PyTorchKR Papers

최신 논문

총 5,358편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2206.07259공식 코드 · Papers with Code

Self-Supervised Learning of Image Scale and Orientation

We study the problem of learning to assign a characteristic pose, i.e., scale and orientation, for an image region of interest. Despite its apparent simplicity, the problem is non-trivial; it is hard to obtain a large-scale set of image regions with explicit pose annotations that a model directly learns from. To tackle the issue, we propose a self-supervised learning framework with a histogram alignment technique. It generates pairs of image patches by random rescaling/rotating and then train an estimator to predict their scale/orientation values so that their relative difference is consistent with the rescaling/rotating used. The estimator learns to predict a non-parametric histogram distribution of scale/orientation without any supervision. Experiments show that it significantly outperforms previous methods in scale/orientation estimation and also improves image matching and 6 DoF camera pose estimation by incorporating our patch poses into a matching process.

arXiv:2206.07692공식 코드 · Papers with Code

A Simple Data Mixing Prior for Improving Self-Supervised Learning

Data mixing (e.g., Mixup, Cutmix, ResizeMix) is an essential component for advancing recognition models. In this paper, we focus on studying its effectiveness in the self-supervised setting. By noticing the mixed images that share the same source images are intrinsically related to each other, we hereby propose SDMP, short for $\textbf{S}$imple $\textbf{D}$ata $\textbf{M}$ixing $\textbf{P}$rior, to capture this straightforward yet essential prior, and position such mixed images as additional $\textbf{positive pairs}$ to facilitate self-supervised representation learning. Our experiments verify that the proposed SDMP enables data mixing to help a set of self-supervised learning frameworks (e.g., MoCo) achieve better accuracy and out-of-distribution robustness. More notably, our SDMP is the first method that successfully leverages data mixing to improve (rather than hurt) the performance of Vision Transformers in the self-supervised setting. Code is publicly available at https://github.com/OliverRensu/SDMP

arXiv:2206.07434

Self-Supervised Implicit Attention: Guided Attention by The Model Itself

We propose Self-Supervised Implicit Attention (SSIA), a new approach that adaptively guides deep neural network models to gain attention by exploiting the properties of the models themselves. SSIA is a novel attention mechanism that does not require any extra parameters, computation, or memory access costs during inference, which is in contrast to existing attention mechanism. In short, by considering attention weights as higher-level semantic information, we reconsidered the implementation of existing attention mechanisms and further propose generating supervisory signals from higher network layers to guide lower network layers for parameter updates. We achieved this by building a self-supervised learning task using the hierarchical features of the network itself, which only works at the training stage. To verify the effectiveness of SSIA, we performed a particular implementation (called an SSIA block) in convolutional neural network models and validated it on several image classification datasets. The experimental results show that an SSIA block can significantly improve the model performance, even outperforms many popular attention methods that require additional parameters and computation costs, such as Squeeze-and-Excitation and Convolutional Block Attention Module. Our implementation will be available on GitHub.

arXiv:2206.07240

Test-Time Adaptation for Visual Document Understanding

For visual document understanding (VDU), self-supervised pretraining has been shown to successfully generate transferable representations, yet, effective adaptation of such representations to distribution shifts at test-time remains to be an unexplored area. We propose DocTTA, a novel test-time adaptation method for documents, that does source-free domain adaptation using unlabeled target document data. DocTTA leverages cross-modality self-supervised learning via masked visual language modeling, as well as pseudo labeling to adapt models learned on a \textit{source} domain to an unlabeled \textit{target} domain at test time. We introduce new benchmarks using existing public datasets for various VDU tasks, including entity recognition, key-value extraction, and document visual question answering. DocTTA shows significant improvements on these compared to the source model performance, up to 1.89\% in (F1 score), 3.43\% (F1 score), and 17.68\% (ANLS score), respectively. Our benchmark datasets are available at \url{https://saynaebrahimi.github.io/DocTTA.html}.

