PyTorchKR Papers

최신 논문

총 6,725편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

논문 ID: 1783.36123

A Prior Instruction Representation Framework for Remote Sensing Image-text Retrieval

This paper presents a prior instruction representation framework (PIR) for remote sensing image-text retrieval, aimed at remote sensing vision-language understanding tasks to solve the semantic noise problem. Our highlight is the proposal of a paradigm that draws on prior knowledge to instruct adaptive learning of vision and text representations. Concretely, two progressive attention encoder (PAE) structures, Spatial-PAE and Temporal-PAE, are proposed to perform long-range dependency modeling to enhance key feature representation. In vision representation, Vision Instruction Representation (VIR) based on Spatial-PAE exploits the prior-guided knowledge of the remote sensing scene recognition by building a belief matrix to select key features for reducing the impact of semantic noise. In text representation, Language Cycle Attention (LCA) based on Temporal-PAE uses the previous time step to cyclically activate the current time step to enhance text representation capability. A cluster-wise affiliation loss is proposed to constrain the inter-classes and to reduce the semantic confusion zones in the common subspace. Comprehensive experiments demonstrate that using prior knowledge instruction could enhance vision and text representations and could outperform the state-of-the-art methods on two benchmark datasets, RSICD and RSITMD.

arXiv:2312.03350

PointMoment:Mixed-Moment-based Self-Supervised Representation Learning for 3D Point Clouds

Large and rich data is a prerequisite for effective training of deep neural networks. However, the irregularity of point cloud data makes manual annotation time-consuming and laborious. Self-supervised representation learning, which leverages the intrinsic structure of large-scale unlabelled data to learn meaningful feature representations, has attracted increasing attention in the field of point cloud research. However, self-supervised representation learning often suffers from model collapse, resulting in reduced information and diversity of the learned representation, and consequently degrading the performance of downstream tasks. To address this problem, we propose PointMoment, a novel framework for point cloud self-supervised representation learning that utilizes a high-order mixed moment loss function rather than the conventional contrastive loss function. Moreover, our framework does not require any special techniques such as asymmetric network architectures, gradient stopping, etc. Specifically, we calculate the high-order mixed moment of the feature variables and force them to decompose into products of their individual moment, thereby making multiple variables more independent and minimizing the feature redundancy. We also incorporate a contrastive learning approach to maximize the feature invariance under different data augmentations of the same point cloud. Experimental results show that our approach outperforms previous unsupervised learning methods on the downstream task of 3D point cloud classification and segmentation.

arXiv:2312.03318

Complementary Benefits of Contrastive Learning and Self-Training Under Distribution Shift

Self-training and contrastive learning have emerged as leading techniques for incorporating unlabeled data, both under distribution shift (unsupervised domain adaptation) and when it is absent (semi-supervised learning). However, despite the popularity and compatibility of these techniques, their efficacy in combination remains unexplored. In this paper, we undertake a systematic empirical investigation of this combination, finding that (i) in domain adaptation settings, self-training and contrastive learning offer significant complementary gains; and (ii) in semi-supervised learning settings, surprisingly, the benefits are not synergistic. Across eight distribution shift datasets (e.g., BREEDs, WILDS), we demonstrate that the combined method obtains 3--8% higher accuracy than either approach independently. We then theoretically analyze these techniques in a simplified model of distribution shift, demonstrating scenarios under which the features produced by contrastive learning can yield a good initialization for self-training to further amplify gains and achieve optimal performance, even when either method alone would fail.

arXiv:2312.02791

Unsupervised learning on spontaneous retinal activity leads to efficient neural representation geometry

