team focused on spin-1/2 chains because they are among the most tunable and theoretically tractable platforms for studying quantum energy storage. Their model is an anisotropic XY chain augmented with a three-spin interaction whose strength and anisotropy can be varied independently. The three-spin term is more than a decorative addition: it breaks the continuous rotational symmetry of the ordinary XX chain and enriches the ground-state phase diagram dramatically. In the isotropic limit, the model supports six distinct non-trivial topological phases, each labeled by a winding number of plus or minus one or two. Switching on XY anisotropy adds a seventh, topologically trivial region and a fifth critical line separating phases whose bulk excitation spectra and edge physics differ fundamentally.

A key technical achievement of the study is that the model remains exactly solvable. Using the Jordan–Wigner transformation, the authors map the interacting spin chain onto a quadratic form of non-interacting fermions—a generalized version of the celebrated Kitaev chain, the paradigmatic model of one-dimensional topological superconductivity. But where the standard Kitaev model has only nearest-neighbor hopping and p-wave pairing, this generalized chain includes next-nearest-neighbor hopping and long-range p-wave pairing, producing a far richer landscape of topological phases. In momentum space, the Hamiltonian decouples into independent two-level blocks for each momentum mode, allowing the researchers to diagonalize it exactly and track every quasiparticle excitation through the charging process.

The charging protocol itself is a quantum quench. The battery is prepared in the ground state of an initial Hamiltonian, which serves as its reference state. At time zero, the Hamiltonian is abruptly switched to a different charging Hamiltonian, and the system evolves unitarily, absorbing energy as quasiparticle modes are populated. After a charging duration, the Hamiltonian is quenched back, decoupling the battery from its charger and trapping the stored energy. This sudden-switch protocol is one of the most natural ways to drive a quantum many-body system out of equilibrium, and it directly probes how the spectral structure of the underlying chain governs energy absorption.

From the exact solution, the authors derive an analytical expression for the stored energy as a sum over momentum modes, each contributing an oscillatory term weighted by the overlap between the initial and final quasiparticle states. This structure defines a geometric factor—an upper bound on the maximum energy the battery can ever hold for a given quench. The team’s central discovery is that when the system evolves for a long time, the dephasing between the different momentum modes drives each oscillatory contribution toward its time average of one half. The steady-state stored energy therefore equals exactly half of the geometric factor, universally, across every quench protocol and every parameter regime they examined. This half-energy rule reflects a rigid geometric constraint imposed by the Hilbert-space structure of the fermionized chain, and numerical simulations on chains of one thousand spins confirm the analytical prediction with striking precision.

Perhaps the most striking result is how sensitive this energy storage capacity is to quantum criticality. The geometric factor exhibits a pronounced maximum precisely at one of the model’s critical points, where the bulk energy gap closes and the system undergoes a topological phase transition. The first derivative of the geometric factor develops sharp cusps at every critical value, revealing that charging efficiency is deeply rooted in the critical fluctuations of the ground state. Moreover, whichever critical line maximizes the stored energy depends on the initial configuration of the quench, meaning the battery’s performance encodes detailed information about where its charging trajectory begins and ends within the topological phase diagram. In effect, the charged battery acts as a readout of the quantum phase transitions it was driven across.

The short-time charging dynamics are equally revealing. A Taylor expansion shows that the stored energy initially grows quadratically with time, with a rate set by the initial quasiparticle dispersion, the square of the final dispersion, and the transition probabilities between eigenstates. The time to reach peak energy is governed by a ratio of spectral moments of the initial and final Hamiltonians, and because the quench populates a broad swath of the excitation spectrum, this peak time varies smoothly and robustly across the parameter range rather than spiking at isolated resonances. The team supplemented their analytical estimate with an empirical polynomial fit that accurately captures the numerical peak time over the full quench interval, providing a practical benchmark for the charging performance of this architecture.

The interplay between topology and coherence becomes even clearer at long times. In generic quenches, energy disperses across many modes, decoherence accumulates, and the post-revival energy maximum falls short of the initial charging peak. But when the final Hamiltonian sits at a critical point, the gap closing enhances mode degeneracy and slows the decay of coherence, allowing quantum revivals to reconstruct stored energy more effectively. At some—but not all—critical points, the revival maximum actually exceeds the short-time peak, a signature of topological protection mitigating decoherence. Meanwhile, the peak charging power density rises essentially monotonically as the quench drives the system deeper into different topological regimes, showing that the location of the final Hamiltonian in the phase diagram directly controls how fast the battery can be charged.

