The problem the researchers tackle is deceptively simple to state and brutally hard to solve. When a machine learning model is trained, it learns the statistical patterns of its training data, the relationships between symptoms, vital signs, and outcomes. But patient populations change, diseases evolve, sensor hardware ages, and clinical practices shift. The underlying data distribution begins to exhibit patterns that differ from those seen during training, and the model’s performance deteriorates over time. In a centralized cloud system, engineers can periodically re-evaluate a model against labeled data. But in fog computing, where intelligence is pushed down to small gateways and edge devices scattered across clinics, ambulances, and patients’ homes, drift can occur independently at every device, and the labeled ground truth needed to measure accuracy is almost never available in real time.

Fog computing has become the dominant architecture for Internet of Things based healthcare precisely because it solves a different problem: latency. Systems such as HealthFog, which performs automatic heart disease diagnosis on integrated IoT and fog platforms, and smart e-health gateways that classify electrocardiogram signals locally, depend on processing patient data close to its source rather than round-tripping it to a distant data center. For time-critical applications such as arrhythmia detection or COVID-19 severity prediction, the milliseconds saved by local processing can matter clinically. But this same decentralization creates a monitoring blind spot. Each fog node runs its own copy of the model, ingests its own stream of patient data, and drifts in its own way. A drift detector that works at the cloud level is useless if the damage happens at the edge.

The core difficulty is that classic drift detection methods assume access to true labels. Techniques such as the Drift Detection Method of Gama and colleagues from 2004, or the Early Drift Detection Method published in 2006, monitor the error rate of a classifier as labeled examples stream in, and raise an alarm when the error rate rises beyond statistically expected bounds. Adaptive windowing approaches such as ADWIN, introduced by Bifet and Gavaldà in 2007, slice the stream into windows and compare performance across them. These methods are elegant and well understood, but in a deployed healthcare system, patient data arrives unlabeled. Nobody knows the true diagnosis the moment a wearable transmits a vital-sign vector. Waiting for a clinician’s confirmation to feed the detector would defeat the purpose of real-time monitoring, and would reintroduce exactly the human workload the automation was meant to reduce.

Ahmed and Masood’s insight is to manufacture an approximate error signal without labels, by making a classifier and a clustering model learn from the same data simultaneously. During training, the system fits a supervised classifier and an unsupervised clustering model on the same dataset. Because both models are exposed to identical examples, a statistical relationship emerges between them: the accuracy of the classifier becomes linked to the purity of the clusters. Cluster purity measures how homogeneous each cluster is with respect to class membership, so when the clusters cleanly separate the classes, the classifier tends to predict accurately, and when the cluster structure blurs, classifier performance degrades in a predictable way.

At deployment time, this relationship becomes the detection mechanism. Every unlabeled instance arriving at the fog node is processed twice: the classifier predicts a label, and the clustering model assigns the instance to a cluster that carries its own class label, derived from the training phase. The detector then simply compares the classifier’s predicted label with the label assigned to the instance’s cluster. When the two agree, the system treats the instance as likely correctly classified; when they disagree, it counts as an approximate error. Streaming these approximate errors over time produces a signal that behaves like a true error rate, and standard drift detection logic can be applied to it. If the approximate error rate climbs beyond expected statistical variation, the system concludes that concept drift has occurred and can trigger adaptation or retraining, without a single human-labeled example.

The elegance of this approach lies in its economy. It requires no additional annotation, no clinician in the loop, and no modification to the underlying prediction model beyond the parallel clustering fit. In a fog architecture, each node can run its own classifier-cluster pair and detect drift locally, which means the method scales naturally across the distributed devices where drift actually happens. The authors report that their experiments show the proposed method reduces the need for human intervention, which is the operational bottleneck in maintaining fleets of healthcare edge devices. Instead of waiting for a technician to notice that a remote monitoring station has gone stale, the system notices on its own.

