Out-of-distribution (OOD) problems in few-shot classification (FSC) occur when novel classes sampled from testing distributions differ from base classes drawn from training distributions, which considerably degrades the performance of deep learning models deployed in real-world applications. Recent studies suggest that the OOD problems in FSC mainly including: (a) cross-domain few-shot classification (CD-FSC) and (b) spurious-correlation few-shot classification (SC-FSC). Specifically, CD-FSC occurs when a classifier learns transferring knowledge from base classes drawn from seen training distributions but recognizes novel classes sampled from unseen testing distributions. In contrast, SC-FSC arises when a classifier relies on non-causal features (or contexts) that happen to be correlated with the labels (or concepts) in base classes but such relationships no longer hold during the model deployment. Despite CD-FSC has been extensively studied, SC-FSC remains understudied due to lack of the corresponding evaluation benchmarks. To this end, we present Meta Concept Context (MetaCoCo), a benchmark with spurious-correlation shifts collected from real-world scenarios. Moreover, to quantify the extent of spurious-correlation shifts of the presented MetaCoCo, we further propose a metric by using CLIP as a pre-trained vision-language model. Extensive experiments on the proposed benchmark are performed to evaluate the state-of-the-art methods in FSC, cross-domain shifts, and self-supervised learning. The experimental results show that the performance of the existing methods degrades significantly in the presence of spurious-correlation shifts. We open-source all codes of our benchmark and hope that the proposed MetaCoCo can facilitate future research on spurious-correlation shifts problems in FSC. The code is available at: https://github.com/remiMZ/MetaCoCo-ICLR24.
In recent years, Event Sound Source Localization has been widely applied in various fields. Recent works typically relying on the contrastive learning framework show impressive performance. However, all work is based on large relatively simple datasets. It's also crucial to understand and analyze human behaviors (actions and interactions of people), voices, and sounds in chaotic events in many applications, e.g., crowd management, and emergency response services. In this paper, we apply the existing model to a more complex dataset, explore the influence of parameters on the model, and propose a semi-supervised improvement method SemiPL. With the increase in data quantity and the influence of label quality, self-supervised learning will be an unstoppable trend. The experiment shows that the parameter adjustment will positively affect the existing model. In particular, SSPL achieved an improvement of 12.2% cIoU and 0.56% AUC in Chaotic World compared to the results provided. The code is available at: https://github.com/ly245422/SSPL
Previous graph neural networks (GNNs) usually assume that the graph data is with clean labels for representation learning, but it is not true in real applications. In this paper, we propose a new multi-teacher distillation method based on bi-level optimization (namely BO-NNC), to conduct noisy node classification on the graph data. Specifically, we first employ multiple self-supervised learning methods to train diverse teacher models, and then aggregate their predictions through a teacher weight matrix. Furthermore, we design a new bi-level optimization strategy to dynamically adjust the teacher weight matrix based on the training progress of the student model. Finally, we design a label improvement module to improve the label quality. Extensive experimental results on real datasets show that our method achieves the best results compared to state-of-the-art methods.
We propose a self-supervised approach for learning physics-based subspaces for real-time simulation. Existing learning-based methods construct subspaces by approximating pre-defined simulation data in a purely geometric way. However, this approach tends to produce high-energy configurations, leads to entangled latent space dimensions, and generalizes poorly beyond the training set. To overcome these limitations, we propose a self-supervised approach that directly minimizes the system's mechanical energy during training. We show that our method leads to learned subspaces that reflect physical equilibrium constraints, resolve overfitting issues of previous methods, and offer interpretable latent space parameters.
