PyTorchKR Papers

최신 논문

총 5,358편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2112.00847

CLAWS: Contrastive Learning with hard Attention and Weak Supervision

Learning effective visual representations without human supervision is a long-standing problem in computer vision. Recent advances in self-supervised learning algorithms have utilized contrastive learning, with methods such as SimCLR, which applies a composition of augmentations to an image, and minimizes a contrastive loss between the two augmented images. In this paper, we present CLAWS, an annotation-efficient learning framework, addressing the problem of manually labeling large-scale agricultural datasets along with potential applications such as anomaly detection and plant growth analytics. CLAWS uses a network backbone inspired by SimCLR and weak supervision to investigate the effect of contrastive learning within class clusters. In addition, we inject a hard attention mask to the cropped input image before maximizing agreement between the image pairs using a contrastive loss function. This mask forces the network to focus on pertinent object features and ignore background features. We compare results between a supervised SimCLR and CLAWS using an agricultural dataset with 227,060 samples consisting of 11 different crop classes. Our experiments and extensive evaluations show that CLAWS achieves a competitive NMI score of 0.7325. Furthermore, CLAWS engenders the creation of low dimensional representations of very large datasets with minimal parameter tuning and forming well-defined clusters, which lends themselves to using efficient, transparent, and highly interpretable clustering methods such as Gaussian Mixture Models.

arXiv:2112.00804

PreViTS: Contrastive Pretraining with Video Tracking Supervision

Videos are a rich source for self-supervised learning (SSL) of visual representations due to the presence of natural temporal transformations of objects. However, current methods typically randomly sample video clips for learning, which results in an imperfect supervisory signal. In this work, we propose PreViTS, an SSL framework that utilizes an unsupervised tracking signal for selecting clips containing the same object, which helps better utilize temporal transformations of objects. PreViTS further uses the tracking signal to spatially constrain the frame regions to learn from and trains the model to locate meaningful objects by providing supervision on Grad-CAM attention maps. To evaluate our approach, we train a momentum contrastive (MoCo) encoder on VGG-Sound and Kinetics-400 datasets with PreViTS. Training with PreViTS outperforms representations learnt by contrastive strategy alone on video downstream tasks, obtaining state-of-the-art performance on action classification. PreViTS helps learn feature representations that are more robust to changes in background and context, as seen by experiments on datasets with background changes. Learning from large-scale videos with PreViTS could lead to more accurate and robust visual feature representations.

arXiv:2112.01528공식 코드 · Papers with Code

A Fast Knowledge Distillation Framework for Visual Recognition

While Knowledge Distillation (KD) has been recognized as a useful tool in many visual tasks, such as supervised classification and self-supervised representation learning, the main drawback of a vanilla KD framework is its mechanism, which consumes the majority of the computational overhead on forwarding through the giant teacher networks, making the entire learning procedure inefficient and costly. ReLabel, a recently proposed solution, suggests creating a label map for the entire image. During training, it receives the cropped region-level label by RoI aligning on a pre-generated entire label map, allowing for efficient supervision generation without having to pass through the teachers many times. However, as the KD teachers are from conventional multi-crop training, there are various mismatches between the global label-map and region-level label in this technique, resulting in performance deterioration. In this study, we present a Fast Knowledge Distillation (FKD) framework that replicates the distillation training phase and generates soft labels using the multi-crop KD approach, while training faster than ReLabel since no post-processes such as RoI align and softmax operations are used. When conducting multi-crop in the same image for data loading, our FKD is even more efficient than the traditional image classification framework. On ImageNet-1K, we obtain 79.8% with ResNet-50, outperforming ReLabel by ~1.0% while being faster. On the self-supervised learning task, we also show that FKD has an efficiency advantage. Our project page: http://zhiqiangshen.com/projects/FKD/index.html, source code and models are available at: https://github.com/szq0214/FKD.