The study situates itself within a rapidly growing literature on drift under constraint. Recent work has pushed toward fully unsupervised drift detection, including benchmarks of unsupervised detectors on real-world data streams and adaptive windowing methods designed specifically for unlabeled data, such as ADWIN-U published in 2025. Others have attacked the problem from the federated learning angle, where concept drift distributed across clients complicates model aggregation, and from the autoencoder angle, using reconstruction error as a drift proxy. Empirical studies on real medical imaging data have confirmed that data drift is not a theoretical worry but an observable clinical reality. What distinguishes RTDDM is its explicit targeting of the fog computing context, where computation is cheap enough to run a parallel clustering model but connectivity is too limited for centralized, label-dependent monitoring.

The experimental foundation of the paper draws on publicly available datasets, including a widely used COVID-19 clinical dataset originating from the Government of Mexico’s epidemiological authority, and the implementation builds on the River library for streaming machine learning in Python, a standard toolkit for this research community. The authors describe their preprocessing steps, model parameters, and evaluation procedure in sufficient detail to support reproducibility, with scripts available from the corresponding author on reasonable request. The work was carried out under the SAFE-RH project, funded by the European Commission through the Erasmus+ programme, reflecting the growing institutional investment in resilient remote healthcare infrastructure.

The implications extend well beyond the specific datasets tested. As healthcare systems worldwide embed machine learning into telemedicine platforms, wearable monitoring networks, and fog-assisted diagnostic gateways, the question of who notices when a model goes wrong becomes a patient-safety issue, not merely an engineering one. A drift detector that works on unlabeled data, at the edge, in real time, converts model maintenance from a reactive, labor-intensive chore into an automated property of the system itself. It is the kind of unglamorous infrastructure research that determines whether clinical AI fulfills its promise or quietly fails in the field. Ahmed and Masood’s method offers a concrete, testable step toward that resilience, and it signals a broader shift in the field: the era of assuming that a trained model is a finished product is ending, replaced by the understanding that intelligence, like the patients it serves, is something that must be continuously monitored as the world changes underneath it.

Subject of Research: Real-time machine learning concept drift detection for unlabeled data streams in fog computing based remote healthcare systems

Article Title: Implementing machine learning model drift detection in fog computing architectures for remote healthcare systems

Article References: Ahmed, S., & Masood, N. (2026). Implementing machine learning model drift detection in fog computing architectures for remote healthcare systems. Cluster Computing, 29(14), Article 790. https://doi.org/10.1007/s10586-026-06545-4

Image Credits: AI Generated

DOI: 10.1007/s10586-026-06545-4

Keywords: machine learning, concept drift, drift detection, fog computing, remote healthcare, unlabeled data, clustering, classifier accuracy, edge computing, telemedicine, data streams, RTDDM

Semi-supervised regression sits at the intersection of two worlds. In supervised learning, every training example comes with a known target value, such as the exact energy output of a solar panel or the measured concentration of a pollutant. In unsupervised learning, the algorithm must find structure without any answers at all. Semi-supervised methods try to have it both ways, using a small labeled set to anchor the model and a much larger unlabeled set to refine it. The promise is enormous, because collecting raw measurements is usually far cheaper than annotating them, and domains from biomedicine to finance to industrial manufacturing are drowning in unannotated numerical data.

The dominant strategies in this field have long followed a conservative philosophy. Early approaches selected only a small number of high-confidence unlabeled examples and folded them into the training data, effectively discarding the rest. This filtering kept the training signal clean, but it also threw away most of the information contained in the unlabeled pool. More recent methods took the opposite approach, using off-the-shelf semi-supervised regressors to generate pseudo-labels, which are the model’s own predictions treated as if they were ground truth, for every unlabeled example. The problem, as the Chinese team points out, is that these methods treat all pseudo-labels uniformly during training, ignoring the inherent quality differences among them and potentially injecting significant noise into the learning process.

FullReg addresses this weakness with two interlocking mechanisms. The first is a confidence-weighting scheme based on data similarity. Rather than accepting every pseudo-label at face value, the framework assigns each one a weight in the loss function that reflects how trustworthy it is likely to be. The intuition is geometric: if an unlabeled example sits close to labeled examples in the input space, its neighbors’ known target values provide meaningful evidence about what its own target should be, so its pseudo-label earns a high weight. If an unlabeled point floats in a sparse region far from any labeled data, the model’s guess about its value is essentially unsupported, and the weight drops accordingly. By modulating the contribution of each pseudo-label to the overall loss, the mechanism dampens the influence of unreliable predictions and preserves the validity of the training signal.

