Recent advances in self-supervised learning have attracted significant attention from both machine learning and neuroscience. This is primarily because self-supervised methods do not require annotated supervisory information, making them applicable to training artificial networks without relying on large amounts of curated data, and potentially offering insights into how the brain adapts to its environment in an unsupervised manner. Although several previous studies have elucidated the correspondence between neural representations in deep convolutional neural networks (DCNNs) and biological systems, the extent to which unsupervised or self-supervised learning can explain the human-like acquisition of categorically structured information remains less explored. In this study, we investigate the correspondence between the internal representations of DCNNs trained using a self-supervised contrastive learning algorithm and human semantics and recognition. To this end, we employ a few-shot learning evaluation procedure, which measures the ability of DCNNs to recognize novel concepts from limited exposure, to examine the inter-categorical structure of the learned representations. Two comparative approaches are used to relate the few-shot learning outcomes to human semantics and recognition, with results suggesting that the representations acquired through contrastive learning are well aligned with human cognition. These findings underscore the potential of self-supervised contrastive learning frameworks to model learning mechanisms similar to those of the human brain, particularly in scenarios where explicit supervision is unavailable, such as in human infants prior to language acquisition.
The rapid growth of unlabeled time-series data in domains such as wireless communications, radar, biomedical engineering, and the Internet of Things (IoT) has driven advancements in unsupervised learning. This review synthesizes recent progress in applying autoencoders and vision transformers for unsupervised signal analysis, focusing on their architectures, applications, and emerging trends. We explore how these models enable feature extraction, anomaly detection, and classification across diverse signal types, including electrocardiograms, radar waveforms, and IoT sensor data. The review highlights the strengths of hybrid architectures and self-supervised learning, while identifying challenges in interpretability, scalability, and domain generalization. By bridging methodological innovations and practical applications, this work offers a roadmap for developing robust, adaptive models for signal intelligence.
Raman spectroscopy serves as a powerful and reliable tool for analyzing the chemical information of substances. The integration of Raman spectroscopy with deep learning methods enables rapid qualitative and quantitative analysis of materials. Most existing approaches adopt supervised learning methods. Although supervised learning has achieved satisfactory accuracy in spectral analysis, it is still constrained by costly and limited well-annotated spectral datasets for training. When spectral annotation is challenging or the amount of annotated data is insufficient, the performance of supervised learning in spectral material identification declines. In order to address the challenge of feature extraction from unannotated spectra, we propose a self-supervised learning paradigm for Raman Spectroscopy based on a Masked AutoEncoder, termed SMAE. SMAE does not require any spectral annotations during pre-training. By randomly masking and then reconstructing the spectral information, the model learns essential spectral features. The reconstructed spectra exhibit certain denoising properties, improving the signal-to-noise ratio (SNR) by more than twofold. Utilizing the network weights obtained from masked pre-training, SMAE achieves clustering accuracy of over 80% for 30 classes of isolated bacteria in a pathogenic bacterial dataset, demonstrating significant improvements compared to classical unsupervised methods and other state-of-the-art deep clustering methods. After fine-tuning the network with a limited amount of annotated data, SMAE achieves an identification accuracy of 83.90% on the test set, presenting competitive performance against the supervised ResNet (83.40%).
Magnetic resonance imaging (MRI) reconstruction has largely been dominated by deep neural networks (DNN); however, many state-of-the-art architectures use black-box structures, which hinder interpretability and improvement. Here, we propose an interpretable DNN architecture for self-supervised MRI reconstruction and denoising by directly parameterizing and learning the classical primal-dual splitting, dubbed LPDSNet. This splitting algorithm allows us to decouple the observation model from the signal prior. Experimentally, we show other interpretable architectures without this decoupling property exhibit failure in the self-supervised learning regime. We report state-of-the-art self-supervised joint MRI reconstruction and denoising performance and novel noise-level generalization capabilities, where in contrast black-box networks fail to generalize.
Accurate prediction of material properties facilitates the discovery of novel materials with tailored functionalities. Deep learning models have recently shown superior accuracy and flexibility in capturing structure-property relationships. However, these models often rely on supervised learning, which requires large, well-annotated datasets an expensive and time-consuming process. Self-supervised learning (SSL) offers a promising alternative by pretraining on large, unlabeled datasets to develop foundation models that can be fine-tuned for material property prediction. In this work, we propose supervised pretraining, where available class information serves as surrogate labels to guide learning, even when downstream tasks involve unrelated material properties. We evaluate this strategy on two state-of-the-art SSL models and introduce a novel framework for supervised pretraining. To further enhance representation learning, we propose a graph-based augmentation technique that injects noise to improve robustness without structurally deforming material graphs. The resulting foundation models are fine-tuned for six challenging material property predictions, achieving significant performance gains over baselines, ranging from 2% to 6.67% improvement in mean absolute error (MAE) and establishing a new benchmark in material property prediction. This study represents the first exploration of supervised pertaining with surrogate labels in material property prediction, advancing methodology and application in the field.