arXiv:2112.01515공식 코드 · Papers with Code

TransFGU: A Top-down Approach to Fine-Grained Unsupervised Semantic Segmentation

Unsupervised semantic segmentation aims to obtain high-level semantic representation on low-level visual features without manual annotations. Most existing methods are bottom-up approaches that try to group pixels into regions based on their visual cues or certain predefined rules. As a result, it is difficult for these bottom-up approaches to generate fine-grained semantic segmentation when coming to complicated scenes with multiple objects and some objects sharing similar visual appearance. In contrast, we propose the first top-down unsupervised semantic segmentation framework for fine-grained segmentation in extremely complicated scenarios. Specifically, we first obtain rich high-level structured semantic concept information from large-scale vision data in a self-supervised learning manner, and use such information as a prior to discover potential semantic categories presented in target datasets. Secondly, the discovered high-level semantic categories are mapped to low-level pixel features by calculating the class activate map (CAM) with respect to certain discovered semantic representation. Lastly, the obtained CAMs serve as pseudo labels to train the segmentation module and produce the final semantic segmentation. Experimental results on multiple semantic segmentation benchmarks show that our top-down unsupervised segmentation is robust to both object-centric and scene-centric datasets under different semantic granularity levels, and outperforms all the current state-of-the-art bottom-up methods. Our code is available at \url{https://github.com/damo-cv/TransFGU}.

arXiv:2112.01511공식 코드 · Papers with Code

The Surprising Effectiveness of Representation Learning for Visual Imitation

While visual imitation learning offers one of the most effective ways of learning from visual demonstrations, generalizing from them requires either hundreds of diverse demonstrations, task specific priors, or large, hard-to-train parametric models. One reason such complexities arise is because standard visual imitation frameworks try to solve two coupled problems at once: learning a succinct but good representation from the diverse visual data, while simultaneously learning to associate the demonstrated actions with such representations. Such joint learning causes an interdependence between these two problems, which often results in needing large amounts of demonstrations for learning. To address this challenge, we instead propose to decouple representation learning from behavior learning for visual imitation. First, we learn a visual representation encoder from offline data using standard supervised and self-supervised learning methods. Once the representations are trained, we use non-parametric Locally Weighted Regression to predict the actions. We experimentally show that this simple decoupling improves the performance of visual imitation models on both offline demonstration datasets and real-robot door opening compared to prior work in visual imitation. All of our generated data, code, and robot videos are publicly available at https://jyopari.github.io/VINN/.

arXiv:2112.00319공식 코드 · Papers with Code

Object-Aware Cropping for Self-Supervised Learning

A core component of the recent success of self-supervised learning is cropping data augmentation, which selects sub-regions of an image to be used as positive views in the self-supervised loss. The underlying assumption is that randomly cropped and resized regions of a given image share information about the objects of interest, which the learned representation will capture. This assumption is mostly satisfied in datasets such as ImageNet where there is a large, centered object, which is highly likely to be present in random crops of the full image. However, in other datasets such as OpenImages or COCO, which are more representative of real world uncurated data, there are typically multiple small objects in an image. In this work, we show that self-supervised learning based on the usual random cropping performs poorly on such datasets. We propose replacing one or both of the random crops with crops obtained from an object proposal algorithm. This encourages the model to learn both object and scene level semantic representations. Using this approach, which we call object-aware cropping, results in significant improvements over scene cropping on classification and object detection benchmarks. For example, on OpenImages, our approach achieves an improvement of 8.8% mAP over random scene-level cropping using MoCo-v2 based pre-training. We also show significant improvements on COCO and PASCAL-VOC object detection and segmentation tasks over the state-of-the-art self-supervised learning approaches. Our approach is efficient, simple and general, and can be used in most existing contrastive and non-contrastive self-supervised learning frameworks.

arXiv:2111.15068

MISS: Multi-Interest Self-Supervised Learning Framework for Click-Through Rate Prediction

CTR prediction is essential for modern recommender systems. Ranging from early factorization machines to deep learning based models in recent years, existing CTR methods focus on capturing useful feature interactions or mining important behavior patterns. Despite the effectiveness, we argue that these methods suffer from the risk of label sparsity (i.e., the user-item interactions are highly sparse with respect to the feature space), label noise (i.e., the collected user-item interactions are usually noisy), and the underuse of domain knowledge (i.e., the pairwise correlations between samples). To address these challenging problems, we propose a novel Multi-Interest Self-Supervised learning (MISS) framework which enhances the feature embeddings with interest-level self-supervision signals. With the help of two novel CNN-based multi-interest extractors,self-supervision signals are discovered with full considerations of different interest representations (point-wise and union-wise), interest dependencies (short-range and long-range), and interest correlations (inter-item and intra-item). Based on that, contrastive learning losses are further applied to the augmented views of interest representations, which effectively improves the feature representation learning. Furthermore, our proposed MISS framework can be used as an plug-in component with existing CTR prediction models and further boost their performances. Extensive experiments on three large-scale datasets show that MISS significantly outperforms the state-of-the-art models, by up to 13.55% in AUC, and also enjoys good compatibility with representative deep CTR models.