This idea of weighting by similarity has deep roots in statistical learning, where kernel methods and locally weighted regression have long recognized that predictions are more reliable near observed data. What FullReg adds is a systematic way to translate that geometric intuition into the training dynamics of a neural network performing semi-supervised regression. The result is a framework that can exploit the entire unlabeled pool, as the newer generation of methods does, while retaining the noise resistance that made the older, selective approaches robust. In effect, the model no longer has to choose between using all of its data and trusting what that data tells it.

The second innovation is a residual-connection mechanism that operates across training epochs rather than across network layers. The name deliberately echoes the residual connections popularized by deep residual networks in computer vision, where skip links allow information to bypass layers and stabilize training. Here, the connection is temporal: at each epoch, the framework blends the model parameters inherited from previous epochs with the parameters being learned in the current one. Instead of letting the network lurch toward whatever solution the latest batch of data suggests, the residual mechanism anchors it to its own history, producing a trajectory of parameter updates that evolves progressively rather than erratically.

This temporal smoothing serves a similar purpose to techniques such as temporal ensembling and weight averaging, which have been shown in prior research to lead neural networks toward wider optima and better generalization. It also echoes the mean teacher paradigm, in which an averaged copy of a model provides steadier training targets than the model itself. By embedding that stabilizing principle directly into the parameter updates of a semi-supervised regression pipeline, FullReg gains resilience against the fluctuations that pseudo-label noise would otherwise introduce. The two mechanisms reinforce each other: confidence weighting reduces the noise entering the loss, while residual connections prevent whatever noise remains from destabilizing the learned parameters.

To test the framework, the researchers ran experiments on benchmark datasets drawn from five distinct domains, spanning biomedical, business, ecology, physical, and life sciences data, sourced from public repositories including the UCI Machine Learning Repository, the Delve repository, and the StatLib archive. They also evaluated the method on a real-world solar photovoltaic dataset, a setting where accurate regression matters for forecasting power generation from grid-connected installations. Across these benchmarks, FullReg was compared against eight state-of-the-art semi-supervised regression algorithms, and it compared favorably in most benchmark settings, suggesting that the combination of full data utilization and noise-aware training translates into measurable predictive gains rather than merely theoretical elegance.

The practical implications extend well beyond benchmark tables. Consider solar power forecasting, where weather stations, inverter readings, and satellite imagery generate torrents of measurements but ground-truth labels for every operating condition are scarce. A framework that can safely exploit all of that unlabeled data, rather than a hand-picked high-confidence subset, could sharpen the forecasts that grid operators rely on to balance supply and demand. Similar logic applies to air temperature mapping, stock price prediction during volatile periods, thermal error compensation in precision manufacturing, and clinical risk prediction, all of which are cited in the study’s bibliography as active application areas for semi-supervised regression. In each case, the bottleneck is the same: labeled examples are few, unlabeled examples are plentiful, and the quality of machine-generated labels varies wildly.

The work also contributes to a broader conversation in machine learning about how models should treat their own outputs. Pseudo-labeling has become a cornerstone of modern semi-supervised learning, powering influential techniques in image classification, semantic segmentation, and few-shot learning, yet the calibration of pseudo-label quality remains an open problem. FullReg’s answer, grounding confidence in data similarity and stabilizing learning through temporal residual connections, offers a template that other researchers may adapt to classification and other tasks. The authors have made their benchmark analysis transparent, drawing on publicly available datasets, with the solar photovoltaic dataset and code available from the corresponding author on reasonable request. As unlabeled data continues to accumulate faster than any labeling effort could match, methods like this one, which learn to distrust their own mistakes in a principled way, may define the next generation of practical machine learning.

