Conventional audio-visual methods for speaker verification rely on large amounts of labeled data and separate modality-specific architectures, which is computationally expensive, limiting their scalability. To address these problems, we propose a self-supervised learning framework based on contrastive learning with asymmetric masking and masked data modeling to obtain robust audiovisual feature representations. In particular, we employ a unified framework for self-supervised audiovisual speaker verification using a single shared backbone for audio and visual inputs, leveraging the versatility of vision transformers. The proposed unified framework can handle audio, visual, or audiovisual inputs using a single shared vision transformer backbone during training and testing while being computationally efficient and robust to missing modalities. Extensive experiments demonstrate that our method achieves competitive performance without labeled data while reducing computational costs compared to traditional approaches.
Keyword Spotting plays a critical role in enabling hands-free interaction for battery-powered edge devices. Few-Shot Keyword Spotting (FS-KWS) addresses the scalability and adaptability challenges of traditional systems by enabling recognition of custom keywords with only a few examples. However, existing FS-KWS systems achieve subpar accuracy at desirable false acceptance rates, particularly in resource-constrained edge environments. To address these issues, we propose a training scheme that leverages self-supervised learning models for robust feature extraction, dimensionality reduction, and knowledge distillation. The teacher model, based on Wav2Vec 2.0 is trained using Sub-center ArcFace loss, which enhances inter-class separability and intra-class compactness. To enable efficient deployment on edge devices, we introduce attention-based dimensionality reduction and train a standard lightweight ResNet15 student model. We evaluate the proposed approach on the English portion of the Multilingual Spoken Words Corpus (MSWC) and the Google Speech Commands (GSC) datasets. Notably, the proposed training method improves the 10-shot classification accuracy from 33.4% to 74.1% on 11 classes at 1% false alarm accuracy on the GSC dataset, thus making it significantly better-suited for a real use case scenario.
Traditional models of accent perception underestimate the role of gradient variations in phonological features which listeners rely upon for their accent judgments. We investigate how pretrained representations from current self-supervised learning (SSL) models of speech encode phonological feature-level variations that influence the perception of segmental accent. We focus on three segments: the labiodental approximant, the rhotic tap, and the retroflex stop, which are uniformly produced in the English of native speakers of Hindi as well as other languages in the Indian sub-continent. We use the CSLU Foreign Accented English corpus (Lander, 2007) to extract, for these segments, phonological feature probabilities using Phonet (V\'asquez-Correa et al., 2019) and pretrained representations from Wav2Vec2-BERT (Barrault et al., 2023) and WavLM (Chen et al., 2022) along with accent judgements by native speakers of American English. Probing analyses show that accent strength is best predicted by a subset of the segment's pretrained representation features, in which perceptually salient phonological features that contrast the expected American English and realized non-native English segments are given prominent weighting. A multinomial logistic regression of pretrained representation-based segment distances from American and Indian English baselines on accent ratings reveals strong associations between the odds of accent strength and distances from the baselines, in the expected directions. These results highlight the value of self-supervised speech representations for modeling accent perception using interpretable phonological features.
Dysarthria is a neurological disorder that significantly impairs speech intelligibility, often rendering affected individuals unable to communicate effectively. This necessitates the development of robust dysarthric-to-regular speech conversion techniques. In this work, we investigate the utility and limitations of self-supervised learning (SSL) features and their quantized representations as an alternative to mel-spectrograms for speech generation. Additionally, we explore methods to mitigate speaker variability by generating clean speech in a single-speaker voice using features extracted from WavLM. To this end, we propose a fully non-autoregressive approach that leverages Conditional Flow Matching (CFM) with Diffusion Transformers to learn a direct mapping from dysarthric to clean speech. Our findings highlight the effectiveness of discrete acoustic units in improving intelligibility while achieving faster convergence compared to traditional mel-spectrogram-based approaches.
