Semi-supervised Deep Linear Discriminant Analysis for Histopathology Image Classification

Deep learning has shown remarkable potential for complex signal recognition, yet its deployment in noisy real-world environments remains a major challenge. This study presents an adaptive deep learning framework for acoustic signal detection, demonstrated through the identification of cetacean whistles in the Strait of Gibraltar, a mandatory migratory bottleneck where several species coexist with intense maritime traffic and complex acoustic conditions. Passive acoustic monitoring (PAM) enables long-term assessment of marine mammal presence and behavior, yet its application remains challenging due to overlapping anthropogenic and environmental sounds and the large volumes of data generated. The proposed framework combines transfer learning, semi-supervised iterative fine-tuning, and confidence-threshold calibration to automate the analysis of PAM data and improve robustness in acoustically variable conditions. Pre-trained bioacoustic models (BirdNET and Perch) were repurposed as feature extractors and coupled with custom classifiers to assess generalization across domains. Model performance was evaluated using standard deep learning metrics, including accuracy, recall, and F1-score. Models were validated using a clean benchmark dataset and a real-world deployment dataset characterized by substantial anthropogenic and geophonic noise. While baseline models achieved over 0.95 accuracy on clean data, their performance degraded under realistic noise conditions (whistle F1-score ≤ 0.10). The best-performing model, fine-tuned with local data and semi-supervised iterative validation, achieved an F1-score of 0.88 and improved recall across deployments. Confidence-threshold optimization further enhanced adaptability, supporting both automated and expert-assisted monitoring workflows. The final model was tested across three independent acoustic deployments spanning different seasons and locations. Results demonstrated robust whistle detection with minimal annotation effort. This work provides a reproducible, AI-driven framework for signal detection in noisy environments, demonstrating how adaptive learning and threshold calibration can extend the applicability of deep neural networks to real-world acoustic monitoring and contribute to the development of automated, ecologically meaningful soundscape analysis systems. • A semi-supervised deep learning framework is presented for acoustic monitoring. • Integrated BirdNET and Perch embeddings with iterative expert validation. • Adaptive threshold optimization (0.79–0.84) balances automation with validation. • Model-assisted annotation reduces labeling effort while preserving quality over 1319 h. • Scalability of an AI-based approach is assessed in a high-impact PAM context.

A semi-supervised deep learning approach for cropped image detection

Automated image classification is an essential task of the computer vision field. The tagging of images into a set of predefined groups is referred to as image classification. The implementation of computer vision to automate image classification would be beneficial because manual image evaluation and identification can be time-consuming, particularly when there are many images of different classes. Deep learning approaches are proven to overperform existing machine learning techniques in many fields in recent years, and computer vision is one of the most notable examples. The very deep neural network (VDCNN) is a powerful deep learning model for image classification, and this paper examines it briefly using MNIST handwritten digit dataset. This dataset is used to prove the efficacy of very deep neural networks over other deep learning models. The proposed study aims to comprehend the very deep neural network architecture used to accomplish a handwritten digit recognition task. The feasibility of the proposed model is evaluated using mean accuracy, validation accuracy, and standard deviation. The study results of the very deep neural network model are compared to a convolutional neural network and convolutional neural network with batch normalization. According to the results of the comparison study, very deep neural networks achieve high accuracy of 99.1% for a handwritten dataset. The outcome of the proposed work is used to interpret how well a very deep neural network performs when compared to the other two models of deep neural networks. This proposed architecture may be used to automate the classification of handwritten digits dataset.

This paper proposes a fault diagnosis method for air handling units (AHUs) that addresses the challenges of limited labeled data, high-dimensional sensor features, and device heterogeneity. The technique consists of a two-stage deviceindependent feature selection strategy and a semi-supervised deep learning framework (SS-TCN). The feature selection strategy utilizes multiple ensemble models and SHAP values to identify key sensor features. The SS-TCN framework employs a teacher-student network based on Temporal Convolutional Networks (TCN) for semi-supervised learning, generating pseudo-labels for unlabeled data via an Exponential Moving Average (EMA) update. Experimental results on the ASHRAE RP-1312 dataset demonstrate that the proposed method achieves a high weighted F1 score of 0.962 using only 10% of the labeled data, while reducing feature dimensions by 64.7 %, showing significant effectiveness and efficiency.