Full waveform inversion (FWI) updates the velocity model by minimizing the discrepancy between observed and simulated data. However, discretization errors in numerical modeling and incomplete seismic data acquisition can introduce noise, which propagates through the adjoint operator and affects the accuracy of the velocity gradient, thereby impacting the FWI inversion accuracy. To mitigate the influence of noise on the gradient, we employ a convolutional neural network (CNN) to refine the velocity model before performing the forward simulation, aiming to reduce noise and provide a more accurate velocity update direction. We use the same data misfit loss to update both the velocity and network parameters, thereby forming a self-supervised learning procedure. We propose two implementation schemes, which differ in whether the velocity update passes through the CNN. In both methodologies, the velocity representation is extended (VRE) by using a neural network in addition to the grid-based velocities. Thus, we refer to this general approach as VRE-FWI. Synthetic and real data tests demonstrate that the proposed VRE-FWI achieves higher velocity inversion accuracy compared to traditional FWI, at a marginal additional computational cost of approximately 1%.
User sequence modeling is crucial for modern large-scale recommendation systems, as it enables the extraction of informative representations of users and items from their historical interactions. These user representations are widely used for a variety of downstream tasks to enhance users' online experience. A key challenge for learning these representations is the lack of labeled training data. While self-supervised learning (SSL) methods have emerged as a promising solution for learning representations from unlabeled data, many existing approaches rely on extensive negative sampling, which can be computationally expensive and may not always be feasible in real-world scenario. In this work, we propose an adaptation of Barlow Twins, a state-of-the-art SSL methods, to user sequence modeling by incorporating suitable augmentation methods. Our approach aims to mitigate the need for large negative sample batches, enabling effective representation learning with smaller batch sizes and limited labeled data. We evaluate our method on the MovieLens-1M, MovieLens-20M, and Yelp datasets, demonstrating that our method consistently outperforms the widely-used dual encoder model across three downstream tasks, achieving an 8%-20% improvement in accuracy. Our findings underscore the effectiveness of our approach in extracting valuable sequence-level information for user modeling, particularly in scenarios where labeled data is scarce and negative examples are limited.
This dissertation establishes the contexture theory to mathematically characterize the mechanism of representation learning, or pretraining. Despite the remarkable empirical success of foundation models, it is not very clear what representations they learn, and why these representations are useful for various downstream tasks. A scientific understanding of representation learning is critical, especially at this point when scaling up the model size is producing diminishing returns, and designing new pretraining methods is imperative for further progress. Prior work treated different representation learning methods quite differently, whereas the contexture theory provides a unified framework for analyzing these methods. The central argument is that a representation is learned from the association between the input X and a context variable A. We prove that if an encoder captures the maximum information of this association, in which case we say that the encoder learns the contexture, then it will be optimal on the class of tasks that are compatible with the context. We also show that a context is the most useful when the association between X and A is neither too strong nor too weak. The important implication of the contexture theory is that increasing the model size alone will achieve diminishing returns, and further advancements require better contexts. We demonstrate that many pretraining objectives can learn the contexture, including supervised learning, self-supervised learning, generative models, etc. Then, we introduce two general objectives -- SVME and KISE, for learning the contexture. We also show how to mix multiple contexts together, an effortless way to create better contexts from existing ones. Then, we prove statistical learning bounds for representation learning. Finally, we discuss the effect of the data distribution shift from pretraining to the downstream task.
Speaker diarization is a fundamental task in speech processing that involves dividing an audio stream by speaker. Although state-of-the-art models have advanced performance in high-resource languages, low-resource languages such as Kurdish pose unique challenges due to limited annotated data, multiple dialects and frequent code-switching. In this study, we address these issues by training the Wav2Vec 2.0 self-supervised learning model on a dedicated Kurdish corpus. By leveraging transfer learning, we adapted multilingual representations learned from other languages to capture the phonetic and acoustic characteristics of Kurdish speech. Relative to a baseline method, our approach reduced the diarization error rate by seven point two percent and improved cluster purity by thirteen percent. These findings demonstrate that enhancements to existing models can significantly improve diarization performance for under-resourced languages. Our work has practical implications for developing transcription services for Kurdish-language media and for speaker segmentation in multilingual call centers, teleconferencing and video-conferencing systems. The results establish a foundation for building effective diarization systems in other understudied languages, contributing to greater equity in speech technology.