The ability of deep neural networks (DNNs) come from extracting and interpreting features from the data provided. By exploiting intermediate features in DNNs instead of relying on hard labels, we craft adversarial perturbation that generalize more effectively, boosting black-box transferability. These features ubiquitously come from supervised learning in previous work. Inspired by the exceptional synergy between self-supervised learning and the Transformer architecture, this paper explores whether exploiting self-supervised Vision Transformer (ViT) representations can improve adversarial transferability. We present dSVA -- a generative dual self-supervised ViT features attack, that exploits both global structural features from contrastive learning (CL) and local textural features from masked image modeling (MIM), the self-supervised learning paradigm duo for ViTs. We design a novel generative training framework that incorporates a generator to create black-box adversarial examples, and strategies to train the generator by exploiting joint features and the attention mechanism of self-supervised ViTs. Our findings show that CL and MIM enable ViTs to attend to distinct feature tendencies, which, when exploited in tandem, boast great adversarial generalizability. By disrupting dual deep features distilled by self-supervised ViTs, we are rewarded with remarkable black-box transferability to models of various architectures that outperform state-of-the-arts. Code available at https://github.com/spencerwooo/dSVA.
Contrastive self-supervised learning based on point-wise comparisons has been widely studied for vision tasks. In the visual cortex of the brain, neuronal responses to distinct stimulus classes are organized into geometric structures known as neural manifolds. Accurate classification of stimuli can be achieved by effectively separating these manifolds, akin to solving a packing problem. We introduce Contrastive Learning As Manifold Packing (CLAMP), a self-supervised framework that recasts representation learning as a manifold packing problem. CLAMP introduces a loss function inspired by the potential energy of short-range repulsive particle systems, such as those encountered in the physics of simple liquids and jammed packings. In this framework, each class consists of sub-manifolds embedding multiple augmented views of a single image. The sizes and positions of the sub-manifolds are dynamically optimized by following the gradient of a packing loss. This approach yields interpretable dynamics in the embedding space that parallel jamming physics, and introduces geometrically meaningful hyperparameters within the loss function. Under the standard linear evaluation protocol, which freezes the backbone and trains only a linear classifier, CLAMP achieves competitive performance with state-of-the-art self-supervised models. Furthermore, our analysis reveals that neural manifolds corresponding to different categories emerge naturally and are effectively separated in the learned representation space, highlighting the potential of CLAMP to bridge insights from physics, neural science, and machine learning.
Self-supervised learning (SSL) has revolutionized audio representations, yet models often remain domain-specific, focusing on either speech or non-speech tasks. In this work, we present Universal Speech and Audio Distillation (USAD), a unified approach to audio representation learning that integrates diverse audio types - speech, sound, and music - into a single model. USAD employs efficient layer-to-layer distillation from domain-specific SSL models to train a student on a comprehensive audio dataset. USAD offers competitive performance across various benchmarks and datasets, including frame and instance-level speech processing tasks, audio tagging, and sound classification, achieving near state-of-the-art results with a single encoder on SUPERB and HEAR benchmarks.
The increasing ubiquity of video content and the corresponding demand for efficient access to meaningful information have elevated video summarization and video highlights as a vital research area. However, many state-of-the-art methods depend heavily either on supervised annotations or on attention-based models, which are computationally expensive and brittle in the face of distribution shifts that hinder cross-domain applicability across datasets. We introduce a pioneering self-supervised video summarization model that captures both spatial and temporal dependencies without the overhead of attention, RNNs, or transformers. Our framework integrates a novel set of Markov process-driven loss metrics and a two-stage self supervised learning paradigm that ensures both performance and efficiency. Our approach achieves state-of-the-art performance on the SUMME and TVSUM datasets, outperforming all existing unsupervised methods. It also rivals the best supervised models, demonstrating the potential for efficient, annotation-free architectures. This paves the way for more generalizable video summarization techniques and challenges the prevailing reliance on complex architectures.