Prior to the onset of vision, neurons in the developing mammalian retina spontaneously fire in correlated activity patterns known as retinal waves. Experimental evidence suggests that retinal waves strongly influence the emergence of sensory representations before visual experience. We aim to model this early stage of functional development by using movies of neurally active developing retinas as pre-training data for neural networks. Specifically, we pre-train a ResNet-18 with an unsupervised contrastive learning objective (SimCLR) on both simulated and experimentally-obtained movies of retinal waves, then evaluate its performance on image classification tasks. We find that pre-training on retinal waves significantly improves performance on tasks that test object invariance to spatial translation, while slightly improving performance on more complex tasks like image classification. Notably, these performance boosts are realized on held-out natural images even though the pre-training procedure does not include any natural image data. We then propose a geometrical explanation for the increase in network performance, namely that the spatiotemporal characteristics of retinal waves facilitate the formation of separable feature representations. In particular, we demonstrate that networks pre-trained on retinal waves are more effective at separating image manifolds than randomly initialized networks, especially for manifolds defined by sets of spatial translations. These findings indicate that the broad spatiotemporal properties of retinal waves prepare networks for higher order feature extraction.

arXiv:2302.06675

Symbolic Discovery of Optimization Algorithms

We present a method to formulate algorithm discovery as program search, and apply it to discover optimization algorithms for deep neural network training. We leverage efficient search techniques to explore an infinite and sparse program space. To bridge the large generalization gap between proxy and target tasks, we also introduce program selection and simplification strategies. Our method discovers a simple and effective optimization algorithm, $\textbf{Lion}$ ($\textit{Evo$\textbf{L}$ved S$\textbf{i}$gn M$\textbf{o}$me$\textbf{n}$tum}$). It is more memory-efficient than Adam as it only keeps track of the momentum. Different from adaptive optimizers, its update has the same magnitude for each parameter calculated through the sign operation. We compare Lion with widely used optimizers, such as Adam and Adafactor, for training a variety of models on different tasks. On image classification, Lion boosts the accuracy of ViT by up to 2% on ImageNet and saves up to 5x the pre-training compute on JFT. On vision-language contrastive learning, we achieve 88.3% $\textit{zero-shot}$ and 91.1% $\textit{fine-tuning}$ accuracy on ImageNet, surpassing the previous best results by 2% and 0.1%, respectively. On diffusion models, Lion outperforms Adam by achieving a better FID score and reducing the training compute by up to 2.3x. For autoregressive, masked language modeling, and fine-tuning, Lion exhibits a similar or better performance compared to Adam. Our analysis of Lion reveals that its performance gain grows with the training batch size. It also requires a smaller learning rate than Adam due to the larger norm of the update produced by the sign function. Additionally, we examine the limitations of Lion and identify scenarios where its improvements are small or not statistically significant. Lion is also successfully deployed in production systems such as Google search ads CTR model.

arXiv:2312.02227

Improving Multimodal Sentiment Analysis: Supervised Angular Margin-based Contrastive Learning for Enhanced Fusion Representation

The effectiveness of a model is heavily reliant on the quality of the fusion representation of multiple modalities in multimodal sentiment analysis. Moreover, each modality is extracted from raw input and integrated with the rest to construct a multimodal representation. Although previous methods have proposed multimodal representations and achieved promising results, most of them focus on forming positive and negative pairs, neglecting the variation in sentiment scores within the same class. Additionally, they fail to capture the significance of unimodal representations in the fusion vector. To address these limitations, we introduce a framework called Supervised Angular-based Contrastive Learning for Multimodal Sentiment Analysis. This framework aims to enhance discrimination and generalizability of the multimodal representation and overcome biases in the fusion vector's modality. Our experimental results, along with visualizations on two widely used datasets, demonstrate the effectiveness of our approach.

arXiv:2312.02619공식 코드 · Papers with Code

Rethinking and Simplifying Bootstrapped Graph Latents

Graph contrastive learning (GCL) has emerged as a representative paradigm in graph self-supervised learning, where negative samples are commonly regarded as the key to preventing model collapse and producing distinguishable representations. Recent studies have shown that GCL without negative samples can achieve state-of-the-art performance as well as scalability improvement, with bootstrapped graph latent (BGRL) as a prominent step forward. However, BGRL relies on a complex architecture to maintain the ability to scatter representations, and the underlying mechanisms enabling the success remain largely unexplored. In this paper, we introduce an instance-level decorrelation perspective to tackle the aforementioned issue and leverage it as a springboard to reveal the potential unnecessary model complexity within BGRL. Based on our findings, we present SGCL, a simple yet effective GCL framework that utilizes the outputs from two consecutive iterations as positive pairs, eliminating the negative samples. SGCL only requires a single graph augmentation and a single graph encoder without additional parameters. Extensive experiments conducted on various graph benchmarks demonstrate that SGCL can achieve competitive performance with fewer parameters, lower time and space costs, and significant convergence speedup.

