The common standard for quality evaluation of automatic speech recognition (ASR) systems is reference-based metrics such as the Word Error Rate (WER), computed using manual ground-truth transcriptions that are time-consuming and expensive to obtain. This work proposes a multi-language referenceless quality metric, which allows comparing the performance of different ASR models on a speech dataset without ground truth transcriptions. To estimate the quality of ASR hypotheses, a pre-trained language model (LM) is fine-tuned with contrastive learning in a self-supervised learning manner. In experiments conducted on several unseen test datasets consisting of outputs from top commercial ASR engines in various languages, the proposed referenceless metric obtains a much higher correlation with WER scores and their ranks than the perplexity metric from the state-of-art multi-lingual LM in all experiments, and also reduces WER by more than $7\%$ when used for ensembling hypotheses. The fine-tuned model and experiments are made available for the reproducibility: https://github.com/aixplain/NoRefER
Transfer learning plays a key role in advancing machine learning models, yet conventional supervised pretraining often undermines feature transferability by prioritizing features that minimize the pretraining loss. In this work, we adapt a self-supervised learning regularization technique from the VICReg method to supervised learning contexts, introducing Variance-Covariance Regularization (VCReg). This adaptation encourages the network to learn high-variance, low-covariance representations, promoting learning more diverse features. We outline best practices for an efficient implementation of our framework, including applying it to the intermediate representations. Through extensive empirical evaluation, we demonstrate that our method significantly enhances transfer learning for images and videos, achieving state-of-the-art performance across numerous tasks and datasets. VCReg also improves performance in scenarios like long-tail learning and hierarchical classification. Additionally, we show its effectiveness may stem from its success in addressing challenges like gradient starvation and neural collapse. In summary, VCReg offers a universally applicable regularization framework that significantly advances transfer learning and highlights the connection between gradient starvation, neural collapse, and feature transferability.
This paper proposes a novel Self-Supervised Intrusion Detection (SSID) framework, which enables a fully online Deep Learning (DL) based Intrusion Detection System (IDS) that requires no human intervention or prior off-line learning. The proposed framework analyzes and labels incoming traffic packets based only on the decisions of the IDS itself using an Auto-Associative Deep Random Neural Network, and on an online estimate of its statistically measured trustworthiness. The SSID framework enables IDS to adapt rapidly to time-varying characteristics of the network traffic, and eliminates the need for offline data collection. This approach avoids human errors in data labeling, and human labor and computational costs of model training and data collection. The approach is experimentally evaluated on public datasets and compared with well-known {machine learning and deep learning} models, showing that this SSID framework is very useful and advantageous as an accurate and online learning DL-based IDS for IoT systems.
Automatic singing voice understanding tasks, such as singer identification, singing voice transcription, and singing technique classification, benefit from data-driven approaches that utilize deep learning techniques. These approaches work well even under the rich diversity of vocal and noisy samples owing to their representation ability. However, the limited availability of labeled data remains a significant obstacle to achieving satisfactory performance. In recent years, self-supervised learning models (SSL models) have been trained using large amounts of unlabeled data in the field of speech processing and music classification. By fine-tuning these models for the target tasks, comparable performance to conventional supervised learning can be achieved with limited training data. Therefore, in this paper, we investigate the effectiveness of SSL models for various singing voice recognition tasks. We report the results of experiments comparing SSL models for three different tasks (i.e., singer identification, singing voice transcription, and singing technique classification) as initial exploration and aim to discuss these findings. Experimental results show that each SSL model achieves comparable performance and sometimes outperforms compared to state-of-the-art methods on each task. We also conducted a layer-wise analysis to further understand the behavior of the SSL models.