Purpose: Paranasal anomalies, frequently identified in routine radiological screenings, exhibit diverse morphological characteristics. Due to the diversity of anomalies, supervised learning methods require large labelled dataset exhibiting diverse anomaly morphology. Self-supervised learning (SSL) can be used to learn representations from unlabelled data. However, there are no SSL methods designed for the downstream task of classifying paranasal anomalies in the maxillary sinus (MS). Methods: Our approach uses a 3D Convolutional Autoencoder (CAE) trained in an unsupervised anomaly detection (UAD) framework. Initially, we train the 3D CAE to reduce reconstruction errors when reconstructing normal maxillary sinus (MS) image. Then, this CAE is applied to an unlabelled dataset to generate coarse anomaly locations by creating residual MS images. Following this, a 3D Convolutional Neural Network (CNN) reconstructs these residual images, which forms our SSL task. Lastly, we fine-tune the encoder part of the 3D CNN on a labelled dataset of normal and anomalous MS images. Results: The proposed SSL technique exhibits superior performance compared to existing generic self-supervised methods, especially in scenarios with limited annotated data. When trained on just 10% of the annotated dataset, our method achieves an Area Under the Precision-Recall Curve (AUPRC) of 0.79 for the downstream classification task. This performance surpasses other methods, with BYOL attaining an AUPRC of 0.75, SimSiam at 0.74, SimCLR at 0.73 and Masked Autoencoding using SparK at 0.75. Conclusion: A self-supervised learning approach that inherently focuses on localizing paranasal anomalies proves to be advantageous, particularly when the subsequent task involves differentiating normal from anomalous maxillary sinuses. Access our code at https://github.com/mtec-tuhh/self-supervised-paranasal-anomaly
This paper presents a novel approach to processing multimodal data for dynamic emotion recognition, named as the Multimodal Masked Autoencoder for Dynamic Emotion Recognition (MultiMAE-DER). The MultiMAE-DER leverages the closely correlated representation information within spatiotemporal sequences across visual and audio modalities. By utilizing a pre-trained masked autoencoder model, the MultiMAEDER is accomplished through simple, straightforward finetuning. The performance of the MultiMAE-DER is enhanced by optimizing six fusion strategies for multimodal input sequences. These strategies address dynamic feature correlations within cross-domain data across spatial, temporal, and spatiotemporal sequences. In comparison to state-of-the-art multimodal supervised learning models for dynamic emotion recognition, MultiMAE-DER enhances the weighted average recall (WAR) by 4.41% on the RAVDESS dataset and by 2.06% on the CREMAD. Furthermore, when compared with the state-of-the-art model of multimodal self-supervised learning, MultiMAE-DER achieves a 1.86% higher WAR on the IEMOCAP dataset.
The classification of sleep stages is a pivotal aspect of diagnosing sleep disorders and evaluating sleep quality. However, the conventional manual scoring process, conducted by clinicians, is time-consuming and prone to human bias. Recent advancements in deep learning have substantially propelled the automation of sleep stage classification. Nevertheless, challenges persist, including the need for large datasets with labels and the inherent biases in human-generated annotations. This paper introduces NeuroNet, a self-supervised learning (SSL) framework designed to effectively harness unlabeled single-channel sleep electroencephalogram (EEG) signals by integrating contrastive learning tasks and masked prediction tasks. NeuroNet demonstrates superior performance over existing SSL methodologies through extensive experimentation conducted across three polysomnography (PSG) datasets. Additionally, this study proposes a Mamba-based temporal context module to capture the relationships among diverse EEG epochs. Combining NeuroNet with the Mamba-based temporal context module has demonstrated the capability to achieve, or even surpass, the performance of the latest supervised learning methodologies, even with a limited amount of labeled data. This study is expected to establish a new benchmark in sleep stage classification, promising to guide future research and applications in the field of sleep analysis.