arXiv:2206.07700

Masked Siamese ConvNets

Self-supervised learning has shown superior performances over supervised methods on various vision benchmarks. The siamese network, which encourages embeddings to be invariant to distortions, is one of the most successful self-supervised visual representation learning approaches. Among all the augmentation methods, masking is the most general and straightforward method that has the potential to be applied to all kinds of input and requires the least amount of domain knowledge. However, masked siamese networks require particular inductive bias and practically only work well with Vision Transformers. This work empirically studies the problems behind masked siamese networks with ConvNets. We propose several empirical designs to overcome these problems gradually. Our method performs competitively on low-shot image classification and outperforms previous methods on object detection benchmarks. We discuss several remaining issues and hope this work can provide useful data points for future general-purpose self-supervised learning.

arXiv:2206.06606공식 코드 · Papers with Code

Astock: A New Dataset and Automated Stock Trading based on Stock-specific News Analyzing Model

Natural Language Processing(NLP) demonstrates a great potential to support financial decision-making by analyzing the text from social media or news outlets. In this work, we build a platform to study the NLP-aided stock auto-trading algorithms systematically. In contrast to the previous work, our platform is characterized by three features: (1) We provide financial news for each specific stock. (2) We provide various stock factors for each stock. (3) We evaluate performance from more financial-relevant metrics. Such a design allows us to develop and evaluate NLP-aided stock auto-trading algorithms in a more realistic setting. In addition to designing an evaluation platform and dataset collection, we also made a technical contribution by proposing a system to automatically learn a good feature representation from various input information. The key to our algorithm is a method called semantic role labeling Pooling (SRLP), which leverages Semantic Role Labeling (SRL) to create a compact representation of each news paragraph. Based on SRLP, we further incorporate other stock factors to make the final prediction. In addition, we propose a self-supervised learning strategy based on SRLP to enhance the out-of-distribution generalization performance of our system. Through our experimental study, we show that the proposed method achieves better performance and outperforms all the baselines' annualized rate of return as well as the maximum drawdown of the CSI300 index and XIN9 index on real trading. Our Astock dataset and code are available at https://github.com/JinanZou/Astock.

arXiv:2206.06593

On Finite-Sample Identifiability of Contrastive Learning-Based Nonlinear Independent Component Analysis

Nonlinear independent component analysis (nICA) aims at recovering statistically independent latent components that are mixed by unknown nonlinear functions. Central to nICA is the identifiability of the latent components, which had been elusive until very recently. Specifically, Hyv\"arinen et al. have shown that the nonlinearly mixed latent components are identifiable (up to often inconsequential ambiguities) under a generalized contrastive learning (GCL) formulation, given that the latent components are independent conditioned on a certain auxiliary variable. The GCL-based identifiability of nICA is elegant, and establishes interesting connections between nICA and popular unsupervised/self-supervised learning paradigms in representation learning, causal learning, and factor disentanglement. However, existing identifiability analyses of nICA all build upon an unlimited sample assumption and the use of ideal universal function learners -- which creates a non-negligible gap between theory and practice. Closing the gap is a nontrivial challenge, as there is a lack of established ``textbook'' routine for finite sample analysis of such unsupervised problems. This work puts forth a finite-sample identifiability analysis of GCL-based nICA. Our analytical framework judiciously combines the properties of the GCL loss function, statistical generalization analysis, and numerical differentiation. Our framework also takes the learning function's approximation error into consideration, and reveals an intuitive trade-off between the complexity and expressiveness of the employed function learner. Numerical experiments are used to validate the theorems.

arXiv:2206.06469

Invariant Structure Learning for Better Generalization and Causal Explainability

Learning the causal structure behind data is invaluable for improving generalization and obtaining high-quality explanations. We propose a novel framework, Invariant Structure Learning (ISL), that is designed to improve causal structure discovery by utilizing generalization as an indication. ISL splits the data into different environments, and learns a structure that is invariant to the target across different environments by imposing a consistency constraint. An aggregation mechanism then selects the optimal classifier based on a graph structure that reflects the causal mechanisms in the data more accurately compared to the structures learnt from individual environments. Furthermore, we extend ISL to a self-supervised learning setting where accurate causal structure discovery does not rely on any labels. This self-supervised ISL utilizes invariant causality proposals by iteratively setting different nodes as targets. On synthetic and real-world datasets, we demonstrate that ISL accurately discovers the causal structure, outperforms alternative methods, and yields superior generalization for datasets with significant distribution shifts.