Self-supervised learning (SSL) has emerged as a promising paradigm that presents self-generated supervisory signals to real-world problems, bypassing the extensive manual labeling burden. SSL is especially attractive for unsupervised tasks such as anomaly detection, where labeled anomalies are often nonexistent and costly to obtain. While self-supervised anomaly detection (SSAD) has seen a recent surge of interest, the literature has failed to treat data augmentation as a hyperparameter. Meanwhile, recent works have reported that the choice of augmentation has significant impact on detection performance. In this paper, we introduce ST-SSAD (Self-Tuning Self-Supervised Anomaly Detection), the first systematic approach to SSAD in regards to rigorously tuning augmentation. To this end, our work presents two key contributions. The first is a new unsupervised validation loss that quantifies the alignment between the augmented training data and the (unlabeled) test data. In principle we adopt transduction, quantifying the extent to which augmentation mimics the true anomaly-generating mechanism, in contrast to augmenting data with arbitrary pseudo anomalies without regard to test data. Second, we present new differentiable augmentation functions, allowing data augmentation hyperparameter(s) to be tuned end-to-end via our proposed validation loss. Experiments on two testbeds with semantic class anomalies and subtle industrial defects show that systematically tuning augmentation offers significant performance gains over current practices.
The existing contrastive learning methods widely adopt one-hot instance discrimination as pretext task for self-supervised learning, which inevitably neglects rich inter-instance similarities among natural images, then leading to potential representation degeneration. In this paper, we propose a novel image mix method, PatchMix, for contrastive learning in Vision Transformer (ViT), to model inter-instance similarities among images. Following the nature of ViT, we randomly mix multiple images from mini-batch in patch level to construct mixed image patch sequences for ViT. Compared to the existing sample mix methods, our PatchMix can flexibly and efficiently mix more than two images and simulate more complicated similarity relations among natural images. In this manner, our contrastive framework can significantly reduce the gap between contrastive objective and ground truth in reality. Experimental results demonstrate that our proposed method significantly outperforms the previous state-of-the-art on both ImageNet-1K and CIFAR datasets, e.g., 3.0% linear accuracy improvement on ImageNet-1K and 8.7% kNN accuracy improvement on CIFAR100. Moreover, our method achieves the leading transfer performance on downstream tasks, object detection and instance segmentation on COCO dataset. The code is available at https://github.com/visresearch/patchmix
Contrastive learning, along with its variations, has been a highly effective self-supervised learning method across diverse domains. Contrastive learning measures the distance between representations using cosine similarity and uses cross-entropy for representation learning. Within the same framework of cosine-similarity-based representation learning, margins have played a significant role in enhancing face and speaker recognition tasks. Interestingly, despite the shared reliance on the same similarity metrics and objective functions, contrastive learning has not actively adopted margins. Furthermore, decision-boundary-based explanations are the only ones that have been used to explain the effect of margins in contrastive learning. In this work, we propose a new perspective to understand the role of margins based on gradient analysis. Based on the new perspective, we analyze how margins affect gradients of contrastive learning and separate the effect into more elemental levels. We separately analyze each and provide possible directions for improving contrastive learning. Our experimental results demonstrate that emphasizing positive samples and scaling gradients depending on positive sample angles and logits are the keys to improving the generalization performance of contrastive learning in both seen and unseen datasets, and other factors can only marginally improve performance.