Learning ability is the basic characteristic of human intelligence. The July 1, 2005 issue of Science published a list of 125 important questions in science. Among them, the question 94 “What are the limits of learning by machines?”. The annotation “Computers can already beat the world’s best chess players, and they have a wealth of information on the Web to draw on. But abstract reasoning is still beyond any machine”. In recent artificial intelligence has made great progresses. In 1997, the rise of the man-machine war, IBM Supercomputer Deep Blue defeated the chess master Garry Kasparov. On February 14, 2011, IBM’s Watson supercomputer won a practice round against Jeopardy champions Ken Jennings and Brad Rutter. In March 2016, Google DeepMind’s AlphaGo sealed a 4-1 victory over a South Korean Go grandmaster Lee Se-dol. This paper focuses on the machine learning methods of AlphaGo, including reinforcement learning, deep learning, deep reinforcement learning, analysis of the existing problems and the latest research progress. Deep reinforcement learning is the combination of deep learning and reinforcement learning, which can realize the learning algorithm from the perception to action. Simply said, this is the same as human behavior, input sensing information such as vision, and then, direct output action through the deep neural network. Deep reinforcement learning has the potential to learn a variety of skills for the robot to achieve full autonomy. Even though reinforcement learning is practiced successfully, but feature states need to manually set, for complex scene is a difficult thing, especially easy to cause the dimension disaster, and expression is not good. In 2010, Sascha Lange and Martin Riedmiller proposed deep auto-encoder neural networks in reinforcement learning to extract feature, which is used to control the visual correlation. In 2013, DeepMind proposed deep Q-network (DQN) in NIPS 2013, using convolution neural network to extract features, and then applied in reinforcement learning. They continue to improve and published an improved version of DQN on Nature in 2015, which has aroused widespread concern. In order to break through the limits of learning by machines, cognitive machine learning is proposed, which is the combination of machine learning and brain cognition, so that the machine intelligence is constantly evolving, and gradually reaches the human level of artificial intelligence. A cognitive model entitled Consciousness And Memory (CAM) is proposed by author, which consists of memory, consciousness, high-level cognitive functions, perception and motor. High-level cognitive functions of the brain include learning, language, thinking, decision making, emotion, and so on. Learning is a course to accept the stimulus through the nervous system and obtain new behavior, habits and accumulation experience. According to the current research progress of brain science and cognitive science, cognitive machine learning may be interested in learning emergence, procedural memory knowledge learning, learning evolution and so on. For intelligence, so-called evolution is refers to the learning of learning and the structure also follows the change. It is important to record the learning result by structure changing and improve the learning method.

Convolutional neural networks (CNNs) are state-of-the-art computer vision techniques for various tasks, particularly for image classification. However, there are domains where the training of classification models that generalize on several datasets is still an open challenge because of the highly heterogeneous data and the lack of large datasets with local annotations of the regions of interest, such as histopathology image analysis. Histopathology concerns the microscopic analysis of tissue specimens processed in glass slides to identify diseases such as cancer. Digital pathology concerns the acquisition, management and automatic analysis of digitized histopathology images that are large, having in the order of 100′0002 pixels per image. Digital histopathology images are highly heterogeneous due to the variability of the image acquisition procedures. Creating locally labeled regions (required for the training) is time-consuming and often expensive in the medical field, as physicians usually have to annotate the data. Despite the advances in deep learning, leveraging strongly and weakly annotated datasets to train classification models is still an unsolved problem, mainly when data are very heterogeneous. Large amounts of data are needed to create models that generalize well. This paper presents a novel approach to train CNNs that generalize to heterogeneous datasets originating from various sources and without local annotations. The data analysis pipeline targets Gleason grading on prostate images and includes two models in sequence, following a teacher/student training paradigm. The teacher model (a high-capacity neural network) automatically annotates a set of pseudo-labeled patches used to train the student model (a smaller network). The two models are trained with two different teacher/student approaches: semi-supervised learning and semi-weekly supervised learning. For each of the two approaches, three student training variants are presented. The baseline is provided by training the student model only with the strongly annotated data. Classification performance is evaluated on the student model at the patch level (using the local annotations of the Tissue Micro-Arrays Zurich dataset) and at the global level (using the TCGA-PRAD, The Cancer Genome Atlas-PRostate ADenocarcinoma, whole slide image Gleason score). The teacher/student paradigm allows the models to better generalize on both datasets, despite the inter-dataset heterogeneity and the small number of local annotations used. The classification performance is improved both at the patch-level (up to κ=0.6127±0.0133 from κ=0.5667±0.0285), at the TMA core-level (Gleason score) (up to κ=0.7645±0.0231 from κ=0.7186±0.0306) and at the WSI-level (Gleason score) (up to κ=0.4529±0.0512 from κ=0.2293±0.1350). The results show that with the teacher/student paradigm, it is possible to train models that generalize on datasets from entirely different sources, despite the inter-dataset heterogeneity and the lack of large datasets with local annotations.