Osteoporosis, characterized by reduced bone mineral density (BMD) and compromised bone microstructure, increases fracture risk in aging populations. While dual-energy X-ray absorptiometry (DXA) is the clinical standard for BMD assessment, its limited accessibility hinders diagnosis in resource-limited regions. Opportunistic computed tomography (CT) analysis has emerged as a promising alternative for osteoporosis diagnosis using existing imaging data. Current approaches, however, face three limitations: (1) underutilization of unlabeled vertebral data, (2) systematic bias from device-specific DXA discrepancies, and (3) insufficient integration of clinical knowledge such as spatial BMD distribution patterns. To address these, we propose a unified deep learning framework with three innovations. First, a self-supervised learning method using radiomic representations to leverage unlabeled CT data and preserve bone texture. Second, a Mixture of Experts (MoE) architecture with learned gating mechanisms to enhance cross-device adaptability. Third, a multi-task learning framework integrating osteoporosis diagnosis, BMD regression, and vertebra location prediction. Validated across three clinical sites and an external hospital, our approach demonstrates superior generalizability and accuracy over existing methods for opportunistic osteoporosis screening and diagnosis.
Satellite remote sensing (RS) enables a wide array of downstream Earth observation (EO) applications, including climate modeling, carbon accounting, and strategies for conservation and sustainable land use. We present TESSERA, a novel Remote Sensing Foundation Model (RSFM) that uses Self-Supervised Learning (SSL) to generate global, robust representations at 10m scale from pixel-level satellite time series data. TESSERA combines information from only optical and SAR data streams using two parallel Transformer-based encoders: one dedicated to Sentinel-1 SAR polarizations and another to Sentinel-2 MSI data (10 selected spectral bands) to create representations that are then fused using a multilayer perceptron (MLP), resulting in a global representation map covering the years 2017 to 2024. Our precomputed representations set a new state-of-the-art performance benchmark and our open-source approach democratizes access to high-performance, high-resolution representations. We benchmark the performance of TESSERA in five diverse tasks, comparing our work with state-of-the-art task-specific models and other foundation models. Our results show that TESSERA outperforms both traditional RS baselines and the leading geospatial foundation models in these diverse downstream tasks.
Automatic fluency assessment (AFA) remains challenging, particularly in capturing speech rhythm, pauses, and disfluencies in non-native speakers. We introduce a chunk-based approach integrating self-supervised learning (SSL) models (Wav2Vec2, HuBERT, and WavLM) selected for their complementary strengths in phonetic, prosodic, and noisy speech modeling, with a hierarchical CNN-BiLSTM framework. Speech is segmented into breath-group chunks using Silero voice activity detection (Silero-VAD), enabling fine-grained temporal analysis while mitigating over-segmentation artifacts. SSL embeddings are fused via a learnable weighted mechanism, balancing acoustic and linguistic features, and enriched with chunk-level fluency markers (e.g., speech rate, pause durations, n-gram repetitions). The CNN-BiLSTM captures local and long-term dependencies across chunks. Evaluated on Avalinguo and Speechocean762, our approach improves F1-score by 2.8 and Pearson correlation by 6.2 points over single SSL baselines on Speechocean762, with gains of 4.2 F1-score and 4.0 Pearson points on Avalinguo, surpassing Pyannote.audio-based segmentation baselines. These findings highlight chunk-based multi-SSL fusion for robust fluency evaluation, though future work should explore generalization to dialects with irregular prosody.
We present FOMO60K, a large-scale, heterogeneous dataset of 60,529 brain Magnetic Resonance Imaging (MRI) scans from 13,900 sessions and 11,187 subjects, aggregated from 16 publicly available sources. The dataset includes both clinical- and research-grade images, multiple MRI sequences, and a wide range of anatomical and pathological variability, including scans with large brain anomalies. Minimal preprocessing was applied to preserve the original image characteristics while reducing barriers to entry for new users. Accompanying code for self-supervised pretraining and finetuning is provided. FOMO60K is intended to support the development and benchmarking of self-supervised learning methods in medical imaging at scale.