General-purpose foundation models have led to recent breakthroughs in artificial intelligence. In remote sensing, self-supervised learning (SSL) and Masked Image Modeling (MIM) have been adopted to build foundation models. However, these models primarily learn low-level features and require annotated data for fine-tuning. Moreover, they are inapplicable for retrieval and zero-shot applications due to the lack of language understanding. To address these limitations, we propose RemoteCLIP, the first vision-language foundation model for remote sensing that aims to learn robust visual features with rich semantics and aligned text embeddings for seamless downstream application. To address the scarcity of pre-training data, we leverage data scaling which converts heterogeneous annotations into a unified image-caption data format based on Box-to-Caption (B2C) and Mask-to-Box (M2B) conversion. By further incorporating UAV imagery, we produce a 12 $\times$ larger pretraining dataset than the combination of all available datasets. RemoteCLIP can be applied to a variety of downstream tasks, including zero-shot image classification, linear probing, $\textit{k}$-NN classification, few-shot classification, image-text retrieval, and object counting in remote sensing images. Evaluation on 16 datasets, including a newly introduced RemoteCount benchmark to test the object counting ability, shows that RemoteCLIP consistently outperforms baseline foundation models across different model scales. Impressively, RemoteCLIP beats the state-of-the-art method by 9.14% mean recall on the RSITMD dataset and 8.92% on the RSICD dataset. For zero-shot classification, our RemoteCLIP outperforms the CLIP baseline by up to 6.39% average accuracy on 12 downstream datasets. Project website: https://github.com/ChenDelong1999/RemoteCLIP
We introduce RedMotion, a transformer model for motion prediction in self-driving vehicles that learns environment representations via redundancy reduction. Our first type of redundancy reduction is induced by an internal transformer decoder and reduces a variable-sized set of local road environment tokens, representing road graphs and agent data, to a fixed-sized global embedding. The second type of redundancy reduction is obtained by self-supervised learning and applies the redundancy reduction principle to embeddings generated from augmented views of road environments. Our experiments reveal that our representation learning approach outperforms PreTraM, Traj-MAE, and GraphDINO in a semi-supervised setting. Moreover, RedMotion achieves competitive results compared to HPTR or MTR++ in the Waymo Motion Prediction Challenge. Our open-source implementation is available at: https://github.com/kit-mrt/future-motion
The computer-assisted radiologic informative report has received increasing research attention to facilitate diagnosis and treatment planning for dental care providers. However, manual interpretation of dental images is limited, expensive, and time-consuming. Another barrier in dental imaging is the limited number of available images for training, which is a challenge in the era of deep learning. This study proposes a novel self-distillation (SD) enhanced self-supervised learning on top of the masked image modeling (SimMIM) Transformer, called SD-SimMIM, to improve the outcome with a limited number of dental radiographs. In addition to the prediction loss on masked patches, SD-SimMIM computes the self-distillation loss on the visible patches. We apply SD-SimMIM on dental panoramic X-rays for teeth numbering, detection of dental restorations and orthodontic appliances, and instance segmentation tasks. Our results show that SD-SimMIM outperforms other self-supervised learning methods. Furthermore, we augment and improve the annotation of an existing dataset of panoramic X-rays.
Self-supervised learning (SSL) has recently achieved impressive performance on various time series tasks. The most prominent advantage of SSL is that it reduces the dependence on labeled data. Based on the pre-training and fine-tuning strategy, even a small amount of labeled data can achieve high performance. Compared with many published self-supervised surveys on computer vision and natural language processing, a comprehensive survey for time series SSL is still missing. To fill this gap, we review current state-of-the-art SSL methods for time series data in this article. To this end, we first comprehensively review existing surveys related to SSL and time series, and then provide a new taxonomy of existing time series SSL methods by summarizing them from three perspectives: generative-based, contrastive-based, and adversarial-based. These methods are further divided into ten subcategories with detailed reviews and discussions about their key intuitions, main frameworks, advantages and disadvantages. To facilitate the experiments and validation of time series SSL methods, we also summarize datasets commonly used in time series forecasting, classification, anomaly detection, and clustering tasks. Finally, we present the future directions of SSL for time series analysis.
Accurate feature matching and correspondence in endoscopic images play a crucial role in various clinical applications, including patient follow-up and rapid anomaly localization through panoramic image generation. However, developing robust and accurate feature matching techniques faces challenges due to the lack of discriminative texture and significant variability between patients. To address these limitations, we propose a novel self-supervised approach that combines Convolutional Neural Networks for capturing local visual appearance and attention-based Graph Neural Networks for modeling spatial relationships between key-points. Our approach is trained in a fully self-supervised scheme without the need for labeled data. Our approach outperforms state-of-the-art handcrafted and deep learning-based methods, demonstrating exceptional performance in terms of precision rate (1) and matching score (99.3%). We also provide code and materials related to this work, which can be accessed at https://github.com/abenhamadou/graph-self-supervised-learning-for-endoscopic-image-matching.