arXiv:2111.15487

FROB: Few-shot ROBust Model for Classification and Out-of-Distribution Detection

Nowadays, classification and Out-of-Distribution (OoD) detection in the few-shot setting remain challenging aims due to rarity and the limited samples in the few-shot setting, and because of adversarial attacks. Accomplishing these aims is important for critical systems in safety, security, and defence. In parallel, OoD detection is challenging since deep neural network classifiers set high confidence to OoD samples away from the training data. To address such limitations, we propose the Few-shot ROBust (FROB) model for classification and few-shot OoD detection. We devise FROB for improved robustness and reliable confidence prediction for few-shot OoD detection. We generate the support boundary of the normal class distribution and combine it with few-shot Outlier Exposure (OE). We propose a self-supervised learning few-shot confidence boundary methodology based on generative and discriminative models. The contribution of FROB is the combination of the generated boundary in a self-supervised learning manner and the imposition of low confidence at this learned boundary. FROB implicitly generates strong adversarial samples on the boundary and forces samples from OoD, including our boundary, to be less confident by the classifier. FROB achieves generalization to unseen OoD with applicability to unknown, in the wild, test sets that do not correlate to the training datasets. To improve robustness, FROB redesigns OE to work even for zero-shots. By including our boundary, FROB reduces the threshold linked to the model's few-shot robustness; it maintains the OoD performance approximately independent of the number of few-shots. The few-shot robustness analysis evaluation of FROB on different sets and on One-Class Classification (OCC) data shows that FROB achieves competitive performance and outperforms benchmarks in terms of robustness to the outlier few-shot sample population and variability.

arXiv:2111.14834공식 코드 · Papers with Code

Self-supervised Autoregressive Domain Adaptation for Time Series Data

Unsupervised domain adaptation (UDA) has successfully addressed the domain shift problem for visual applications. Yet, these approaches may have limited performance for time series data due to the following reasons. First, they mainly rely on large-scale dataset (i.e., ImageNet) for the source pretraining, which is not applicable for time-series data. Second, they ignore the temporal dimension on the feature space of the source and target domains during the domain alignment step. Last, most of prior UDA methods can only align the global features without considering the fine-grained class distribution of the target domain. To address these limitations, we propose a Self-supervised Autoregressive Domain Adaptation (SLARDA) framework. In particular, we first design a self-supervised learning module that utilizes forecasting as an auxiliary task to improve the transferability of the source features. Second, we propose a novel autoregressive domain adaptation technique that incorporates temporal dependency of both source and target features during domain alignment. Finally, we develop an ensemble teacher model to align the class-wise distribution in the target domain via a confident pseudo labeling approach. Extensive experiments have been conducted on three real-world time series applications with 30 cross-domain scenarios. Results demonstrate that our proposed SLARDA method significantly outperforms the state-of-the-art approaches for time series domain adaptation.

arXiv:2111.15454공식 코드 · Papers with Code

Boosting Discriminative Visual Representation Learning with Scenario-Agnostic Mixup

Mixup is a well-known data-dependent augmentation technique for DNNs, consisting of two sub-tasks: mixup generation and classification. However, the recent dominant online training method confines mixup to supervised learning (SL), and the objective of the generation sub-task is limited to selected sample pairs instead of the whole data manifold, which might cause trivial solutions. To overcome such limitations, we comprehensively study the objective of mixup generation and propose \textbf{S}cenario-\textbf{A}gnostic \textbf{Mix}up (SAMix) for both SL and Self-supervised Learning (SSL) scenarios. Specifically, we hypothesize and verify the objective function of mixup generation as optimizing local smoothness between two mixed classes subject to global discrimination from other classes. Accordingly, we propose $\eta$-balanced mixup loss for complementary learning of the two sub-objectives. Meanwhile, a label-free generation sub-network is designed, which effectively provides non-trivial mixup samples and improves transferable abilities. Moreover, to reduce the computational cost of online training, we further introduce a pre-trained version, SAMix$^\mathcal{P}$, achieving more favorable efficiency and generalizability. Extensive experiments on nine SL and SSL benchmarks demonstrate the consistent superiority and versatility of SAMix compared with existing methods.