arXiv:2312.02235

CryoGEM: Physics-Informed Generative Cryo-Electron Microscopy

In the past decade, deep conditional generative models have revolutionized the generation of realistic images, extending their application from entertainment to scientific domains. Single-particle cryo-electron microscopy (cryo-EM) is crucial in resolving near-atomic resolution 3D structures of proteins, such as the SARS- COV-2 spike protein. To achieve high-resolution reconstruction, a comprehensive data processing pipeline has been adopted. However, its performance is still limited as it lacks high-quality annotated datasets for training. To address this, we introduce physics-informed generative cryo-electron microscopy (CryoGEM), which for the first time integrates physics-based cryo-EM simulation with a generative unpaired noise translation to generate physically correct synthetic cryo-EM datasets with realistic noises. Initially, CryoGEM simulates the cryo-EM imaging process based on a virtual specimen. To generate realistic noises, we leverage an unpaired noise translation via contrastive learning with a novel mask-guided sampling scheme. Extensive experiments show that CryoGEM is capable of generating authentic cryo-EM images. The generated dataset can used as training data for particle picking and pose estimation models, eventually improving the reconstruction resolution.

arXiv:2312.02240

Contrastive Learning-Based Spectral Knowledge Distillation for Multi-Modality and Missing Modality Scenarios in Semantic Segmentation

Improving the performance of semantic segmentation models using multispectral information is crucial, especially for environments with low-light and adverse conditions. Multi-modal fusion techniques pursue either the learning of cross-modality features to generate a fused image or engage in knowledge distillation but address multimodal and missing modality scenarios as distinct issues, which is not an optimal approach for multi-sensor models. To address this, a novel multi-modal fusion approach called CSK-Net is proposed, which uses a contrastive learning-based spectral knowledge distillation technique along with an automatic mixed feature exchange mechanism for semantic segmentation in optical (EO) and infrared (IR) images. The distillation scheme extracts detailed textures from the optical images and distills them into the optical branch of CSK-Net. The model encoder consists of shared convolution weights with separate batch norm (BN) layers for both modalities, to capture the multi-spectral information from different modalities of the same objects. A Novel Gated Spectral Unit (GSU) and mixed feature exchange strategy are proposed to increase the correlation of modality-shared information and decrease the modality-specific information during the distillation process. Comprehensive experiments show that CSK-Net surpasses state-of-the-art models in multi-modal tasks and for missing modalities when exclusively utilizing IR data for inference across three public benchmarking datasets. For missing modality scenarios, the performance increase is achieved without additional computational costs compared to the baseline segmentation models.

arXiv:2312.02185

Virtual Fusion with Contrastive Learning for Single Sensor-based Activity Recognition

Various types of sensors can be used for Human Activity Recognition (HAR), and each of them has different strengths and weaknesses. Sometimes a single sensor cannot fully observe the user's motions from its perspective, which causes wrong predictions. While sensor fusion provides more information for HAR, it comes with many inherent drawbacks like user privacy and acceptance, costly set-up, operation, and maintenance. To deal with this problem, we propose Virtual Fusion - a new method that takes advantage of unlabeled data from multiple time-synchronized sensors during training, but only needs one sensor for inference. Contrastive learning is adopted to exploit the correlation among sensors. Virtual Fusion gives significantly better accuracy than training with the same single sensor, and in some cases, it even surpasses actual fusion using multiple sensors at test time. We also extend this method to a more general version called Actual Fusion within Virtual Fusion (AFVF), which uses a subset of training sensors during inference. Our method achieves state-of-the-art accuracy and F1-score on UCI-HAR and PAMAP2 benchmark datasets. Implementation is available upon request.

