Machine learning, an essential subset of artificial intelligence, empowers computers to identify subtle and intricate patterns in data that frequently elude even the most experienced human experts. Yet, despite incredible computational power, computers acquire knowledge very differently from humans. For example, a young child can effortlessly recognize faces and discern unusual events in complex environments with minimal instruction. Conversely, computer vision systems often rely on vast datasets, meticulous annotation, and extensive iterative training. These systems must be rigorously tested to ensure robustness amid changing conditions such as variable lighting, shifting camera perspectives, or environmental noise — a process that can be painstakingly slow but essential for dependable performance.

This paradox — in which machines eventually outperform humans on many perceptual tasks despite requiring significantly more extensive learning — was a central theme in Dr. Abu’s keynote address during the Second Ateneo Breakthroughs lecture held on February 26, 2026. Her presentation, entitled “Smarter Sight: Building Intelligent Visual Systems for Public Good,” not only elucidated the current limitations and challenges of machine learning but also highlighted the critical importance of interdisciplinary collaboration in bridging the gap between theoretical models and real-life applications.

Dr. Abu underscored that the reliability of any machine learning system depends on a fusion of domain expertise and computational rigor. This necessitates close communication between subject matter experts — doctors, urban planners, engineers — and computer scientists who translate complex, often messy real-world environments into mathematical models. Such partnerships ensure that developed algorithms are valid beyond laboratory contexts and maintain high performance in operational settings, where unpredictability and variability are the norm.

Within ALIVE, Dr. Abu and her research team have concentrated efforts on computer vision and image processing driven by deep learning techniques that have broad applications. In healthcare, for instance, they developed a dental imaging support system designed to augment a dentist’s ability to spot diagnostic clues that may be subtle or easily missed during busy clinical routines. Additionally, patch-based deep learning models have been engineered to detect bone metastasis in medical images, providing oncologists with vital, early diagnostic support tools.

Beyond the clinical realm, ALIVE’s innovations extend to public infrastructure through projects such as V-PROBE — a versatile platform that analyzes real-time data on vehicle and pedestrian movement to monitor traffic conditions, anticipate parking availability, and proactively flag emerging congestion risks. By delivering timely insights, V-PROBE has the potential to enhance urban mobility and reduce the often significant socio-economic costs of traffic gridlocks.

The success of these projects hinges on ongoing collaboration with stakeholders who operate within complex, dynamic environments. Algorithmic models cannot remain confined to glorified demonstrations but must adapt and respond to operational realities, ranging from hardware limitations and privacy concerns to the heterogeneity of deployment environments and stringent public expectations around performance and reliability.

ALIVE’s strategic focus has shifted toward deepening ties with industry partners. Such collaborations provide crucial access to large-scale data pipelines and real-world operational settings where ALIVE’s research can be rigorously evaluated against benchmarks of speed, security, robustness, and scalability. Industry experts also inform the teams about end-user needs, enabling a transition from embryonic ideas in research labs to practical innovations that offer tangible benefits.

Dr. Abu’s leadership exemplifies the necessity of integrating artificial intelligence efforts with domain-specific knowledge to engineer intelligent visual systems that serve public interests. By adopting an inclusive, collaborative approach, ALIVE propels machine learning closer to embedding itself effectively in medical diagnostics, urban management, and beyond.

Intrinsically, the story of ALIVE at Ateneo de Manila University reflects a broader narrative unfolding in artificial intelligence research today: the imperative to transcend academic silos and integrate theoretical advances with societal needs. Through innovation, cooperation, and persistent refinement, machine learning systems can be shaped not only to recognize patterns but to respond responsibly within complex human environments, paving the way for smarter, safer, and more adaptive visual technologies.

For specimen demonstrations and to experience the full potential of these advancements, Dr. Patricia Abu’s lecture is available for public viewing at ateneo.edu/breakthroughs. Meanwhile, researchers, industry stakeholders, and media are encouraged to foster dialogue and partnerships by contacting Dr. Abu directly at pabu@ateneo.edu.

As artificial intelligence continues its ascent, the interplay between sophisticated algorithms, domain expertise, and operational realities will define the trajectory of smart visual systems worldwide. ALIVE’s work stands as a beacon of how such interdisciplinary collaboration can unlock the powerful possibilities of machine learning to transform healthcare, urban living, and beyond.

Subject of Research: Interdisciplinary machine learning approaches for computer vision and intelligent visual systems development.

Article Title: Smarter Sight: How Machine Learning is Transforming Visual Intelligence for Public Good

News Publication Date: February 26, 2026

Web References: http://ateneo.edu/breakthroughs, http://archium.ateneo.edu

Image Credits: OAVP-RCWI, 2026

Keywords: machine learning, computer vision, artificial intelligence, interdisciplinary collaboration, healthcare imaging, urban traffic systems, deep learning, intelligent visual systems, pattern recognition, ALIVE, Ateneo de Manila University, real-world AI applications

Flexible electronics, long limited by the trade-offs between durability, stretchability, and bio-compatibility, receive a quantum leap forward through the dual-network nature of this hydrogel. Engineered with an entangled polymer matrix interspersed with liquid metal induction cross-linking, the PGEH material exhibits extraordinary mechanical properties. It withstands elongations of up to 1643% strain and endures tensile stresses as high as 366 kPa. These parameters closely mimic the behavior of natural human skin under deformation, ensuring that the sensor maintains integrity in highly dynamic environments such as joint movements or facial expressions, vital for practical wearable applications.