Depth map enhancement using paired high-resolution RGB images offers a cost-effective solution for improving low-resolution depth data from lightweight ToF sensors. Nevertheless, naively adopting a depth estimation pipeline to fuse the two modalities requires groundtruth depth maps for supervision. To address this, we propose a self-supervised learning framework, SelfToF, which generates detailed and scale-aware depth maps. Starting from an image-based self-supervised depth estimation pipeline, we add low-resolution depth as inputs, design a new depth consistency loss, propose a scale-recovery module, and finally obtain a large performance boost. Furthermore, since the ToF signal sparsity varies in real-world applications, we upgrade SelfToF to SelfToF* with submanifold convolution and guided feature fusion. Consequently, SelfToF* maintain robust performance across varying sparsity levels in ToF data. Overall, our proposed method is both efficient and effective, as verified by extensive experiments on the NYU and ScanNet datasets. The code is available at \href{https://github.com/denyingmxd/selftof}{https://github.com/denyingmxd/selftof}.
Image-based cell profiling aims to create informative representations of cell images. This technique is critical in drug discovery and has greatly advanced with recent improvements in computer vision. Inspired by recent developments in non-contrastive Self-Supervised Learning (SSL), this paper provides an initial exploration into training a generalizable feature extractor for cell images using such methods. However, there are two major challenges: 1) There is a large difference between the distributions of cell images and natural images, causing the view-generation process in existing SSL methods to fail; and 2) Unlike typical scenarios where each representation is based on a single image, cell profiling often involves multiple input images, making it difficult to effectively combine all available information. To overcome these challenges, we propose SSLProfiler, a non-contrastive SSL framework specifically designed for cell profiling. We introduce specialized data augmentation and representation post-processing methods tailored to cell images, which effectively address the issues mentioned above and result in a robust feature extractor. With these improvements, SSLProfiler won the Cell Line Transferability challenge at CVPR 2025.
Recent advancements in speech synthesis technologies have led to increasingly advanced spoofing attacks, posing significant challenges for automatic speaker verification systems. While systems based on self-supervised learning (SSL) models, particularly the XLSR-Conformer model, have demonstrated remarkable performance in synthetic speech detection, there remains room for architectural improvements. In this paper, we propose a novel approach that replaces the traditional Multi-Layer Perceptron in the XLSR-Conformer model with a Kolmogorov-Arnold Network (KAN), a novel architecture based on the Kolmogorov-Arnold representation theorem. Our results on ASVspoof2021 demonstrate that integrating KAN into the SSL-based models can improve the performance by 60.55% relatively on LA and DF sets, further achieving 0.70% EER on the 21LA set. These findings suggest that incorporating KAN into SSL-based models is a promising direction for advances in synthetic speech detection.
In the face of pressing environmental issues in the 21st century, monitoring surface changes on Earth is more important than ever. Large-scale remote sensing, such as satellite imagery, is an important tool for this task. However, using supervised methods to detect changes is difficult because of the lack of satellite data annotated with change labels, especially for rare categories of change. Annotation proves challenging due to the sparse occurrence of changes in satellite images. Even within a vast collection of images, only a small fraction may exhibit persistent changes of interest. To address this challenge, we introduce OPTIMUS, a self-supervised learning method based on an intuitive principle: if a model can recover information about the relative order of images in the time series, then that implies that there are long-lasting changes in the images. OPTIMUS demonstrates this principle by using change point detection methods on model outputs in a time series. We demonstrate that OPTIMUS can directly detect interesting changes in satellite images, achieving an improvement in AUROC score from 56.3% to 87.6% at distinguishing changed time series from unchanged ones compared to baselines. Our code and dataset are available at https://huggingface.co/datasets/optimus-change/optimus-dataset/.