Spatial-temporal graph learning has emerged as a promising solution for modeling structured spatial-temporal data and learning region representations for various urban sensing tasks such as crime forecasting and traffic flow prediction. However, most existing models are vulnerable to the quality of the generated region graph due to the inaccurate graph-structured information aggregation schema. The ubiquitous spatial-temporal data noise and incompleteness in real-life scenarios pose challenges in generating high-quality region representations. To address this challenge, we propose a new spatial-temporal graph learning model (GraphST) for enabling effective self-supervised learning. Our proposed model is an adversarial contrastive learning paradigm that automates the distillation of crucial multi-view self-supervised information for robust spatial-temporal graph augmentation. We empower GraphST to adaptively identify hard samples for better self-supervision, enhancing the representation discrimination ability and robustness. In addition, we introduce a cross-view contrastive learning paradigm to model the inter-dependencies across view-specific region representations and preserve underlying relation heterogeneity. We demonstrate the superiority of our proposed GraphST method in various spatial-temporal prediction tasks on real-life datasets. We release our model implementation via the link: \url{https://github.com/HKUDS/GraphST}.
Contrastive learning (CL) has become the de-facto learning paradigm in self-supervised learning on graphs, which generally follows the "augmenting-contrasting" learning scheme. However, we observe that unlike CL in computer vision domain, CL in graph domain performs decently even without augmentation. We conduct a systematic analysis of this phenomenon and argue that homophily, i.e., the principle that "like attracts like", plays a key role in the success of graph CL. Inspired to leverage this property explicitly, we propose HomoGCL, a model-agnostic framework to expand the positive set using neighbor nodes with neighbor-specific significances. Theoretically, HomoGCL introduces a stricter lower bound of the mutual information between raw node features and node embeddings in augmented views. Furthermore, HomoGCL can be combined with existing graph CL models in a plug-and-play way with light extra computational overhead. Extensive experiments demonstrate that HomoGCL yields multiple state-of-the-art results across six public datasets and consistently brings notable performance improvements when applied to various graph CL methods. Code is avilable at https://github.com/wenzhilics/HomoGCL.
Reinforcement learning is able to solve complex sequential decision-making tasks but is currently limited by sample efficiency and required computation. To improve sample efficiency, recent work focuses on model-based RL which interleaves model learning with planning. Recent methods further utilize policy learning, value estimation, and, self-supervised learning as auxiliary objectives. In this paper we show that, surprisingly, a simple representation learning approach relying only on a latent dynamics model trained by latent temporal consistency is sufficient for high-performance RL. This applies when using pure planning with a dynamics model conditioned on the representation, but, also when utilizing the representation as policy and value function features in model-free RL. In experiments, our approach learns an accurate dynamics model to solve challenging high-dimensional locomotion tasks with online planners while being 4.1 times faster to train compared to ensemble-based methods. With model-free RL without planning, especially on high-dimensional tasks, such as the DeepMind Control Suite Humanoid and Dog tasks, our approach outperforms model-free methods by a large margin and matches model-based methods' sample efficiency while training 2.4 times faster.
Multiple Object Tracking (MOT) aims to find bounding boxes and identities of targeted objects in consecutive video frames. While fully-supervised MOT methods have achieved high accuracy on existing datasets, they cannot generalize well on a newly obtained dataset or a new unseen domain. In this work, we first address the MOT problem from the cross-domain point of view, imitating the process of new data acquisition in practice. Then, a new cross-domain MOT adaptation from existing datasets is proposed without any pre-defined human knowledge in understanding and modeling objects. It can also learn and update itself from the target data feedback. The intensive experiments are designed on four challenging settings, including MOTSynth to MOT17, MOT17 to MOT20, MOT17 to VisDrone, and MOT17 to DanceTrack. We then prove the adaptability of the proposed self-supervised learning strategy. The experiments also show superior performance on tracking metrics MOTA and IDF1, compared to fully supervised, unsupervised, and self-supervised state-of-the-art methods.