The unique reversible adhesion mechanism hinges on temperature-activated bonding kinetics. When applied to skin, the patch adheres firmly under human body temperatures ranging from 30 to 40 °C, generating adhesion forces up to 104 kPa. This adhesion is not permanent; it can be gently and painlessly released with a simple rinse of cold water around 10 °C, dramatically reducing trauma and irritation typically associated with adhesive biomedical devices. Moreover, the patch’s reusable adhesion capacity extends beyond 30 cycles without loss of efficacy, heralding a sustainable and user-friendly interface for long-term wear.

Electrochemical performance dramatically elevates the potential of this hydrogel in electrophysiological monitoring. The PGEH capacitive sensor boasts ultralow impedance of approximately 310 ohms at 100 Hz, a significant improvement over conventional silver/silver chloride (Ag/AgCl) electrodes which often degrade within six hours of continuous use. This reduced impedance boosts signal fidelity, evidenced by a high signal-to-noise ratio of 25.2 dB, allowing the capture of subtle EEG voltage variations in the microvolt range over sustained periods of up to 48 hours, an unprecedented benchmark in wearable EEG technology.

Integration of this sensor with artificial intelligence underscores the multidimensional innovation of the system. Utilizing the lightweight deep learning architecture EEGNet, the device classifies cognitive states such as focused attention, distraction, and fatigue with astonishing accuracy surpassing 91%. This real-time monitoring capability paves the way for responsive neurofeedback systems that can adapt user environments or workflows dynamically, holding promise for education, clinical neurorehabilitation, and occupations where sustained attention is critical.

Such a sensor ushers in revolutionary applications beyond traditional EEG recording. The researchers demonstrated encrypted communication via finger-tapping Morse or binary code modulated by changes in capacitance, enabling secure, hands-free messaging paradigms. This creative interface taps into subtle physiological signals for nonverbal communication, potentially transformative in accessibility technologies or covert communications.

Moreover, the sensor’s utility extends to continuous health monitoring, capturing electrocardiogram (ECG) and electromyogram (EMG) signals with clinical-grade fidelity for cardiac and muscular diagnostics. This capability, integrated in a flexible, skin-conforming form factor, facilitates prolonged monitoring periods without the discomfort or skin damage posed by rigid electrodes and bulky cables, signaling a new era in patient-centered healthcare devices.

Underlying the technological triumph is an elegantly engineered material platform. The hydrogel’s entangled network is cross-linked in the presence of eutectic gallium-indium (EGaIn) liquid metal particles, which impart liquid-metal conductivity while maintaining softness and flexibility. This composite synergy allows for the hydrogel to retain high electrical conductance while enduring mechanical deformation and repeated adhesion cycles, a challenge that has stymied the development of prior flexible sensing interfaces.

The mechanical robustness and skin-mimicking elasticity of the PGEH also position it as a comfortable medium for prolonged use. Unlike many biomedical adhesives which irritate or cause allergic reactions upon repeated application, this hydrogel sensor offers a biocompatible alternative with minimal skin irritation and no residue, validated through multiple reuse cycles. This quality, combined with reversible adhesion, streamlines user experience by reducing downtime and barrier to adoption in diverse user populations.

Adding to its versatility, the hydrogel patch is manufactured as an ultrathin film compatible with existing wearable design paradigms. This slim footprint reduces bulk and enhances conformal contact against irregular skin surfaces, optimizing signal acquisition and wearer comfort. It can be fashioned into headbands or patches integrated seamlessly into everyday accessories, blurring the line between medical device and consumer electronics.

Beyond its impressive material and engineering feats, the fusion of AI-driven analytics with such a robust sensor network represents a pivotal paradigm shift. Real-time EEG feedback captured through this device could facilitate individualized cognitive training, fatigue management in high-risk professions such as aviation or transportation, and early detection of neurological abnormalities. These capabilities underscore the hydrogel’s potential impact across healthcare, occupational safety, and cognitive enhancement industries.

In conclusion, the PGEH hydrogel sensor embodies a transformative approach to wearable biomedical technology. By harmonizing remarkable mechanical properties, reversible skin adhesion, ultra-sensitive electrophysiological monitoring, and AI-powered cognitive state classification, this platform breaks longstanding barriers in flexible electronics. As researchers move towards commercialization, this convergence of materials innovation and machine learning could profoundly alter how we monitor, interpret, and interact with human physiology in real-time.

Subject of Research: Experimental study on a machine learning-enabled, reusable adhesion hydrogel for long-term, high-fidelity EEG recording and attention assessment.