Every year, millions of women across the globe are diagnosed with breast cancer (BC), an illness that is both common and potentially fatal. To provide effective therapy and enhance patient outcomes, it is essential to make an accurate diagnosis as soon as possible. In recent years, deep-learning (DL) approaches have shown great effectiveness in a variety of medical imaging applications, including the processing of histopathological images. Using DL techniques, the objective of this study is to recover the detection of BC by merging qualitative and quantitative data. Using deep mutual learning (DML), the emphasis of this research was on BC. In addition, a wide variety of breast cancer imaging modalities were investigated to assess the distinction between aggressive and benign BC. Based on this, deep convolutional neural networks (DCNNs) have been established to assess histopathological images of BC. In terms of the Break His-200×, BACH, and PUIH datasets, the results of the trials indicate that the level of accuracy achieved by the DML model is 98.97%, 96.78, and 96.34, respectively. This indicates that the DML model outperforms and has the greatest value among the other methodologies. To be more specific, it improves the results of localization without compromising the performance of the classification, which is an indication of its increased utility. We intend to proceed with the development of the diagnostic model to make it more applicable to clinical settings.

Gait recognition is a noncontact biometric procedure that determines the identity or health status of a person by analyzing his or her walking posture and habits, including skeletal and joint movements. The most remarkable feature of this method is the possibility of conducting recognition without demanding much cooperation from participants. Therefore, this recognition technique has attracted much attention from scholars. Additionally, because of the rapid development of graphics processing unit technology, related hardware and computation performance, the applications of deep-learning technology are considerably enhanced. The objective of this study was to apply a deep neural network (DNN), which employs deep-learning technology, to achieve gait-based automatic pedestrian detection and recognition. In contrast to using wearable devices to precisely capture skeletal and joint movements, pedestrian color-image sequences were used as input in this study. Subsequently, a pretraining convolutional neural network (CNN) was employed to capture pedestrian location and extract pedestrian dense optical flow to serve as concrete low-level feature inputs. Then, a finely-tuned DNN based on the wide residual network was employed to extract high-level abstract features. In addition, to overcome the difficulty of obtaining local temporal features by using a 2D CNN, part of the 3D convolutional structure was introduced into the CNN. This design enabled use of limited memory to acquire more effective features and enhance the DNN performance. The experimental results show that the proposed method has exceptional performance for pedestrian detection and recognition.

Deep Learning, developed from the multi-layer ANN concept, is a machine learning method that compiles every detail of the learning process to obtain more abstract data, multi-level data, and more complex data features through the composition of various mathematical functions. Deep MLP, another name for DNN (Deep Neural Network), is MLP, which has more than three layers. DNN's ability will increase in solving problems when more layers are used. Another advantage of DNN is the variety of layer types used, including fully connected layers, convolution layers, softmax layers, recurrent layers, and others. Autoencoder is a type of ANN that trained to reconstruct input patterns in such a way that the output of the deepest hidden layer is a vector resulting from the reduction of dimensions of the input pattern. This study proposes a novel DNN architecture called Deep Auto-Encoder Semi Convolutional Neural Network (DAESCNN) to improve the performance of conventional ANN performance. Annual rainfall data will be used to test forecast performance using DAESCNN. In this study, annual rainfall data were obtained from weather stations of Samarinda city, East Kalimantan, Indonesia in the period 2006–2016. In this study, all 2006–2014 data points were used as training input data, all 2007–2015 data points were used as training target data, while 2016 data points were used as validation of training outcomes. The DAESCNN training process executed through a program built using MATLAB. The training process also carried out with varying training parameters to show its performance. The results shown that DAESCNN generally has excellent performance, is above 99%.