The authors emphasize that their model could plausibly be engineered on programmable quantum simulation platforms, including Rydberg-atom arrays, where recent experiments have demonstrated tunable spin Hamiltonians with controllable anisotropies and Floquet-engineering schemes have been proposed for realizing generalized spin-exchange interactions. A full experimental implementation of the three-spin Hamiltonian remains beyond the scope of the current work, but the framework establishes a concrete route toward quantum batteries whose performance is deliberately steered by topological design. Beyond energy storage, the exquisite sensitivity of charging dynamics to critical lines suggests a dual use: such devices could simultaneously function as precision probes of quantum phase transitions, turning a future quantum battery into both a power source and a diagnostic instrument for the quantum materials it is built from.

Subject of Research: Energy storage limits and charging dynamics in an anisotropic three-spin interaction quantum battery

Article Title: Energy storage limits and criticality in an anisotropic three-spin interaction quantum battery

Article References: Beygi, V. Y., Mahdavifar, S., Mirmasoudi, F., & Ahadpour, S. (2026). Energy storage limits and criticality in an anisotropic three-spin interaction quantum battery. Results in Physics, 88, Article 108748. https://doi.org/10.1016/j.rinp.2026.108748

Image Credits: AI Generated

DOI: 10.1016/j.rinp.2026.108748

Keywords: quantum battery, three-spin interaction, topological phase transition, spin chain, quantum criticality, ergotropy, quantum quench, charging power, Jordan-Wigner transformation, Kitaev chain, stored energy, Rydberg atoms

Machine vision systems combine image acquisition hardware, such as RGB cameras, multispectral and hyperspectral sensors, infrared thermography and even X-ray imaging, with software pipelines that clean, enhance and analyze the resulting images. The survey traces the classical workflow in detail. Raw images are first converted between color spaces or reduced to grayscale, then enhanced through techniques such as histogram equalization and its many adaptive variants, which stretch contrast while preserving brightness and structural detail. Noise introduced by dust, vibration and inconsistent field lighting is suppressed with Gaussian, median and Wiener-style filters, some of them optimized for real-time performance on embedded processors. Only after this preprocessing can the harder tasks begin: segmenting plants from soil, extracting features such as color, texture and shape, and classifying what the camera has actually seen.

Those downstream tasks have been transformed by the deep learning revolution. The survey documents the field’s migration from hand-engineered classifiers such as k-nearest neighbors, support vector machines, logistic regression and random forests toward convolutional neural networks, including landmark architectures such as AlexNet and the YOLO family of real-time object detectors, and more recently toward transformer-based and hybrid convolutional-transformer models. In plant disease detection alone, the authors cite systematic reviews showing that deep learning approaches now dominate the literature, with models trained on leaf imagery achieving rapid, automated diagnosis across crops as varied as tomato, grape, citrus, papaya and blueberry. Explainable deep vision frameworks have even been applied to plant stress phenotyping, giving breeders not just a prediction but a spatial map of where stress manifests on the plant.

Pest identification and monitoring emerges as one of the most mature application areas. Early systems relied on color cues to distinguish weeds from crops, while large-scale investigations demonstrated that machine vision could identify weed seeds with high accuracy. More recent work combines k-means clustering with convolutional neural networks for weed identification, enabling precision sprayers that apply herbicide only where weeds are detected rather than across entire fields. Smartphone-based systems now allow aphid identification and counting in the field, and light-attracted pest traps fitted with vision modules can automatically recognize and tally insect catches at high altitude in orchards. The practical payoff is a reduction in chemical inputs, lower costs and less environmental burden, all of which align with the sustainability goals that motivate precision agriculture in the first place.

The survey also charts how vision systems track crop growth and estimate yield, a problem with direct economic consequences. Researchers have measured seedling growth rates from images as early as the 1990s, and subsequent systems have monitored greenhouse vegetables, mushrooms and chrysanthemums non-destructively over time. Yield mapping began with citrus, where cameras counted fruit on the tree, and has since expanded to tomato yield estimation and fruit maturity detection using machine vision pipelines. Crop-load estimation with YOLOv8 illustrates the current state of the art: a single neural network counts fruit in real time from imagery captured on the move, giving growers a data-driven forecast of harvest volume before a single crate is filled. Systematic reviews of machine learning for crop yield prediction confirm that such vision-derived features are increasingly central to these forecasting models.