To reduce the need for skilled clinicians in heart sound interpretation, recent studies on automating cardiac auscultation have explored deep learning approaches. However, despite the demands for large data for deep learning, the size of the heart sound datasets is limited, and no pre-trained model is available. On the contrary, many pre-trained models for general audio tasks are available as general-purpose audio representations. This study explores the potential of general-purpose audio representations pre-trained on large-scale datasets for transfer learning in heart murmur detection. Experiments on the CirCor DigiScope heart sound dataset show that the recent self-supervised learning Masked Modeling Duo (M2D) outperforms previous methods with the results of a weighted accuracy of 0.832 and an unweighted average recall of 0.713. Experiments further confirm improved performance by ensembling M2D with other models. These results demonstrate the effectiveness of general-purpose audio representation in processing heart sounds and open the way for further applications. Our code is available online which runs on a 24 GB consumer GPU at https://github.com/nttcslab/m2d/tree/master/app/circor
Self-Supervised Learning (SSL) is a valuable and robust training methodology for contemporary Deep Neural Networks (DNNs), enabling unsupervised pretraining on a 'pretext task' that does not require ground-truth labels/annotation. This allows efficient representation learning from massive amounts of unlabeled training data, which in turn leads to increased accuracy in a 'downstream task' by exploiting supervised transfer learning. Despite the relatively straightforward conceptualization and applicability of SSL, it is not always feasible to collect and/or to utilize very large pretraining datasets, especially when it comes to real-world application settings. In particular, in cases of specialized and domain-specific application scenarios, it may not be achievable or practical to assemble a relevant image pretraining dataset in the order of millions of instances or it could be computationally infeasible to pretrain at this scale, e.g., due to unavailability of sufficient computational resources that SSL methods typically require to produce improved visual analysis results. This situation motivates an investigation on the effectiveness of common SSL pretext tasks, when the pretraining dataset is of relatively limited/constrained size. This work briefly introduces the main families of modern visual SSL methods and, subsequently, conducts a thorough comparative experimental evaluation in the low-data regime, targeting to identify: a) what is learnt via low-data SSL pretraining, and b) how do different SSL categories behave in such training scenarios. Interestingly, for domain-specific downstream tasks, in-domain low-data SSL pretraining outperforms the common approach of large-scale pretraining on general datasets.
In an era where the volume of data drives the effectiveness of self-supervised learning, the specificity and clarity of data semantics play a crucial role in model training. Addressing this, we introduce HYPerbolic Entailment filtering (HYPE), a novel methodology designed to meticulously extract modality-wise meaningful and well-aligned data from extensive, noisy image-text pair datasets. Our approach leverages hyperbolic embeddings and the concept of entailment cones to evaluate and filter out samples with meaningless or underspecified semantics, focusing on enhancing the specificity of each data sample. HYPE not only demonstrates a significant improvement in filtering efficiency but also sets a new state-of-the-art in the DataComp benchmark when combined with existing filtering techniques. This breakthrough showcases the potential of HYPE to refine the data selection process, thereby contributing to the development of more accurate and efficient self-supervised learning models. Additionally, the image specificity $\epsilon_{i}$ can be independently applied to induce an image-only dataset from an image-text or image-only data pool for training image-only self-supervised models and showed superior performance when compared to the dataset induced by CLIP score.
Humans effortlessly interpret images by parsing them into part-whole hierarchies; deep learning excels in learning multi-level feature spaces, but they often lack explicit coding of part-whole relations, a prominent property of medical imaging. To overcome this limitation, we introduce Adam-v2, a new self-supervised learning framework extending Adam [79] by explicitly incorporating part-whole hierarchies into its learning objectives through three key branches: (1) Localizability, acquiring discriminative representations to distinguish different anatomical patterns; (2) Composability, learning each anatomical structure in a parts-to-whole manner; and (3) Decomposability, comprehending each anatomical structure in a whole-to-parts manner. Experimental results across 10 tasks, compared to 11 baselines in zero-shot, few-shot transfer, and full fine-tuning settings, showcase Adam-v2's superior performance over large-scale medical models and existing SSL methods across diverse downstream tasks. The higher generality and robustness of Adam-v2's representations originate from its explicit construction of hierarchies for distinct anatomical structures from unlabeled medical images. Adam-v2 preserves a semantic balance of anatomical diversity and harmony in its embedding, yielding representations that are both generic and semantically meaningful, yet overlooked in existing SSL methods. All code and pretrained models are available at https://github.com/JLiangLab/Eden.
Human Activity Recognition (HAR) based on the sensors of mobile/wearable devices aims to detect the physical activities performed by humans in their daily lives. Although supervised learning methods are the most effective in this task, their effectiveness is constrained to using a large amount of labeled data during training. While collecting raw unlabeled data can be relatively easy, annotating data is challenging due to costs, intrusiveness, and time constraints. To address these challenges, this paper explores alternative approaches for accurate HAR using a limited amount of labeled data. In particular, we have adapted recent Self-Supervised Learning (SSL) algorithms to the HAR domain and compared their effectiveness. We investigate three state-of-the-art SSL techniques of different families: contrastive, generative, and predictive. Additionally, we evaluate the impact of the underlying neural network on the recognition rate by comparing state-of-the-art CNN and transformer architectures. Our results show that a Masked Auto Encoder (MAE) approach significantly outperforms other SSL approaches, including SimCLR, commonly considered one of the best-performing SSL methods in the HAR domain. The code and the pre-trained SSL models are publicly available for further research and development.