Subject of Research: Semi-supervised regression using confidence-weighted pseudo-labels and residual parameter connections

Article Title: Semi-supervised regression via confidence-weighting and residual-connection

Article References: Liu, L., Mao, Y., Lu, X., & Min, F. (2026). Semi-supervised regression via confidence-weighting and residual-connection. Applied Intelligence, 56(15), Article 440. https://doi.org/10.1007/s10489-026-07489-3

Image Credits: AI Generated

DOI: 10.1007/s10489-026-07489-3

Keywords: semi-supervised regression, pseudo-labels, confidence weighting, data similarity, residual connections, neural networks, machine learning, unlabeled data, solar photovoltaic forecasting, Applied Intelligence, Southwest Petroleum University, loss function

The study, published open access in the journal Machine Learning by Sintija Stevanoska, Christian L. Camacho Villalón, Sašo Džeroski and Katharina Dost, tackles a well-known pain point in applied machine learning. Labels are expensive: healthcare systems accumulate millions of patient records, yet expert-annotated outcomes remain scarce. Self-supervised learning promises to bridge that gap by defining pretext tasks from the intrinsic structure of data — masking features and reconstructing them, corrupting values and asking a model to spot the damage, or learning invariant representations across contrastive views. For tables, however, these methods inherit an awkward problem. Unlike images or sentences, tabular rows lack spatial or sequential structure, mix numerical and categorical features, and often contain missing values and tangled inter-dependencies. And pretraining on very large unlabeled pools is computationally expensive while potentially importing noisy or unrepresentative samples that actively degrade learning.

The researchers asked a deceptively simple question: is a small, well-chosen subset of unlabeled data sufficient for effective pretraining? To answer it, they ran what they describe as the first large-scale evaluation of unlabeled data subsampling for tabular self-supervised learning, spanning 25 benchmark datasets drawn from the OpenML-CC18 suite and covering domains from biology and finance to medicine and climate science. Four representative self-supervised models were tested — TabNet, VIME, SCARF and SAINT — deliberately chosen to span the field’s three main paradigms. TabNet and VIME learn by masking or corrupting features and reconstructing them, SCARF relies on contrastive views generated by resampling features from their empirical distributions, and SAINT combines contrastive and denoising objectives inside a transformer that attends over both rows and columns. This diversity of pretext tasks and encoder architectures ensured the conclusions would not hinge on one particular model family.

At the heart of the study lies a family of selection strategies borrowed and adapted from active learning, the branch of machine learning concerned with identifying the most informative examples to label. The team compared random sampling against three principled approaches. Disagreement sampling, based on the classic query-by-committee method, trains a committee of 100 decision trees on bootstrap samples of the labeled data, restricting each tree to a random subset of features in the manner of random forests, and scores every unlabeled point by the normalized entropy of the committee’s votes — a value from 0, indicating complete agreement, to 1, indicating uniform disagreement. Diversity sampling applies k-means clustering to the unlabeled pool and picks one representative per cluster, the point closest to each centroid, enforcing a minimum cluster size to avoid crowning representatives of tiny outlier groups. Optimal transport sampling treats the labeled and unlabeled sets as empirical distributions, computes an entropically regularized transport plan between them with the Sinkhorn algorithm, and assigns each unlabeled point its expected transport cost to the labeled distribution — a low score signaling strong alignment with the labeled data, a high score signaling underrepresented regions of the feature space.

Crucially, all of these strategies are deliberately model-agnostic. They are computed before any self-supervised pretraining begins, using no gradients, losses, embeddings or augmentations from the downstream model. Score-based methods partition the unlabeled pool into percentile brackets — low, medium and high — and samples are then drawn from a chosen bracket, either uniformly at random or via the diversity-based cluster representatives. This design allowed the authors to evaluate the sampling approaches consistently across all four pretext objectives, isolating the effect of data selection from the idiosyncrasies of any particular architecture.

The experimental scale was formidable. Labeled subsets of 100, 500 and 1000 examples were drawn under both unbiased, stratified conditions and a deliberately biased regime in which labeled examples were preferentially selected from one branch of a decision stump, simulating real-world selection bias. Unlabeled subsets of 100 to 1000 examples were drawn under every combination of strategy, subset size, bracket and five random seeds. Performance was measured through three lenses: downstream classification quality using macro-averaged area under the precision-recall curve, total wall-clock time including both selection and training, and representation quality assessed by linear probing — training a simple logistic regression classifier on frozen pretrained embeddings to test whether class information is linearly encoded in the learned latent space.