Ego-motion estimation is vital for drones when flying in GPS-denied environments. Vision-based methods struggle when flight speed increases and close-by objects lead to difficult visual conditions with considerable motion blur and large occlusions. To tackle this, vision is typically complemented by state estimation filters that combine a drone model with inertial measurements. However, these drone models are currently learned in a supervised manner with ground-truth data from external motion capture systems, limiting scalability to different environments and drones. In this work, we propose a self-supervised learning scheme to train a neural-network-based drone model using only onboard monocular video and flight controller data (IMU and motor feedback). We achieve this by first training a self-supervised relative pose estimation model, which then serves as a teacher for the drone model. To allow this to work at high speed close to obstacles, we propose an improved occlusion handling method for training self-supervised pose estimation models. Due to this method, the root mean squared error of resulting odometry estimates is reduced by an average of 15%. Moreover, the student neural drone model can be successfully obtained from the onboard data. It even becomes more accurate at higher speeds compared to its teacher, the self-supervised vision-based model. We demonstrate the value of the neural drone model by integrating it into a traditional filter-based VIO system (ROVIO), resulting in superior odometry accuracy on aggressive 3D racing trajectories near obstacles. Self-supervised learning of ego-motion estimation represents a significant step toward bridging the gap between flying in controlled, expensive lab environments and real-world drone applications. The fusion of vision and drone models will enable higher-speed flight and improve state estimation, on any drone in any environment.
We present Variational Self-Supervised Learning (VSSL), a novel framework that combines variational inference with self-supervised learning to enable efficient, decoder-free representation learning. Unlike traditional VAEs that rely on input reconstruction via a decoder, VSSL symmetrically couples two encoders with Gaussian outputs. A momentum-updated teacher network defines a dynamic, data-dependent prior, while the student encoder produces an approximate posterior from augmented views. The reconstruction term in the ELBO is replaced with a cross-view denoising objective, preserving the analytical tractability of Gaussian KL divergence. We further introduce cosine-based formulations of KL and log-likelihood terms to enhance semantic alignment in high-dimensional latent spaces. Experiments on CIFAR-10, CIFAR-100, and ImageNet-100 show that VSSL achieves competitive or superior performance to leading self-supervised methods, including BYOL and MoCo V3. VSSL offers a scalable, probabilistically grounded approach to learning transferable representations without generative reconstruction, bridging the gap between variational modeling and modern self-supervised techniques.
Pancreatic cancer, characterized by its notable prevalence and mortality rates, demands accurate lesion delineation for effective diagnosis and therapeutic interventions. The generalizability of extant methods is frequently compromised due to the pronounced variability in imaging and the heterogeneous characteristics of pancreatic lesions, which may mimic normal tissues and exhibit significant inter-patient variability. Thus, we propose a generalization framework that synergizes pixel-level classification and regression tasks, to accurately delineate lesions and improve model stability. This framework not only seeks to align segmentation contours with actual lesions but also uses regression to elucidate spatial relationships between diseased and normal tissues, thereby improving tumor localization and morphological characterization. Enhanced by the reciprocal transformation of task outputs, our approach integrates additional regression supervision within the segmentation context, bolstering the model's generalization ability from a dual-task perspective. Besides, dual self-supervised learning in feature spaces and output spaces augments the model's representational capability and stability across different imaging views. Experiments on 594 samples composed of three datasets with significant imaging differences demonstrate that our generalized pancreas segmentation results comparable to mainstream in-domain validation performance (Dice: 84.07%). More importantly, it successfully improves the results of the highly challenging cross-lesion generalized pancreatic cancer segmentation task by 9.51%. Thus, our model constitutes a resilient and efficient foundational technological support for pancreatic disease management and wider medical applications. The codes will be released at https://github.com/SJTUBME-QianLab/Dual-Task-Seg.