Anomaly detection plays a crucial role in industrial settings, particularly in maintaining the reliability and optimal performance of cooling systems. Traditional anomaly detection methods often face challenges in handling diverse data characteristics and variations in noise levels, resulting in limited effectiveness. And yet traditional anomaly detection often relies on application of single models. This work proposes a novel, robust approach using five heterogeneous independent models combined with a dual ensemble fusion of voting techniques. Diverse models capture various system behaviors, while the fusion strategy maximizes detection effectiveness and minimizes false alarms. Each base autoencoder model learns a unique representation of the data, leveraging their complementary strengths to improve anomaly detection performance. To increase the effectiveness and reliability of final anomaly prediction, dual ensemble technique is applied. This approach outperforms in maximizing the coverage of identifying anomalies. Experimental results on a real-world dataset of industrial cooling system data demonstrate the effectiveness of the proposed approach. This approach can be extended to other industrial applications where anomaly detection is critical for ensuring system reliability and preventing potential malfunctions.
Recent advancements in self-supervised learning in the point cloud domain have demonstrated significant potential. However, these methods often suffer from drawbacks, including lengthy pre-training time, the necessity of reconstruction in the input space, or the necessity of additional modalities. In order to address these issues, we introduce Point-JEPA, a joint embedding predictive architecture designed specifically for point cloud data. To this end, we introduce a sequencer that orders point cloud patch embeddings to efficiently compute and utilize their proximity based on the indices during target and context selection. The sequencer also allows shared computations of the patch embeddings' proximity between context and target selection, further improving the efficiency. Experimentally, our method achieves competitive results with state-of-the-art methods while avoiding the reconstruction in the input space or additional modality.
Etruscan mirrors constitute a significant category within Etruscan art and, therefore, undergo systematic examinations to obtain insights into ancient times. A crucial aspect of their analysis involves the labor-intensive task of manually tracing engravings from the backside. Additionally, this task is inherently challenging due to the damage these mirrors have sustained, introducing subjectivity into the process. We address these challenges by automating the process through photometric-stereo scanning in conjunction with deep segmentation networks which, however, requires effective usage of the limited data at hand. We accomplish this by incorporating predictions on a per-patch level, and various data augmentations, as well as exploring self-supervised learning. Compared to our baseline, we improve predictive performance w.r.t. the pseudo-F-Measure by around 16%. When assessing performance on complete mirrors against a human baseline, our approach yields quantitative similar performance to a human annotator and significantly outperforms existing binarization methods. With our proposed methodology, we streamline the annotation process, enhance its objectivity, and reduce overall workload, offering a valuable contribution to the examination of these historical artifacts and other non-traditional documents.
The Vision Transformer (ViT) has demonstrated remarkable performance in Self-Supervised Learning (SSL) for 3D medical image analysis. Masked AutoEncoder (MAE) for feature pre-training can further unleash the potential of ViT on various medical vision tasks. However, due to large spatial sizes with much higher dimensions of 3D medical images, the lack of hierarchical design for MAE may hinder the performance of downstream tasks. In this paper, we propose a novel \textit{Mask in Mask (MiM)} pre-training framework for 3D medical images, which aims to advance MAE by learning discriminative representation from hierarchical visual tokens across varying scales. We introduce multiple levels of granularity for masked inputs from the volume, which are then reconstructed simultaneously ranging at both fine and coarse levels. Additionally, a cross-level alignment mechanism is applied to adjacent level volumes to enforce anatomical similarity hierarchically. Furthermore, we adopt a hybrid backbone to enhance the hierarchical representation learning efficiently during the pre-training. MiM was pre-trained on a large scale of available 3D volumetric images, \textit{i.e.,} Computed Tomography (CT) images containing various body parts. Extensive experiments on thirteen public datasets demonstrate the superiority of MiM over other SSL methods in organ/lesion/tumor segmentation and disease classification. We further scale up the MiM to large pre-training datasets with more than 10k volumes, showing that large-scale pre-training can further enhance the performance of downstream tasks. The improvement also concluded that the research community should pay more attention to the scale of the pre-training dataset towards the healthcare foundation model for 3D medical images.