Due to the major operating restrictions, ensuring security is the fundamental problem of Wireless Sensor Networks (WSNs). Because of their inadequate security mechanisms, WSNs are indeed a simple point for malware (worms, viruses, malicious code, etc.). According to the epidemic nature of worm propagation, it is critical to develop a worm defense mechanism in the network. This concept aims to establish novel malware detection in WSN that consists of several phases: “(i) Preprocessing, (ii) feature extraction, as well as (iii) detection”. At first, the input data is subjected for preprocessing phase. Then, the feature extraction takes place, in which principal component analysis (PCA), improved linear discriminant analysis (LDA), and autoencoder-based characteristics are retrieved. Moreover, the retrieved characteristics are subjected to the detection phase. The detection is performed employing combined shallow learning and DL. Further, the shallow learning includes decision tree (DT), logistic regression (LR), and Naive Bayes (NB); the deep learning (DL) includes deep neural network (DNN), convolutional neural network (CNN), and recurrent neural network (RNN). Here, the DT output is given to the DNN, LR output is subjected to CNN, and the NB output is given to the RNN, respectively. Eventually, the DNN, CNN, and RNN outputs are averaged to generate a successful outcome. The combination can be thought of as an Ensemble classifier. The weight of the RNN is optimally tuned through the Self Improved Shark Smell Optimization with Opposition Learning (SISSOOL) model to improve detection precision and accuracy. Lastly, the outcomes of the suggested approach are computed in terms of different measures.

Congenital adrenal hyperplasia (CAH) is the most common primary adrenal insufficiency in children, involving excess androgens secondary to disrupted steroidogenesis as early as the seventh gestational week of life. Although structural brain abnormalities are seen in CAH, little is known about facial morphology. To investigate differences in facial morphologic features between patients with CAH and control individuals with use of machine learning. This cross-sectional study was performed at a pediatric tertiary center in Southern California, from November 2017 to December 2019. Patients younger than 30 years with a biochemical diagnosis of classical CAH due to 21-hydroxylase deficiency and otherwise healthy controls were recruited from the clinic, and face images were acquired. Additional controls were selected from public face image data sets. The main outcome was prediction of CAH, as performed by machine learning (linear discriminant analysis, random forests, deep neural networks). Handcrafted features and learned representations were studied for CAH score prediction, and deformation analysis of facial landmarks and regionwise analyses were performed. A 6-fold cross-validation strategy was used to avoid overfitting and bias. The study included 102 patients with CAH (62 [60.8%] female; mean [SD] age, 11.6 [7.1] years) and 59 controls (30 [50.8%] female; mean [SD] age, 9.0 [5.2] years) from the clinic and 85 controls (48 [60%] female; age, <29 years) from face databases. With use of deep neural networks, a mean (SD) AUC of 92% (3%) was found for accurately predicting CAH over 6 folds. With use of classical machine learning and handcrafted facial features, mean (SD) AUCs of 86% (5%) in linear discriminant analysis and 83% (3%) in random forests were obtained for predicting CAH over 6 folds. There was a deviation of facial features between groups using deformation fields generated from facial landmark templates. Regionwise analysis and class activation maps (deep learning of regions) revealed that the nose and upper face were most contributory (mean [SD] AUC: 69% [17%] and 71% [13%], respectively). The findings suggest that facial morphologic features in patients with CAH is distinct and that deep learning can discover subtle facial features to predict CAH. Longitudinal study of facial morphology as a phenotypic biomarker may help expand understanding of adverse lifespan outcomes for patients with CAH.

Deep learning is being incorporated in many modern software systems. Deep learning approaches train a deep neural network (DNN) model using training examples, and then use the DNN model for prediction. While the structure of a DNN model as layers is observable, the model is treated in its entirety as a monolithic component. To change the logic implemented by the model, e.g. to add/remove logic that recognizes inputs belonging to a certain class, or to replace the logic with an alternative, the training examples need to be changed and the DNN needs to be retrained using the new set of examples. We argue that decomposing a DNN into DNN modules— akin to decomposing a monolithic software code into modules—can bring the benefits of modularity to deep learning. In this work, we develop a methodology for decomposing DNNs for multi-class problems into DNN modules. For four canonical problems, namely MNIST, EMNIST, FMNIST, and KMNIST, we demonstrate that such decomposition enables reuse of DNN modules to create different DNNs, enables replacement of one DNN module in a DNN with another without needing to retrain. The DNN models formed by composing DNN modules are at least as good as traditional monolithic DNNs in terms of test accuracy for our problems.