Self-supervised pre-trained audio networks have seen widespread adoption in real-world systems, particularly in multi-modal large language models. These networks are often employed in a frozen state, under the assumption that the SSL pre-training has sufficiently equipped them to handle real-world audio. However, a critical question remains: how well do these models actually perform in real-world conditions, where audio is typically polyphonic and complex, involving multiple overlapping sound sources? Current audio SSL methods are often benchmarked on datasets predominantly featuring monophonic audio, such as environmental sounds, and speech. As a result, the ability of SSL models to generalize to polyphonic audio, a common characteristic in natural scenarios, remains underexplored. This limitation raises concerns about the practical robustness of SSL models in more realistic audio settings. To address this gap, we introduce Self-Supervised Learning from Audio Mixtures (SSLAM), a novel direction in audio SSL research, designed to improve, designed to improve the model's ability to learn from polyphonic data while maintaining strong performance on monophonic data. We thoroughly evaluate SSLAM on standard audio SSL benchmark datasets which are predominantly monophonic and conduct a comprehensive comparative analysis against SOTA methods using a range of high-quality, publicly available polyphonic datasets. SSLAM not only improves model performance on polyphonic audio, but also maintains or exceeds performance on standard audio SSL benchmarks. Notably, it achieves up to a 3.9\% improvement on the AudioSet-2M (AS-2M), reaching a mean average precision (mAP) of 50.2. For polyphonic datasets, SSLAM sets new SOTA in both linear evaluation and fine-tuning regimes with performance improvements of up to 9.1\% (mAP).
Self-supervised learning (SSL) has achieved major advances in natural images and video understanding, but challenges remain in domains like echocardiography (heart ultrasound) due to subtle anatomical structures, complex temporal dynamics, and the current lack of domain-specific pre-trained models. Existing SSL approaches such as contrastive, masked modeling, and clustering-based methods struggle with high intersample similarity, sensitivity to low PSNR inputs common in ultrasound, or aggressive augmentations that distort clinically relevant features. We present DISCOVR (Distilled Image Supervision for Cross Modal Video Representation), a self-supervised dual branch framework for cardiac ultrasound video representation learning. DISCOVR combines a clustering-based video encoder that models temporal dynamics with an online image encoder that extracts fine-grained spatial semantics. These branches are connected through a semantic cluster distillation loss that transfers anatomical knowledge from the evolving image encoder to the video encoder, enabling temporally coherent representations enriched with fine-grained semantic understanding. Evaluated on six echocardiography datasets spanning fetal, pediatric, and adult populations, DISCOVR outperforms both specialized video anomaly detection methods and state-of-the-art video-SSL baselines in zero-shot and linear probing setups, and achieves superior segmentation transfer.
The growing prevalence of digital health technologies has led to the generation of complex multi-modal data, such as physical activity measurements simultaneously collected from various sensors of mobile and wearable devices. These data hold immense potential for advancing health studies, but current methods predominantly rely on supervised learning, requiring extensive labeled datasets that are often expensive or impractical to obtain, especially in clinical studies. To address this limitation, we propose a self-supervised learning framework called Multi-modal Cross-masked Autoencoder (MoCA) that leverages cross-modality masking and the Transformer autoencoder architecture to utilize both temporal correlations within modalities and cross-modal correlations between data streams. We also provide theoretical guarantees to support the effectiveness of the cross-modality masking scheme in MoCA. Comprehensive experiments and ablation studies demonstrate that our method outperforms existing approaches in both reconstruction and downstream tasks. We release open-source code for data processing, pre-training, and downstream tasks in the supplementary materials. This work highlights the transformative potential of self-supervised learning in digital health and multi-modal data.
Transfer learning, a technique commonly used in generative artificial intelligence, allows neural network models to bring prior knowledge to bear when learning a new task. This study demonstrates that transfer learning significantly enhances the accuracy of sleep-stage decoding from peripheral wearable devices by leveraging neural network models pretrained on electroencephalographic (EEG) signals. Consumer wearable technologies typically rely on peripheral physiological signals such as pulse plethysmography (PPG) and respiratory data, which, while convenient, lack the fidelity of clinical electroencephalography (EEG) for detailed sleep-stage classification. We pretrained a transformer-based neural network on a large, publicly available EEG dataset and subsequently fine-tuned this model on noisier peripheral signals. Our transfer learning approach improved overall classification accuracy from 67.6\% (baseline model trained solely on peripheral signals) to 76.6\%. Notable accuracy improvements were observed across sleep stages, particularly lighter sleep stages such as REM and N1. These results highlight transfer learning's potential to substantially enhance the accuracy and utility of consumer wearable devices without altering existing hardware. Future integration of self-supervised learning methods may further boost performance, facilitating more precise, longitudinal sleep monitoring for personalized health applications.