Transformers have gained popularity in time series forecasting for their ability to capture long-sequence interactions. However, their high memory and computing requirements pose a critical bottleneck for long-term forecasting. To address this, we propose TSMixer, a lightweight neural architecture exclusively composed of multi-layer perceptron (MLP) modules for multivariate forecasting and representation learning on patched time series. Inspired by MLP-Mixer's success in computer vision, we adapt it for time series, addressing challenges and introducing validated components for enhanced accuracy. This includes a novel design paradigm of attaching online reconciliation heads to the MLP-Mixer backbone, for explicitly modeling the time-series properties such as hierarchy and channel-correlations. We also propose a novel Hybrid channel modeling and infusion of a simple gating approach to effectively handle noisy channel interactions and generalization across diverse datasets. By incorporating these lightweight components, we significantly enhance the learning capability of simple MLP structures, outperforming complex Transformer models with minimal computing usage. Moreover, TSMixer's modular design enables compatibility with both supervised and masked self-supervised learning methods, making it a promising building block for time-series Foundation Models. TSMixer outperforms state-of-the-art MLP and Transformer models in forecasting by a considerable margin of 8-60%. It also outperforms the latest strong benchmarks of Patch-Transformer models (by 1-2%) with a significant reduction in memory and runtime (2-3X). The source code of our model is officially released as PatchTSMixer in the HuggingFace. Model: https://huggingface.co/docs/transformers/main/en/model_doc/patchtsmixer Examples: https://github.com/ibm/tsfm/#notebooks-links
In this paper, we evaluate feature extraction models for predicting speech quality. We also propose a model architecture to compare embeddings of supervised learning and self-supervised learning models with embeddings of speaker verification models to predict the metric MOS. Our experiments were performed on the VCC2018 dataset and a Brazilian-Portuguese dataset called BRSpeechMOS, which was created for this work. The results show that the Whisper model is appropriate in all scenarios: with both the VCC2018 and BRSpeech- MOS datasets. Among the supervised and self-supervised learning models using BRSpeechMOS, Whisper-Small achieved the best linear correlation of 0.6980, and the speaker verification model, SpeakerNet, had linear correlation of 0.6963. Using VCC2018, the best supervised and self-supervised learning model, Whisper-Large, achieved linear correlation of 0.7274, and the best model speaker verification, TitaNet, achieved a linear correlation of 0.6933. Although the results of the speaker verification models are slightly lower, the SpeakerNet model has only 5M parameters, making it suitable for real-time applications, and the TitaNet model produces an embedding of size 192, the smallest among all the evaluated models. The experiment results are reproducible with publicly available source-code1 .
Deep neural networks (DNNs) have been widely applied in medical image classification and achieve remarkable classification performance. These achievements heavily depend on large-scale accurately annotated training data. However, label noise is inevitably introduced in the medical image annotation, as the labeling process heavily relies on the expertise and experience of annotators. Meanwhile, DNNs suffer from overfitting noisy labels, degrading the performance of models. Therefore, in this work, we innovatively devise noise-robust training approach to mitigate the adverse effects of noisy labels in medical image classification. Specifically, we incorporate contrastive learning and intra-group attention mixup strategies into the vanilla supervised learning. The contrastive learning for feature extractor helps to enhance visual representation of DNNs. The intra-group attention mixup module constructs groups and assigns self-attention weights for group-wise samples, and subsequently interpolates massive noisy-suppressed samples through weighted mixup operation. We conduct comparative experiments on both synthetic and real-world noisy medical datasets under various noise levels. Rigorous experiments validate that our noise-robust method with contrastive learning and attention mixup can effectively handle with label noise, and is superior to state-of-the-art methods. An ablation study also shows that both components contribute to boost model performance. The proposed method demonstrates its capability of curb label noise and has certain potential toward real-world clinic applications.
Self-supervised learning (SSL) has shown significant progress in speech processing tasks. However, despite the intrinsic randomness in the Transformer structure, such as dropout variants and layer-drop, improving the model-level consistency remains under-explored in the speech SSL literature. To address this, we propose a new pre-training method that uses consistency regularization to improve Data2vec 2.0, the recent state-of-the-art (SOTA) SSL model. Specifically, the proposed method involves sampling two different student sub-models within the Data2vec 2.0 framework, enabling two output variants derived from a single input without additional parameters. Subsequently, we regularize the outputs from the student sub-models to be consistent and require them to predict the representation of the teacher model. Our experimental results demonstrate that the proposed approach improves the SSL model's robustness and generalization ability, resulting in SOTA results on the SUPERB benchmark.