arXiv:2206.06757공식 코드 · Papers with Code

RoSGAS: Adaptive Social Bot Detection with Reinforced Self-Supervised GNN Architecture Search

Social bots are referred to as the automated accounts on social networks that make attempts to behave like human. While Graph Neural Networks (GNNs) has been massively applied to the field of social bot detection, a huge amount of domain expertise and prior knowledge is heavily engaged in the state-of-the art approaches to design a dedicated neural network architecture for a specific classification task. Involving oversized nodes and network layers in the model design, however, usually causes the over-smoothing problem and the lack of embedding discrimination. In this paper, we propose RoSGAS, a novel Reinforced and Self-supervised GNN Architecture Search framework to adaptively pinpoint the most suitable multi-hop neighborhood and the number of layers in the GNN architecture. More specifically, we consider the social bot detection problem as a user-centric subgraph embedding and classification task. We exploit heterogeneous information network to present the user connectivity by leveraging account metadata, relationships, behavioral features and content features. RoSGAS uses a multi-agent deep reinforcement learning (RL) mechanism for navigating the search of optimal neighborhood and network layers to learn individually the subgraph embedding for each target user. A nearest neighbor mechanism is developed for accelerating the RL training process, and RoSGAS can learn more discriminative subgraph embedding with the aid of self-supervised learning. Experiments on 5 Twitter datasets show that RoSGAS outperforms the state-of-the-art approaches in terms of accuracy, training efficiency and stability, and has better generalization when handling unseen samples.

arXiv:2206.06663공식 코드 · Papers with Code

Quantitative Imaging Principles Improves Medical Image Learning

Fundamental differences between natural and medical images have recently favored the use of self-supervised learning (SSL) over ImageNet transfer learning for medical image applications. Differences between image types are primarily due to the imaging modality and medical images utilize a wide range of physics based techniques while natural images are captured using only visible light. While many have demonstrated that SSL on medical images has resulted in better downstream task performance, our work suggests that more performance can be gained. The scientific principles which are used to acquire medical images are not often considered when constructing learning problems. For this reason, we propose incorporating quantitative imaging principles during generative SSL to improve image quality and quantitative biological accuracy. We show that this training schema results in better starting states for downstream supervised training on limited data. Our model also generates images that validate on clinical quantitative analysis software.

arXiv:2206.06490

Learning Task-Independent Game State Representations from Unlabeled Images

Self-supervised learning (SSL) techniques have been widely used to learn compact and informative representations from high-dimensional complex data. In many computer vision tasks, such as image classification, such methods achieve state-of-the-art results that surpass supervised learning approaches. In this paper, we investigate whether SSL methods can be leveraged for the task of learning accurate state representations of games, and if so, to what extent. For this purpose, we collect game footage frames and corresponding sequences of games' internal state from three different 3D games: VizDoom, the CARLA racing simulator and the Google Research Football Environment. We train an image encoder with three widely used SSL algorithms using solely the raw frames, and then attempt to recover the internal state variables from the learned representations. Our results across all three games showcase significantly higher correlation between SSL representations and the game's internal state compared to pre-trained baseline models such as ImageNet. Such findings suggest that SSL-based visual encoders can yield general -- not tailored to a specific task -- yet informative game representations solely from game pixel information. Such representations can, in turn, form the basis for boosting the performance of downstream learning tasks in games, including gameplaying, content generation and player modeling.

arXiv:2206.06959

AuxMix: Semi-Supervised Learning with Unconstrained Unlabeled Data

Semi-supervised learning (SSL) has seen great strides when labeled data is scarce but unlabeled data is abundant. Critically, most recent work assume that such unlabeled data is drawn from the same distribution as the labeled data. In this work, we show that state-of-the-art SSL algorithms suffer a degradation in performance in the presence of unlabeled auxiliary data that does not necessarily possess the same class distribution as the labeled set. We term this problem as Auxiliary-SSL and propose AuxMix, an algorithm that leverages self-supervised learning tasks to learn generic features in order to mask auxiliary data that are not semantically similar to the labeled set. We also propose to regularize learning by maximizing the predicted entropy for dissimilar auxiliary samples. We show an improvement of 5% over existing baselines on a ResNet-50 model when trained on CIFAR10 dataset with 4k labeled samples and all unlabeled data is drawn from the Tiny-ImageNet dataset. We report competitive results on several datasets and conduct ablation studies.