arXiv:2111.15340

MC-SSL0.0: Towards Multi-Concept Self-Supervised Learning

Self-supervised pretraining is the method of choice for natural language processing models and is rapidly gaining popularity in many vision tasks. Recently, self-supervised pretraining has shown to outperform supervised pretraining for many downstream vision applications, marking a milestone in the area. This superiority is attributed to the negative impact of incomplete labelling of the training images, which convey multiple concepts, but are annotated using a single dominant class label. Although Self-Supervised Learning (SSL), in principle, is free of this limitation, the choice of pretext task facilitating SSL is perpetuating this shortcoming by driving the learning process towards a single concept output. This study aims to investigate the possibility of modelling all the concepts present in an image without using labels. In this aspect the proposed SSL frame-work MC-SSL0.0 is a step towards Multi-Concept Self-Supervised Learning (MC-SSL) that goes beyond modelling single dominant label in an image to effectively utilise the information from all the concepts present in it. MC-SSL0.0 consists of two core design concepts, group masked model learning and learning of pseudo-concept for data token using a momentum encoder (teacher-student) framework. The experimental results on multi-label and multi-class image classification downstream tasks demonstrate that MC-SSL0.0 not only surpasses existing SSL methods but also outperforms supervised transfer learning. The source code will be made publicly available for community to train on bigger corpus.

arXiv:2111.15124

In-Bed Human Pose Estimation from Unseen and Privacy-Preserving Image Domains

Medical applications have benefited greatly from the rapid advancement in computer vision. Considering patient monitoring in particular, in-bed human posture estimation offers important health-related metrics with potential value in medical condition assessments. Despite great progress in this domain, it remains challenging due to substantial ambiguity during occlusions, and the lack of large corpora of manually labeled data for model training, particularly with domains such as thermal infrared imaging which are privacy-preserving, and thus of great interest. Motivated by the effectiveness of self-supervised methods in learning features directly from data, we propose a multi-modal conditional variational autoencoder (MC-VAE) capable of reconstructing features from missing modalities seen during training. This approach is used with HRNet to enable single modality inference for in-bed pose estimation. Through extensive evaluations, we demonstrate that body positions can be effectively recognized from the available modality, achieving on par results with baseline models that are highly dependent on having access to multiple modes at inference time. The proposed framework supports future research towards self-supervised learning that generates a robust model from a single source, and expects it to generalize over many unknown distributions in clinical environments.

arXiv:2111.14595

Overcoming the Domain Gap in Contrastive Learning of Neural Action Representations

A fundamental goal in neuroscience is to understand the relationship between neural activity and behavior. For example, the ability to extract behavioral intentions from neural data, or neural decoding, is critical for developing effective brain machine interfaces. Although simple linear models have been applied to this challenge, they cannot identify important non-linear relationships. Thus, a self-supervised means of identifying non-linear relationships between neural dynamics and behavior, in order to compute neural representations, remains an important open problem. To address this challenge, we generated a new multimodal dataset consisting of the spontaneous behaviors generated by fruit flies, Drosophila melanogaster -- a popular model organism in neuroscience research. The dataset includes 3D markerless motion capture data from six camera views of the animal generating spontaneous actions, as well as synchronously acquired two-photon microscope images capturing the activity of descending neuron populations that are thought to drive actions. Standard contrastive learning and unsupervised domain adaptation techniques struggle to learn neural action representations (embeddings computed from the neural data describing action labels) due to large inter-animal differences in both neural and behavioral modalities. To overcome this deficiency, we developed simple yet effective augmentations that close the inter-animal domain gap, allowing us to extract behaviorally relevant, yet domain agnostic, information from neural data. This multimodal dataset and our new set of augmentations promise to accelerate the application of self-supervised learning methods in neuroscience.