Printed Circuit Board (PCB) defect inspection is critical in electronics manufacturing, yet annotating large training sets is labor-intensive and costly. This paper proposes a semi-supervised PCB defect detection framework based on YOLOv5, which leverages a small labeled dataset and a larger pool of unlabeled images through an iterative self-training pipeline. A YOLOv5 detector is first trained on the limited labeled data, then used to generate pseudo-labels on unlabeled images; high-confidence detections are retained as defect annotations, and the model is retrained on the combination of labeled and pseudo-labeled data. This pseudo-labeling and retraining cycle is repeated for multiple iterations to progressively refine the detector. The approach is evaluated on a PCB defect dataset with 100 labeled and 1,000 unlabeled images, and shows significant gains over a fully supervised baseline. The proposed semi-supervised YOLOv5 achieves 91.6% mAP@0.5 with only 100 labeled images, outperforming both the baseline (87.0% mAP) and prior semi-supervised methods, while substantially improving recall and precision. The results demonstrate that the method effectively reduces annotation effort while maintaining high detection accuracy, providing a simple, confidence-thresholded self-training strategy for deploying PCB defect detectors under limited labeling resources. This work directly supports SDG 9 by enabling cost-efficient, high-accuracy AI-based PCB inspection that strengthens intelligent and sustainable manufacturing systems.

Deep neural networks (DNNs) have achieved unprecedented success across many scientific and engineering fields in the last decades. Despite its empirical success, unfortunately, recent studies have shown that there are various failure modes and blindspots in DNN models which may result in unexpected serious failures and potential harms, e.g. the existence of adversarial examples and small perturbations. This is not acceptable especially for safety critical and high stakes applications in the real-world, including healthcare, self-driving cars, aircraft control systems, hiring and malware detection protocols. Moreover, it has been challenging to understand why and when DNNs will fail due to their complicated structures and black-box behaviors. Lacking interpretability is one critical issue that may seriously hinder the deployment of DNNs in high-stake applications, which need interpretability to trust the prediction, to understand potential failures, and to be able to mitigate harms and eliminate biases in the model. To make DNNs trustworthy and reliable for deployment, it is necessary and urgent to develop methods and tools that can (i) quantify and improve their robustness against adversarial and natural perturbations, and (ii) understand their underlying behaviors and further correct errors to prevent injuries and damages. These are the important first steps to enable Trustworthy AI and Trustworthy Machine Learning. In this talk, I will survey a series of research efforts in my lab contributed to tackling the grand challenges in (i) and (ii). In the first part of my talk, I will overview our research effort in Robust Machine Learning since 2017, where we have proposed the first attack-agnostic robustness evaluation metric, the first efficient robustness certification algorithms for various types of perturbations, and efficient robust learning algorithms across supervised learning to deep reinforcement learning. In the second part of my talk, I will survey a series of exciting results in my lab on accelerating interpretable machine learning and explainable AI. Specifically, I will show how we could bring interpretability into deep learning by leveraging recent advances in multi-modal models. I'll present recent works in our group on automatically dissecting neural networks with open vocabulary concepts, designing interpretable neural networks without concept labels, and briefly overview our recent efforts on demystifying black-box DNN training process, automated neuron explanations for Large Language Models and the first robustness evaluation of a family of neuron-level interpretation techniques.

In a previous paper, we introduced a deep learning neural network that should be able to detect the existence of very shallow periodic planetary transits in the presence of red noise. The network in that feasibility study would not provide any further details about the detected transits. The current paper completes this missing part. We present a neural network that tags samples that were obtained during transits. This is essentially similar to the task of identifying the semantic context of each pixel in an image—an important task in computer vision, called “semantic segmentation,” which is often performed by deep neural networks. The neural network we present makes use of novel deep learning concepts such as U-Nets, Generative Adversarial Networks, and adversarial loss. The resulting segmentation should allow further studies of the light curves that are tagged as containing transits. This approach toward the detection and study of very shallow transits is bound to play a significant role in future space-based transit surveys such as PLATO, which are specifically aimed to detect those extremely difficult cases of long-period shallow transits. Our segmentation network also adds to the growing toolbox of deep learning approaches that are being increasingly used in the study of exoplanets; but, so far mainly for vetting transits, rather than their initial detection.