Multiple Instance Learning (MIL) has emerged as the best solution for Whole Slide Image (WSI) classification. It consists of dividing each slide into patches, which are treated as a bag of instances labeled with a global label. MIL includes two main approaches: instance-based and embedding-based. In the former, each patch is classified independently, and then the patch scores are aggregated to predict the bag label. In the latter, bag classification is performed after aggregating patch embeddings. Even if instance-based methods are naturally more interpretable, embedding-based MILs have usually been preferred in the past due to their robustness to poor feature extractors. However, recently, the quality of feature embeddings has drastically increased using self-supervised learning (SSL). Nevertheless, many authors continue to endorse the superiority of embedding-based MIL. To investigate this further, we conduct 710 experiments across 4 datasets, comparing 10 MIL strategies, 6 self-supervised methods with 4 backbones, 4 foundation models, and various pathology-adapted techniques. Furthermore, we introduce 4 instance-based MIL methods never used before in the pathology domain. Through these extensive experiments, we show that with a good SSL feature extractor, simple instance-based MILs, with very few parameters, obtain similar or better performance than complex, state-of-the-art (SOTA) embedding-based MIL methods, setting new SOTA results on the BRACS and Camelyon16 datasets. Since simple instance-based MIL methods are naturally more interpretable and explainable to clinicians, our results suggest that more effort should be put into well-adapted SSL methods for WSI rather than into complex embedding-based MIL methods.
Stroke is a major public health problem, affecting millions worldwide. Deep learning has recently demonstrated promise for enhancing the diagnosis and risk prediction of stroke. However, existing methods rely on costly medical imaging modalities, such as computed tomography. Recent studies suggest that retinal imaging could offer a cost-effective alternative for cerebrovascular health assessment due to the shared clinical pathways between the retina and the brain. Hence, this study explores the impact of leveraging retinal images and clinical data for stroke detection and risk prediction. We propose a multimodal deep neural network that processes Optical Coherence Tomography (OCT) and infrared reflectance retinal scans, combined with clinical data, such as demographics, vital signs, and diagnosis codes. We pretrained our model using a self-supervised learning framework using a real-world dataset consisting of $37$ k scans, and then fine-tuned and evaluated the model using a smaller labeled subset. Our empirical findings establish the predictive ability of the considered modalities in detecting lasting effects in the retina associated with acute stroke and forecasting future risk within a specific time horizon. The experimental results demonstrate the effectiveness of our proposed framework by achieving $5$\% AUROC improvement as compared to the unimodal image-only baseline, and $8$\% improvement compared to an existing state-of-the-art foundation model. In conclusion, our study highlights the potential of retinal imaging in identifying high-risk patients and improving long-term outcomes.
Video quality assessment (VQA) is essential for quantifying perceptual quality in various video processing workflows, spanning from camera capture systems to over-the-top streaming platforms. While recent supervised VQA models have made substantial progress, the reliance on manually annotated datasets -- a process that is labor-intensive, costly, and difficult to scale up -- has hindered further optimization of their generalization to unseen video content and distortions. To bridge this gap, we introduce a self-supervised learning framework for VQA to learn quality assessment capabilities from large-scale, unlabeled web videos. Our approach leverages a \textbf{learning-to-rank} paradigm to train a large multimodal model (LMM) on video pairs automatically labeled via two manners, including quality pseudo-labeling by existing VQA models and relative quality ranking based on synthetic distortion simulations. Furthermore, we introduce a novel \textbf{iterative self-improvement training strategy}, where the trained model acts an improved annotator to iteratively refine the annotation quality of training data. By training on a dataset $10\times$ larger than the existing VQA benchmarks, our model: (1) achieves zero-shot performance on in-domain VQA benchmarks that matches or surpasses supervised models; (2) demonstrates superior out-of-distribution (OOD) generalization across diverse video content and distortions; and (3) sets a new state-of-the-art when fine-tuned on human-labeled datasets. Extensive experimental results validate the effectiveness of our self-supervised approach in training generalized VQA models. The datasets and code will be publicly released to facilitate future research.
Automating leaf manipulation in agricultural settings faces significant challenges, including the variability of plant morphologies and deformable leaves. We propose a novel hybrid geometric-neural approach for autonomous leaf grasping that combines traditional computer vision with neural networks through self-supervised learning. Our method integrates YOLOv8 for instance segmentation and RAFT-Stereo for 3D depth estimation to build rich leaf representations, which feed into both a geometric feature scoring pipeline and a neural refinement module (GraspPointCNN). The key innovation is our confidence-weighted fusion mechanism that dynamically balances the contribution of each approach based on prediction certainty. Our self-supervised framework uses the geometric pipeline as an expert teacher to automatically generate training data. Experiments demonstrate that our approach achieves an 88.0% success rate in controlled environments and 84.7% in real greenhouse conditions, significantly outperforming both purely geometric (75.3%) and neural (60.2%) methods. This work establishes a new paradigm for agricultural robotics where domain expertise is seamlessly integrated with machine learning capabilities, providing a foundation for fully automated crop monitoring systems.