arXiv:2111.14791공식 코드 · Papers with Code

Self-Supervised Pre-Training of Swin Transformers for 3D Medical Image Analysis

Vision Transformers (ViT)s have shown great performance in self-supervised learning of global and local representations that can be transferred to downstream applications. Inspired by these results, we introduce a novel self-supervised learning framework with tailored proxy tasks for medical image analysis. Specifically, we propose: (i) a new 3D transformer-based model, dubbed Swin UNEt TRansformers (Swin UNETR), with a hierarchical encoder for self-supervised pre-training; (ii) tailored proxy tasks for learning the underlying pattern of human anatomy. We demonstrate successful pre-training of the proposed model on 5,050 publicly available computed tomography (CT) images from various body organs. The effectiveness of our approach is validated by fine-tuning the pre-trained models on the Beyond the Cranial Vault (BTCV) Segmentation Challenge with 13 abdominal organs and segmentation tasks from the Medical Segmentation Decathlon (MSD) dataset. Our model is currently the state-of-the-art (i.e. ranked 1st) on the public test leaderboards of both MSD and BTCV datasets. Code: https://monai.io/research/swin-unetr

arXiv:2111.14726공식 코드 · Papers with Code

Do Invariances in Deep Neural Networks Align with Human Perception?

An evaluation criterion for safe and trustworthy deep learning is how well the invariances captured by representations of deep neural networks (DNNs) are shared with humans. We identify challenges in measuring these invariances. Prior works used gradient-based methods to generate identically represented inputs (IRIs), ie, inputs which have identical representations (on a given layer) of a neural network, and thus capture invariances of a given network. One necessary criterion for a network's invariances to align with human perception is for its IRIs look 'similar' to humans. Prior works, however, have mixed takeaways; some argue that later layers of DNNs do not learn human-like invariances (\cite{jenelle2019metamers}) yet others seem to indicate otherwise (\cite{mahendran2014understanding}). We argue that the loss function used to generate IRIs can heavily affect takeaways about invariances of the network and is the primary reason for these conflicting findings. We propose an adversarial regularizer on the IRI generation loss that finds IRIs that make any model appear to have very little shared invariance with humans. Based on this evidence, we argue that there is scope for improving models to have human-like invariances, and further, to have meaningful comparisons between models one should use IRIs generated using the regularizer-free loss. We then conduct an in-depth investigation of how different components (eg architectures, training losses, data augmentations) of the deep learning pipeline contribute to learning models that have good alignment with humans. We find that architectures with residual connections trained using a (self-supervised) contrastive loss with $\ell_p$ ball adversarial data augmentation tend to learn invariances that are most aligned with humans. Code: \url{github.com/nvedant07/Human-NN-Alignment}.

arXiv:2111.14629

Improving Zero-shot Generalization in Offline Reinforcement Learning using Generalized Similarity Functions

Reinforcement learning (RL) agents are widely used for solving complex sequential decision making tasks, but still exhibit difficulty in generalizing to scenarios not seen during training. While prior online approaches demonstrated that using additional signals beyond the reward function can lead to better generalization capabilities in RL agents, i.e. using self-supervised learning (SSL), they struggle in the offline RL setting, i.e. learning from a static dataset. We show that performance of online algorithms for generalization in RL can be hindered in the offline setting due to poor estimation of similarity between observations. We propose a new theoretically-motivated framework called Generalized Similarity Functions (GSF), which uses contrastive learning to train an offline RL agent to aggregate observations based on the similarity of their expected future behavior, where we quantify this similarity using \emph{generalized value functions}. We show that GSF is general enough to recover existing SSL objectives while also improving zero-shot generalization performance on a complex offline RL benchmark, offline Procgen.

arXiv:2111.14585공식 코드 · Papers with Code

Similarity Contrastive Estimation for Self-Supervised Soft Contrastive Learning

Contrastive representation learning has proven to be an effective self-supervised learning method. Most successful approaches are based on Noise Contrastive Estimation (NCE) and use different views of an instance as positives that should be contrasted with other instances, called negatives, that are considered as noise. However, several instances in a dataset are drawn from the same distribution and share underlying semantic information. A good data representation should contain relations, or semantic similarity, between the instances. Contrastive learning implicitly learns relations but considering all negatives as noise harms the quality of the learned relations. To circumvent this issue, we propose a novel formulation of contrastive learning using semantic similarity between instances called Similarity Contrastive Estimation (SCE). Our training objective is a soft contrastive learning one. Instead of hard classifying positives and negatives, we estimate from one view of a batch a continuous distribution to push or pull instances based on their semantic similarities. This target similarity distribution is sharpened to eliminate noisy relations. The model predicts for each instance, from another view, the target distribution while contrasting its positive with negatives. Experimental results show that SCE is Top-1 on the ImageNet linear evaluation protocol at 100 pretraining epochs with 72.1% accuracy and is competitive with state-of-the-art algorithms by reaching 75.4% for 200 epochs with multi-crop. We also show that SCE is able to generalize to several tasks. Source code is available here: https://github.com/CEA-LIST/SCE.