arXiv:2206.05518

Investigation of Ensemble features of Self-Supervised Pretrained Models for Automatic Speech Recognition

Self-supervised learning (SSL) based models have been shown to generate powerful representations that can be used to improve the performance of downstream speech tasks. Several state-of-the-art SSL models are available, and each of these models optimizes a different loss which gives rise to the possibility of their features being complementary. This paper proposes using an ensemble of such SSL representations and models, which exploits the complementary nature of the features extracted by the various pretrained models. We hypothesize that this results in a richer feature representation and shows results for the ASR downstream task. To this end, we use three SSL models that have shown excellent results on ASR tasks, namely HuBERT, Wav2vec2.0, and WaveLM. We explore the ensemble of models fine-tuned for the ASR task and the ensemble of features using the embeddings obtained from the pre-trained models for a downstream ASR task. We get improved performance over individual models and pre-trained features using Librispeech(100h) and WSJ dataset for the downstream tasks.

arXiv:2206.05390

Transformer-based Self-Supervised Fish Segmentation in Underwater Videos

Underwater fish segmentation to estimate fish body measurements is still largely unsolved due to the complex underwater environment. Relying on fully-supervised segmentation models requires collecting per-pixel labels, which is time-consuming and prone to overfitting. Self-supervised learning methods can help avoid the requirement of large annotated training datasets, however, to be useful in real-world applications, they should achieve good segmentation quality. In this paper, we introduce a Transformer-based method that uses self-supervision for high-quality fish segmentation. Our proposed model is trained on videos -- without any annotations -- to perform fish segmentation in underwater videos taken in situ in the wild. We show that when trained on a set of underwater videos from one dataset, the proposed model surpasses previous CNN-based and Transformer-based self-supervised methods and achieves performance relatively close to supervised methods on two new unseen underwater video datasets. This demonstrates the great generalisability of our model and the fact that it does not need a pre-trained model. In addition, we show that, due to its dense representation learning, our model is compute-efficient. We provide quantitative and qualitative results that demonstrate our model's significant capabilities.

arXiv:2206.05982

Learning Fashion Compatibility from In-the-wild Images

Complementary fashion recommendation aims at identifying items from different categories (e.g. shirt, footwear, etc.) that "go well together" as an outfit. Most existing approaches learn representation for this task using labeled outfit datasets containing manually curated compatible item combinations. In this work, we propose to learn representations for compatibility prediction from in-the-wild street fashion images through self-supervised learning by leveraging the fact that people often wear compatible outfits. Our pretext task is formulated such that the representations of different items worn by the same person are closer compared to those worn by other people. Additionally, to reduce the domain gap between in-the-wild and catalog images during inference, we introduce an adversarial loss that minimizes the difference in feature distribution between the two domains. We conduct our experiments on two popular fashion compatibility benchmarks - Polyvore and Polyvore-Disjoint outfits, and outperform existing self-supervised approaches, particularly significant in cross-dataset setting where training and testing images are from different sources.

arXiv:2206.06023

Virtual embeddings and self-consistency for self-supervised learning

Self-supervised Learning (SSL) has recently gained much attention due to the high cost and data limitation in the training of supervised learning models. The current paradigm in the SSL is to utilize data augmentation at the input space to create different views of the same images and train a model to maximize the representations between similar images and minimize them for different ones. While this approach achieves state-of-the-art (SOTA) results in various downstream tasks, it still lakes the opportunity to investigate the latent space augmentation. This paper proposes TriMix, a novel concept for SSL that generates virtual embeddings through linear interpolation of the data, thus providing the model with novel representations. Our strategy focuses on training the model to extract the original embeddings from virtual ones, hence, better representation learning. Additionally, we propose a self-consistency term that improves the consistency between the virtual and actual embeddings. We validate TriMix on eight benchmark datasets consisting of natural and medical images with an improvement of 2.71% and 0.41% better than the second-best models for both data types. Further, our approach outperformed the current methods in semi-supervised learning, particularly in low data regimes. Besides, our pre-trained models showed better transfer to other datasets.