arXiv:2312.02545

Graph Information Bottleneck for Remote Sensing Segmentation

Remote sensing segmentation has a wide range of applications in environmental protection, and urban change detection, etc. Despite the success of deep learning-based remote sensing segmentation methods (e.g., CNN and Transformer), they are not flexible enough to model irregular objects. In addition, existing graph contrastive learning methods usually adopt the way of maximizing mutual information to keep the node representations consistent between different graph views, which may cause the model to learn task-independent redundant information. To tackle the above problems, this paper treats images as graph structures and introduces a simple contrastive vision GNN (SC-ViG) architecture for remote sensing segmentation. Specifically, we construct a node-masked and edge-masked graph view to obtain an optimal graph structure representation, which can adaptively learn whether to mask nodes and edges. Furthermore, this paper innovatively introduces information bottleneck theory into graph contrastive learning to maximize task-related information while minimizing task-independent redundant information. Finally, we replace the convolutional module in UNet with the SC-ViG module to complete the segmentation and classification tasks of remote sensing images. Extensive experiments on publicly available real datasets demonstrate that our method outperforms state-of-the-art remote sensing image segmentation methods.

arXiv:2312.01629

CLAMP: Contrastive LAnguage Model Prompt-tuning

Large language models (LLMs) have emerged as powerful general-purpose interfaces for many machine learning problems. Recent work has adapted LLMs to generative visual tasks like image captioning, visual question answering, and visual chat, using a relatively small amount of instruction-tuning data. In this paper, we explore whether modern LLMs can also be adapted to classifying an image into a set of categories. First, we evaluate multimodal LLMs that are tuned for generative tasks on zero-shot image classification and find that their performance is far below that of specialized models like CLIP. We then propose an approach for light fine-tuning of LLMs using the same contrastive image-caption matching objective as CLIP. Our results show that LLMs can, indeed, achieve good image classification performance when adapted this way. Our approach beats state-of-the-art mLLMs by 13% and slightly outperforms contrastive learning with a custom text model, while also retaining the LLM's generative abilities. LLM initialization appears to particularly help classification in domains under-represented in the visual pre-training data.

arXiv:2312.01529

T3D: Advancing 3D Medical Vision-Language Pre-training by Learning Multi-View Visual Consistency

While 3D visual self-supervised learning (vSSL) shows promising results in capturing visual representations, it overlooks the clinical knowledge from radiology reports. Meanwhile, 3D medical vision-language pre-training (MedVLP) remains underexplored due to the lack of a large-scale, publicly available 3D medical image-report dataset. To bridge this gap, we introduce **CT-3DVLP**, the first and largest **public** 3D volume-report dataset, establishing a comprehensive benchmark for 3D MedVLP research. Meanwhile, we propose the **T3D** framework, which enhances 3D MedVLP beyond naive CLIP-style alignment that directly pairs volumes with reports but neglects local visual representations. Instead, we introduce **Text-informed Multi-view Alignment (TMA)**, a novel approach that clusters volumetric data while enforcing consistency across different views of the same volume-report pair. TMA integrates textual features into fine-grained visual representations, ensuring contextual coherence across views. We evaluate T3D across multiple downstream tasks in both unimodal and cross-modal settings, including zero-shot and fine-tuned classification, cross-modal retrieval, report generation, and semantic segmentation. Our results show that T3D consistently outperforms existing vSSL and multimodal methods, demonstrating superior zero-shot and fine-tuning capabilities and setting a new benchmark for 3D medical image understanding.