arXiv:2111.13301

Simple Contrastive Representation Adversarial Learning for NLP Tasks

Self-supervised learning approach like contrastive learning is attached great attention in natural language processing. It uses pairs of training data augmentations to build a classification task for an encoder with well representation ability. However, the construction of learning pairs over contrastive learning is much harder in NLP tasks. Previous works generate word-level changes to form pairs, but small transforms may cause notable changes on the meaning of sentences as the discrete and sparse nature of natural language. In this paper, adversarial training is performed to generate challenging and harder learning adversarial examples over the embedding space of NLP as learning pairs. Using contrastive learning improves the generalization ability of adversarial training because contrastive loss can uniform the sample distribution. And at the same time, adversarial training also enhances the robustness of contrastive learning. Two novel frameworks, supervised contrastive adversarial learning (SCAL) and unsupervised SCAL (USCAL), are proposed, which yields learning pairs by utilizing the adversarial training for contrastive learning. The label-based loss of supervised tasks is exploited to generate adversarial examples while unsupervised tasks bring contrastive loss. To validate the effectiveness of the proposed framework, we employ it to Transformer-based models for natural language understanding, sentence semantic textual similarity and adversarial learning tasks. Experimental results on GLUE benchmark tasks show that our fine-tuned supervised method outperforms BERT$_{base}$ over 1.75\%. We also evaluate our unsupervised method on semantic textual similarity (STS) tasks, and our method gets 77.29\% with BERT$_{base}$. The robustness of our approach conducts state-of-the-art results under multiple adversarial datasets on NLI tasks.

arXiv:2111.12855공식 코드 · Papers with Code

Robust Equivariant Imaging: a fully unsupervised framework for learning to image from noisy and partial measurements

Deep networks provide state-of-the-art performance in multiple imaging inverse problems ranging from medical imaging to computational photography. However, most existing networks are trained with clean signals which are often hard or impossible to obtain. Equivariant imaging (EI) is a recent self-supervised learning framework that exploits the group invariance present in signal distributions to learn a reconstruction function from partial measurement data alone. While EI results are impressive, its performance degrades with increasing noise. In this paper, we propose a Robust Equivariant Imaging (REI) framework which can learn to image from noisy partial measurements alone. The proposed method uses Stein's Unbiased Risk Estimator (SURE) to obtain a fully unsupervised training loss that is robust to noise. We show that REI leads to considerable performance gains on linear and nonlinear inverse problems, thereby paving the way for robust unsupervised imaging with deep networks. Code is available at: https://github.com/edongdongchen/REI.

arXiv:2111.12958공식 코드 · Papers with Code

Self-Distilled Self-Supervised Representation Learning

State-of-the-art frameworks in self-supervised learning have recently shown that fully utilizing transformer-based models can lead to performance boost compared to conventional CNN models. Striving to maximize the mutual information of two views of an image, existing works apply a contrastive loss to the final representations. Motivated by self-distillation in the supervised regime, we further exploit this by allowing the intermediate representations to learn from the final layer via the contrastive loss. Through self-distillation, the intermediate layers are better suited for instance discrimination, making the performance of an early-exited sub-network not much degraded from that of the full network. This renders the pretext task easier also for the final layer, leading to better representations. Our method, Self-Distilled Self-Supervised Learning (SDSSL), outperforms competitive baselines (SimCLR, BYOL and MoCo v3) using ViT on various tasks and datasets. In the linear evaluation and k-NN protocol, SDSSL not only leads to superior performance in the final layers, but also in most of the lower layers. Furthermore, qualitative and quantitative analyses show how representations are formed more effectively along the transformer layers. Code is available at https://github.com/hagiss/SDSSL.