Article Title: Machine Learning Enabled Reusable Adhesion, Entangled Network-Based Hydrogel for Long-Term, High-Fidelity EEG Recording and Attention Assessment

News Publication Date: 29-May-2025

Web References: http://dx.doi.org/10.1007/s40820-025-01780-7

Image Credits: Kai Zheng, Chengcheng Zheng, Lixian Zhu, Bihai Yang, Xiaokun Jin, Su Wang, Zikai Song, Jingyu Liu, Yan Xiong, Fuze Tian, Ran Cai, Bin Hu.

Keywords: Hydrogels, Flexible Electronics, EEG Sensor, Machine Learning, Wearable Neurotechnology, Liquid Metal, Reusable Adhesives, Electrophysiological Monitoring, AI Neurofeedback.

Mammography has long served as the frontline screening technique for breast cancer detection, pivotal in reducing mortality rates through early diagnosis. However, interpreting mammograms remains a complex challenge due to the subtle nature of abnormalities that can often evade even the most trained human eyes. In light of this, the integration of sophisticated deep learning models provides a promising solution to augment medical professionals’ capabilities in accurately identifying potential anomalies in radiological images.

The authors embark upon a detailed description of Convolutional Neural Networks, a staple in image recognition tasks. CNNs operate by automatically detecting patterns in image data, leveraging layers of convolutional filters that progressively extract relevant features. This multi-layered architecture allows CNNs to capture intricate details in mammograms, facilitating the identification of varying shapes and textures indicative of malignant lesions. The efficiency in training these networks hinges on their ability to learn directly from pixel data, rendering CNNs a natural fit for image-focused applications.

On the other side of the spectrum lies Vision Transformers, an architecture that has recently gained traction in the field of computer vision. ViTs eschew the traditional convolutional approach in favor of transformer models that were initially designed for natural language processing tasks. By segmenting images into patches and applying self-attention mechanisms, ViTs can learn long-range dependencies within visual data. This novel approach has been shown to provide competitive performance against CNNs while often requiring significantly less data to achieve optimal results.

In the study, Sharma et al. meticulously compare the performance of CNNs and ViTs on a widely recognized dataset comprising mammogram images. Each architecture is subject to rigorous training and validation processes to assess their accuracy, sensitivity, and specificity in classifying mammograms as benign or malignant. Their findings illustrate that while both models exhibit commendable performance, nuanced differences emerge, particularly under varied conditions prevalent in clinical environments.

One significant aspect of the study is the exploration of transfer learning—a method of reusing a pre-trained model on a new problem. The authors detail how both CNNs and ViTs benefit from transfer learning, as it allows them to leverage existing knowledge from large datasets, alleviating the need for vast labeled datasets in mammography specifically. The implications of this are profound, especially in scenarios where acquiring and annotating medical images can be labor-intensive and costly.

As healthcare continues to digitize, the role of deep learning in interpreting medical images is elevating patient care. Automated classification systems powered by these neural network architectures can speed up the diagnostic process, allowing radiologists to focus their expertise on cases that require more intensive review. The value of rapid, accurate assessments cannot be overstated, particularly in reducing the anxiety that often accompanies waiting for test results.

Moreover, this comparative analysis sheds light on the computational demands of each model. While CNNs are generally less resource-intensive and quicker to train, ViTs may offer greater accuracy despite their need for more computational power and time. The balance between performance and resource allocation becomes a vital consideration for medical institutions, particularly in low-resource settings where every bit of computational efficiency counts.

The authors further discuss the ethical implications of deploying these technologies in clinical practice. As AI models begin to take on greater roles in diagnostic procedures, the onus remains on developers and healthcare providers to ensure that these systems are trained on diverse datasets. Minimizing biases in training data is crucial to avoid disparities in diagnostic accuracy across different demographic groups.

As the medical community grapples with these innovations, the ongoing development of deep learning models represents a paradigm shift in radiological practices. The findings presented by Sharma, Singh, and Choudhury serve to spark conversations about the future of AI in diagnostics and how these advancements can be harnessed to improve patient outcomes. The potential for increased early detection rates and, consequently, improved survival rates from breast cancer is an exciting prospect that warrants further exploration.

In conclusion, the research conducted by Sharma et al. showcases the power of deep learning in revolutionizing mammography classification. The contrasting performances of CNNs and ViTs highlight the importance of ongoing investigations into machine learning frameworks that can bolster diagnostic accuracy. As researchers continue to push these boundaries, the convergence of artificial intelligence and healthcare holds the promise of reshaping the landscape of medical diagnostics for decades to come.

Subject of Research: Deep learning architectures for mammography classification

Article Title: Advanced deep learning architectures for enhanced mammography classification: a comparative study of CNNs and ViT

Article References: Sharma, S., Singh, Y. & Choudhury, T. Advanced deep learning architectures for enhanced mammography classification: a comparative study of CNNs and ViT. Discov Artif Intell 5, 187 (2025). https://doi.org/10.1007/s44163-025-00426-2

Image Credits: AI Generated

DOI: 10.1007/s44163-025-00426-2

Keywords: deep learning, mammography classification, CNNs, Vision Transformers, artificial intelligence.