Perhaps the most visually dramatic applications involve harvesting robots, which must find fruit, localize it in three dimensions and guide a manipulator to pick it without damaging the crop. The review covers recognition and localization methods for fruit-picking robots across cucumber, apple, cotton, strawberry and citrus systems, including approaches that distinguish fruit from branch in cluttered natural scenes using support vector machines, and methods that reconstruct 3D models of fruit for precise grasping. Hyperspectral imaging paired with deep learning can even spot early bruises on apples that are invisible to the human eye, while X-ray and machine vision combinations probe internal fruit quality non-destructively. These capabilities matter because a robot that cannot reliably see ripe, undamaged fruit in variable lighting is a robot that cannot harvest at all.

Beyond the field, machine vision governs the quality grading and sorting lines that decide which products reach consumers. The survey documents multispectral real-time inspection of citrus dating back to the early 2000s, defect segmentation on apples, quality evaluation of soybeans, maturity prediction for harvested mangoes, and automatic grading of eggs, hairy crabs, walnuts, dragon fruit and litchi. Classical statistical tools such as principal component analysis and Gabor features once powered these systems; today, weighted k-means clustering, AlexNet-derived networks and automated machine learning pipelines sort produce by size, color, shape and surface defects at production-line speeds. Seed quality inspection has followed the same arc, with spectral detection of maize seed vigor and machine vision classification of seed defects enabling pre-planting screening that was previously impossible at scale.

Visual navigation for agricultural robots rounds out the survey’s application landscape. By extracting crop rows, navigation baselines and linear targets from camera imagery, machines can drive themselves through fields, orchards and paddy fields, often fusing vision with GPS for robustness. Stereo vision provides obstacle detection for off-road vehicles, and autonomous robotic mowers have demonstrated navigation and obstacle avoidance in orchards using purely visual cues. The authors note that this capability is converging with broader cyber-physical and Internet of Things architectures, in which vision-equipped machines, cloud analytics and renewable-energy-powered sensor networks form integrated cyber-agricultural systems capable of closing the loop from perception to action across entire farms.

The survey is candid about the obstacles that remain. Field lighting is notoriously inconsistent, motivating engineering fixes such as overcurrent-driven LEDs that guarantee stable image color and brightness. Datasets are often imbalanced or too small for the deep models being applied, and occlusion, clutter and the sheer biological variability of living crops continue to challenge even state-of-the-art detectors. The authors also flag the computational cost of running heavy neural networks on the embedded hardware that agricultural machinery can realistically carry, and the need for interpretable models that farmers can trust. Their forward-looking section points toward transformer architectures, multimodal sensor fusion combining hyperspectral and multispectral imagery, and tighter integration of vision with the cyber-physical systems that will define the next generation of autonomous agriculture.

What emerges from the full sweep of the review is a discipline in transition. The foundational image processing techniques of the 1990s and 2000s, from thresholding and edge detection to early neural classifiers, laid the groundwork; the deep learning era supplied the accuracy and generality that made commercial deployment plausible; and the current wave of transformers, explainable AI and cyber-physical integration is pushing machine vision toward farms that monitor, decide and act with minimal human intervention. For a world that must produce more food with fewer inputs on less land under a changing climate, the authors argue, teaching machines to see may prove one of the most consequential technologies agriculture has ever adopted.

Subject of Research: Machine vision applications in precision agriculture, including crop disease detection, yield estimation, robotic harvesting, quality grading and visual navigation

Article Title: A comprehensive survey on machine vision applications in precision agriculture: current trends and future perspectives

Article References: A comprehensive survey on machine vision applications in precision agriculture: current trends and future perspectives. (n.d.). https://doi.org/10.1007/s41060-026-01278-4

Image Credits: AI Generated

DOI: 10.1007/s41060-026-01278-4

Keywords: machine vision, precision agriculture, deep learning, computer vision, crop disease detection, yield estimation, harvesting robots, agricultural robotics, image processing, hyperspectral imaging, smart farming, artificial intelligence