arXiv:2206.05184공식 코드 · Papers with Code

SERE: Exploring Feature Self-relation for Self-supervised Transformer

Learning representations with self-supervision for convolutional networks (CNN) has been validated to be effective for vision tasks. As an alternative to CNN, vision transformers (ViT) have strong representation ability with spatial self-attention and channel-level feedforward networks. Recent works reveal that self-supervised learning helps unleash the great potential of ViT. Still, most works follow self-supervised strategies designed for CNN, e.g., instance-level discrimination of samples, but they ignore the properties of ViT. We observe that relational modeling on spatial and channel dimensions distinguishes ViT from other networks. To enforce this property, we explore the feature SElf-RElation (SERE) for training self-supervised ViT. Specifically, instead of conducting self-supervised learning solely on feature embeddings from multiple views, we utilize the feature self-relations, i.e., spatial/channel self-relations, for self-supervised learning. Self-relation based learning further enhances the relation modeling ability of ViT, resulting in stronger representations that stably improve performance on multiple downstream tasks. Our source code is publicly available at: https://github.com/MCG-NKU/SERE.

arXiv:2206.05093

Federated Momentum Contrastive Clustering

We present federated momentum contrastive clustering (FedMCC), a learning framework that can not only extract discriminative representations over distributed local data but also perform data clustering. In FedMCC, a transformed data pair passes through both the online and target networks, resulting in four representations over which the losses are determined. The resulting high-quality representations generated by FedMCC can outperform several existing self-supervised learning methods for linear evaluation and semi-supervised learning tasks. FedMCC can easily be adapted to ordinary centralized clustering through what we call momentum contrastive clustering (MCC). We show that MCC achieves state-of-the-art clustering accuracy results in certain datasets such as STL-10 and ImageNet-10. We also present a method to reduce the memory footprint of our clustering schemes.

arXiv:2206.04846공식 코드 · Papers with Code

Masked Autoencoders are Robust Data Augmentors

Deep neural networks are capable of learning powerful representations to tackle complex vision tasks but expose undesirable properties like the over-fitting issue. To this end, regularization techniques like image augmentation are necessary for deep neural networks to generalize well. Nevertheless, most prevalent image augmentation recipes confine themselves to off-the-shelf linear transformations like scale, flip, and colorjitter. Due to their hand-crafted property, these augmentations are insufficient to generate truly hard augmented examples. In this paper, we propose a novel perspective of augmentation to regularize the training process. Inspired by the recent success of applying masked image modeling to self-supervised learning, we adopt the self-supervised masked autoencoder to generate the distorted view of the input images. We show that utilizing such model-based nonlinear transformation as data augmentation can improve high-level recognition tasks. We term the proposed method as \textbf{M}ask-\textbf{R}econstruct \textbf{A}ugmentation (MRA). The extensive experiments on various image classification benchmarks verify the effectiveness of the proposed augmentation. Specifically, MRA consistently enhances the performance on supervised, semi-supervised as well as few-shot classification.

arXiv:2206.05266공식 코드 · Papers with Code

Does Self-supervised Learning Really Improve Reinforcement Learning from Pixels?

We investigate whether self-supervised learning (SSL) can improve online reinforcement learning (RL) from pixels. We extend the contrastive reinforcement learning framework (e.g., CURL) that jointly optimizes SSL and RL losses and conduct an extensive amount of experiments with various self-supervised losses. Our observations suggest that the existing SSL framework for RL fails to bring meaningful improvement over the baselines only taking advantage of image augmentation when the same amount of data and augmentation is used. We further perform evolutionary searches to find the optimal combination of multiple self-supervised losses for RL, but find that even such a loss combination fails to meaningfully outperform the methods that only utilize carefully designed image augmentations. After evaluating these approaches together in multiple different environments including a real-world robot environment, we confirm that no single self-supervised loss or image augmentation method can dominate all environments and that the current framework for joint optimization of SSL and RL is limited. Finally, we conduct the ablation study on multiple factors and demonstrate the properties of representations learned with different approaches.