Machine learning techniques have shown remarkable accuracy in localization tasks, but their dependency on vast amounts of labeled data, particularly Channel State Information (CSI) and corresponding coordinates, remains a bottleneck. Self-supervised learning techniques alleviate the need for labeled data, a potential that remains largely untapped and underexplored in existing research. Addressing this gap, we propose a pioneering approach that leverages self-supervised pretraining on unlabeled data to boost the performance of supervised learning for user localization based on CSI. We introduce two pretraining Auto Encoder (AE) models employing Multi Layer Perceptrons (MLPs) and Convolutional Neural Networks (CNNs) to glean representations from unlabeled data via self-supervised learning. Following this, we utilize the encoder portion of the AE models to extract relevant features from labeled data, and finetune an MLP-based Position Estimation Model to accurately deduce user locations. Our experimentation on the CTW-2020 dataset, which features a substantial volume of unlabeled data but limited labeled samples, demonstrates the viability of our approach. Notably, the dataset covers a vast area spanning over 646x943x41 meters, and our approach demonstrates promising results even for such expansive localization tasks.
Self-Supervised Learning (SSL) frameworks became the standard for learning robust class representations by benefiting from large unlabeled datasets. For Speaker Verification (SV), most SSL systems rely on contrastive-based loss functions. We explore different ways to improve the performance of these techniques by revisiting the NT-Xent contrastive loss. Our main contribution is the definition of the NT-Xent-AM loss and the study of the importance of Additive Margin (AM) in SimCLR and MoCo SSL methods to further separate positive from negative pairs. Despite class collisions, we show that AM enhances the compactness of same-speaker embeddings and reduces the number of false negatives and false positives on SV. Additionally, we demonstrate the effectiveness of the symmetric contrastive loss, which provides more supervision for the SSL task. Implementing these two modifications to SimCLR improves performance and results in 7.85% EER on VoxCeleb1-O, outperforming other equivalent methods.
This paper explores the integration of human-like emotions and ethical considerations into Large Language Models (LLMs). We first model eight fundamental human emotions, presented as opposing pairs, and employ collaborative LLMs to reinterpret and express these emotions across a spectrum of intensity. Our focus extends to embedding a latent ethical dimension within LLMs, guided by a novel self-supervised learning algorithm with human feedback (SSHF). This approach enables LLMs to perform self-evaluations and adjustments concerning ethical guidelines, enhancing their capability to generate content that is not only emotionally resonant but also ethically aligned. The methodologies and case studies presented herein illustrate the potential of LLMs to transcend mere text and image generation, venturing into the realms of empathetic interaction and principled decision-making, thereby setting a new precedent in the development of emotionally aware and ethically conscious AI systems.
The majority of existing hyperspectral anomaly detection (HAD) methods use the low-rank representation (LRR) model to separate the background and anomaly components, where the anomaly component is optimized by handcrafted sparse priors (e.g., $\ell_{2,1}$-norm). However, this may not be ideal since they overlook the spatial structure present in anomalies and make the detection result largely dependent on manually set sparsity. To tackle these problems, we redefine the optimization criterion for the anomaly component in the LRR model with a self-supervised network called self-supervised anomaly prior (SAP). This prior is obtained by the pretext task of self-supervised learning, which is customized to learn the characteristics of hyperspectral anomalies. Specifically, this pretext task is a classification task to distinguish the original hyperspectral image (HSI) and the pseudo-anomaly HSI, where the pseudo-anomaly is generated from the original HSI and designed as a prism with arbitrary polygon bases and arbitrary spectral bands. In addition, a dual-purified strategy is proposed to provide a more refined background representation with an enriched background dictionary, facilitating the separation of anomalies from complex backgrounds. Extensive experiments on various hyperspectral datasets demonstrate that the proposed SAP offers a more accurate and interpretable solution than other advanced HAD methods.