Deep learning models have shown encouraging capabilities for mapping accurately forests at medium resolution with TanDEM-X interferometric SAR data. Such models, as most of current state-of-the-art deep learning techniques in remote sensing, are trained in a fully-supervised way, which requires a large amount of labeled data for training and validation. In this work, our aim is to exploit the high-resolution capabilities of the TanDEM-X mission to map forests at 6 m. The goal is to overcome the intrinsic limitations posed by midresolution products, which affect, e.g., the detection of narrow roads within vegetated areas and the precise delineation of forested regions contours. To cope with the lack of extended reliable reference datasets at such a high resolution, we investigate self-supervised learning techniques for extracting highly informative representations from the input features, followed by a supervised training step with a significantly smaller number of reliable labels. A 1 m resolution forest/non-forest reference map over Pennsylvania, USA, allows for comparing different training approaches for the development of an effective forest mapping framework with limited labeled samples. We select the best-performing approach over this test region and apply it in a real-case forest mapping scenario over the Amazon rainforest, where only very few labeled data at high resolution are available. In this challenging scenario, the proposed self-supervised framework significantly enhances the classification accuracy with respect to fully-supervised methods, trained using the same amount of labeled data, representing an extremely promising starting point for large-scale, very high-resolution forest mapping with TanDEM-X data.
Audiovisual emotion recognition (AVER) aims to infer human emotions from nonverbal visual-audio (VA) cues, offering modality-complementary and language-agnostic advantages. However, AVER remains challenging due to the inherent ambiguity of emotional expressions, cross-modal expressive disparities, and the scarcity of reliably annotated data. Recent self-supervised AVER approaches have introduced strong multimodal representations, yet they predominantly rely on modality-specific encoders and coarse content-level alignment, limiting fine-grained emotional semantic modeling. To address these issues, we propose VAEmo, an efficient two-stage framework for emotion-centric joint VA representation learning with external knowledge injection. In Stage 1, a unified and lightweight representation network is pre-trained on large-scale speaker-centric VA corpora via masked reconstruction and contrastive objectives, mitigating the modality gap and learning expressive, complementary representations without emotion labels. In Stage 2, multimodal large language models automatically generate detailed affective descriptions according to our well-designed chain-of-thought prompting for only a small subset of VA samples; these rich textual semantics are then injected by aligning their corresponding embeddings with VA representations through dual-path contrastive learning, further bridging the emotion gap. Extensive experiments on multiple downstream AVER benchmarks show that VAEmo achieves state-of-the-art performance with a compact design, highlighting the benefit of unified cross-modal encoding and emotion-aware semantic guidance for efficient, generalizable VA emotion representations.
The acoustic analysis of marmoset (Callithrix jacchus) vocalizations is often used to understand the evolutionary origins of human language. Currently, the analysis is largely carried out in a manual or semi-manual manner. Thus, there is a need to develop automatic call analysis methods. In that direction, research has been limited to the development of analysis methods with small amounts of data or for specific scenarios. Furthermore, there is lack of prior knowledge about what type of information is relevant for different call analysis tasks. To address these issues, as a first step, this paper explores different feature representation methods, namely, HCTSA-based hand-crafted features Catch22, pre-trained self supervised learning (SSL) based features extracted from neural networks trained on human speech and end-to-end acoustic modeling for call-type classification, caller identification and caller sex identification. Through an investigation on three different marmoset call datasets, we demonstrate that SSL-based feature representations and end-to-end acoustic modeling tend to lead to better systems than Catch22 features for call-type and caller classification. Furthermore, we also highlight the impact of signal bandwidth on the obtained task performances.
Self-supervised learning (SSL) has demonstrated remarkable success in 3D point cloud analysis, particularly through masked autoencoders (MAEs). However, existing MAE-based methods lack rotation invariance, leading to significant performance degradation when processing arbitrarily rotated point clouds in real-world scenarios. To address this limitation, we introduce Handcrafted Feature-Based Rotation-Invariant Masked Autoencoder (HFBRI-MAE), a novel framework that refines the MAE design with rotation-invariant handcrafted features to ensure stable feature learning across different orientations. By leveraging both rotation-invariant local and global features for token embedding and position embedding, HFBRI-MAE effectively eliminates rotational dependencies while preserving rich geometric structures. Additionally, we redefine the reconstruction target to a canonically aligned version of the input, mitigating rotational ambiguities. Extensive experiments on ModelNet40, ScanObjectNN, and ShapeNetPart demonstrate that HFBRI-MAE consistently outperforms existing methods in object classification, segmentation, and few-shot learning, highlighting its robustness and strong generalization ability in real-world 3D applications.