arXiv:2312.01753

Long-Tail Learning with Rebalanced Contrastive Loss

Integrating supervised contrastive loss to cross entropy-based communication has recently been proposed as a solution to address the long-tail learning problem. However, when the class imbalance ratio is high, it requires adjusting the supervised contrastive loss to support the tail classes, as the conventional contrastive learning is biased towards head classes by default. To this end, we present Rebalanced Contrastive Learning (RCL), an efficient means to increase the long tail classification accuracy by addressing three main aspects: 1. Feature space balancedness - Equal division of the feature space among all the classes, 2. Intra-Class compactness - Reducing the distance between same-class embeddings, 3. Regularization - Enforcing larger margins for tail classes to reduce overfitting. RCL adopts class frequency-based SoftMax loss balancing to supervised contrastive learning loss and exploits scalar multiplied features fed to the contrastive learning loss to enforce compactness. We implement RCL on the Balanced Contrastive Learning (BCL) Framework, which has the SOTA performance. Our experiments on three benchmark datasets demonstrate the richness of the learnt embeddings and increased top-1 balanced accuracy RCL provides to the BCL framework. We further demonstrate that the performance of RCL as a standalone loss also achieves state-of-the-art level accuracy.

arXiv:2312.00966

Spectral Temporal Contrastive Learning

Learning useful data representations without requiring labels is a cornerstone of modern deep learning. Self-supervised learning methods, particularly contrastive learning (CL), have proven successful by leveraging data augmentations to define positive pairs. This success has prompted a number of theoretical studies to better understand CL and investigate theoretical bounds for downstream linear probing tasks. This work is concerned with the temporal contrastive learning (TCL) setting where the sequential structure of the data is used instead to define positive pairs, which is more commonly used in RL and robotics contexts. In this paper, we adapt recent work on Spectral CL to formulate Spectral Temporal Contrastive Learning (STCL). We discuss a population loss based on a state graph derived from a time-homogeneous reversible Markov chain with uniform stationary distribution. The STCL loss enables to connect the linear probing performance to the spectral properties of the graph, and can be estimated by considering previously observed data sequences as an ensemble of MCMC chains.

arXiv:2312.02151공식 코드 · Papers with Code

Guarding Barlow Twins Against Overfitting with Mixed Samples

Self-supervised Learning (SSL) aims to learn transferable feature representations for downstream applications without relying on labeled data. The Barlow Twins algorithm, renowned for its widespread adoption and straightforward implementation compared to its counterparts like contrastive learning methods, minimizes feature redundancy while maximizing invariance to common corruptions. Optimizing for the above objective forces the network to learn useful representations, while avoiding noisy or constant features, resulting in improved downstream task performance with limited adaptation. Despite Barlow Twins' proven effectiveness in pre-training, the underlying SSL objective can inadvertently cause feature overfitting due to the lack of strong interaction between the samples unlike the contrastive learning approaches. From our experiments, we observe that optimizing for the Barlow Twins objective doesn't necessarily guarantee sustained improvements in representation quality beyond a certain pre-training phase, and can potentially degrade downstream performance on some datasets. To address this challenge, we introduce Mixed Barlow Twins, which aims to improve sample interaction during Barlow Twins training via linearly interpolated samples. This results in an additional regularization term to the original Barlow Twins objective, assuming linear interpolation in the input space translates to linearly interpolated features in the feature space. Pre-training with this regularization effectively mitigates feature overfitting and further enhances the downstream performance on CIFAR-10, CIFAR-100, TinyImageNet, STL-10, and ImageNet datasets. The code and checkpoints are available at: https://github.com/wgcban/mix-bt.git

arXiv:2312.01601

Local-Global History-aware Contrastive Learning for Temporal Knowledge Graph Reasoning

Temporal knowledge graphs (TKGs) have been identified as a promising approach to represent the dynamics of facts along the timeline. The extrapolation of TKG is to predict unknowable facts happening in the future, holding significant practical value across diverse fields. Most extrapolation studies in TKGs focus on modeling global historical fact repeating and cyclic patterns, as well as local historical adjacent fact evolution patterns, showing promising performance in predicting future unknown facts. Yet, existing methods still face two major challenges: (1) They usually neglect the importance of historical information in KG snapshots related to the queries when encoding the local and global historical information; (2) They exhibit weak anti-noise capabilities, which hinders their performance when the inputs are contaminated with noise.To this end, we propose a novel \blue{Lo}cal-\blue{g}lobal history-aware \blue{C}ontrastive \blue{L}earning model (\blue{LogCL}) for TKG reasoning, which adopts contrastive learning to better guide the fusion of local and global historical information and enhance the ability to resist interference. Specifically, for the first challenge, LogCL proposes an entity-aware attention mechanism applied to the local and global historical facts encoder, which captures the key historical information related to queries. For the latter issue, LogCL designs four historical query contrast patterns, effectively improving the robustness of the model. The experimental results on four benchmark datasets demonstrate that LogCL delivers better and more robust performance than the state-of-the-art baselines.