The headline finding is striking. Subsampling only 100 to 1000 unlabeled examples often matched or exceeded full-data pretraining across both labeled-data regimes. In the low-label, unbiased setting, the best sampling strategies reached up to 119 percent of the full-data baseline and 114 percent of a purely supervised baseline that skipped pretraining entirely. Under biased labeled selection the gains grew even more pronounced, climbing to 123 percent of the full-data baseline and 115 percent of supervised learning. The effect was strongest exactly where it matters most: when labels were scarcest, and the choice of pretraining data had the greatest opportunity to shape what the model learned. Optimal transport distance and disagreement-based sampling performed most consistently, but the humble random sampler was surprisingly competitive, hinting that for small labeled sets, simply reducing redundancy and avoiding overfitting to the unlabeled distribution may itself act as a form of regularization.

The computational story is equally compelling. Pretraining on the full unlabeled pool consistently demanded 60 to 70 percent more time than the sampling strategies, and the selection step itself added a negligible share of the total runtime. Random sampling typically finished in 29 to 42 percent of full-data time, with optimal transport close behind; diversity-based selection was the most expensive to compute, consuming up to 70 percent of the full-data baseline’s runtime without delivering consistent accuracy dividends. In other words, competitive or better predictive performance was achieved at roughly 30 to 40 percent of the computational cost — a margin that alone would justify subsampling in resource-constrained environments even without accuracy improvements. Linear probing reinforced the picture: targeted subsampling consistently matched or exceeded the representation quality of full-data pretraining, and in many configurations full-data self-supervision actually fell below the supervised baseline, an adverse effect the selective strategies appeared to mitigate, particularly under label bias.

To understand when and why subsampling pays off, the team conducted a meta-analysis linking dataset characteristics to observed gains. They extracted statistical and complexity-related metafeatures for each dataset — among them the performance of a nearest-neighbor classifier built on the single most important feature, the stability of local neighborhood structure across cross-validation folds, the minimum support of decision-tree leaves, the median absolute deviation of the features, and trimmed means capturing distributional asymmetry. The patterns that emerged were model-specific rather than universal. TabNet and SCARF gained most from subsampling on datasets where a single dominant feature carries the predictive signal, while SAINT and VIME tended to benefit more on structurally complex tasks; datasets containing a feature with strongly asymmetric central mass favored using all the data. No single metafeature predicted gains across all architectures, pointing to a decisive interplay between dataset geometry and each model’s inductive bias.

The authors are candid about limitations and open questions. The number and size of unlabeled subsets and the granularity of the score brackets were fixed by design, so the study does not determine an optimal data budget for a given dataset, and the meta-analysis, based on 25 datasets, is exploratory rather than confirmatory. Future directions include adaptive budget selection guided by validation performance or meta-learned dataset properties, curriculum-style ordering of the unlabeled pool, incorporating self-supervised model feedback into the selection loop, and training a meta-learner to recommend strategy and subset size automatically. Even so, the message for practitioners is already actionable and, in its own way, subversive: in tabular self-supervised learning, the brute-force assumption that more unlabeled data is always better does not survive scrutiny. A few hundred well-chosen — or even randomly chosen — examples can deliver richer representations and better predictions at a fraction of the cost, a result that could reshape how resource-conscious teams approach pretraining on the structured data that quietly runs the world.

Subject of Research: Unlabeled data selection for efficient self-supervised learning on tabular data

Article Title: Less is More: Unlabeled Data Selection for Efficient Tabular Self-supervised Learning

Article References: Stevanoska, S., Camacho Villalón, C. L., Džeroski, S., & Dost, K. (2026). Less is More: Unlabeled Data Selection for Efficient Tabular Self-supervised Learning. Machine Learning, 115(9), Article 208. https://doi.org/10.1007/s10994-026-07133-8

Image Credits: AI Generated

DOI: 10.1007/s10994-026-07133-8

Keywords: self-supervised learning, tabular data, data selection, unlabeled data, active learning, optimal transport, query-by-committee, computational efficiency, representation learning, machine learning benchmarks, TabNet, VIME