arXiv:2312.01241

Just-in-Time Detection of Silent Security Patches

Open-source code is pervasive. In this setting, embedded vulnerabilities are spreading to downstream software at an alarming rate. While such vulnerabilities are generally identified and addressed rapidly, inconsistent maintenance policies may lead security patches to go unnoticed. Indeed, security patches can be {\em silent}, i.e., they do not always come with comprehensive advisories such as CVEs. This lack of transparency leaves users oblivious to available security updates, providing ample opportunity for attackers to exploit unpatched vulnerabilities. Consequently, identifying silent security patches just in time when they are released is essential for preventing n-day attacks, and for ensuring robust and secure maintenance practices. With LLMDA we propose to (1) leverage large language models (LLMs) to augment patch information with generated code change explanations, (2) design a representation learning approach that explores code-text alignment methodologies for feature combination, (3) implement a label-wise training with labelled instructions for guiding the embedding based on security relevance, and (4) rely on a probabilistic batch contrastive learning mechanism for building a high-precision identifier of security patches. We evaluate LLMDA on the PatchDB and SPI-DB literature datasets and show that our approach substantially improves over the state-of-the-art, notably GraphSPD by 20% in terms of F-Measure on the SPI-DB benchmark.

arXiv:2312.01191공식 코드 · Papers with Code

Bootstrapping Interactive Image-Text Alignment for Remote Sensing Image Captioning

Recently, remote sensing image captioning has gained significant attention in the remote sensing community. Due to the significant differences in spatial resolution of remote sensing images, existing methods in this field have predominantly concentrated on the fine-grained extraction of remote sensing image features, but they cannot effectively handle the semantic consistency between visual features and textual features. To efficiently align the image-text, we propose a novel two-stage vision-language pre-training-based approach to bootstrap interactive image-text alignment for remote sensing image captioning, called BITA, which relies on the design of a lightweight interactive Fourier Transformer to better align remote sensing image-text features. The Fourier layer in the interactive Fourier Transformer is capable of extracting multi-scale features of remote sensing images in the frequency domain, thereby reducing the redundancy of remote sensing visual features. Specifically, the first stage involves preliminary alignment through image-text contrastive learning, which aligns the learned multi-scale remote sensing features from the interactive Fourier Transformer with textual features. In the second stage, the interactive Fourier Transformer connects the frozen image encoder with a large language model. Then, prefix causal language modeling is utilized to guide the text generation process using visual features. Ultimately, across the UCM-caption, RSICD, and NWPU-caption datasets, the experimental results clearly demonstrate that BITA outperforms other advanced comparative approaches. The code is available at https://github.com/yangcong356/BITA.

arXiv:2312.00552공식 코드 · Papers with Code

Improving Unsupervised Relation Extraction by Augmenting Diverse Sentence Pairs

Unsupervised relation extraction (URE) aims to extract relations between named entities from raw text without requiring manual annotations or pre-existing knowledge bases. In recent studies of URE, researchers put a notable emphasis on contrastive learning strategies for acquiring relation representations. However, these studies often overlook two important aspects: the inclusion of diverse positive pairs for contrastive learning and the exploration of appropriate loss functions. In this paper, we propose AugURE with both within-sentence pairs augmentation and augmentation through cross-sentence pairs extraction to increase the diversity of positive pairs and strengthen the discriminative power of contrastive learning. We also identify the limitation of noise-contrastive estimation (NCE) loss for relation representation learning and propose to apply margin loss for sentence pairs. Experiments on NYT-FB and TACRED datasets demonstrate that the proposed relation representation learning and a simple K-Means clustering achieves state-of-the-art performance.