PyTorchKR Papers

최신 논문

총 5,360편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2306.04186공식 코드 · Papers with Code

Self-supervised Audio Teacher-Student Transformer for Both Clip-level and Frame-level Tasks

Self-supervised learning (SSL) has emerged as a popular approach for learning audio representations. One goal of audio self-supervised pre-training is to transfer knowledge to downstream audio tasks, generally including clip-level and frame-level tasks. While frame-level tasks are important for fine-grained acoustic scene/event understanding, prior studies primarily evaluate on clip-level downstream tasks. In order to tackle both clip-level and frame-level tasks, this paper proposes Audio Teacher-Student Transformer (ATST), with a clip-level version (named ATST-Clip) and a frame-level version (named ATST-Frame), responsible for learning clip-level and frame-level representations, respectively. Both methods use a Transformer encoder and a teacher-student training scheme. We have carefully designed the view creation strategy for ATST-Clip and ATST-Frame. Specifically, ATST-Clip uses segment-wise data augmentations, and ATST-Frame integrates frame-wise data augmentations and masking. Experimental results show that our ATST-Frame model obtains state-of-the-art (SOTA) performances on most of the clip-level and frame-level downstream tasks. Especially, it outperforms other models by a large margin on the frame-level sound event detection task. In addition, the performance can be further improved by combining the two models through knowledge distillation. Our code is available online.

arXiv:2306.04099

NTKCPL: Active Learning on Top of Self-Supervised Model by Estimating True Coverage

High annotation cost for training machine learning classifiers has driven extensive research in active learning and self-supervised learning. Recent research has shown that in the context of supervised learning different active learning strategies need to be applied at various stages of the training process to ensure improved performance over the random baseline. We refer to the point where the number of available annotations changes the suitable active learning strategy as the phase transition point. In this paper, we establish that when combining active learning with self-supervised models to achieve improved performance, the phase transition point occurs earlier. It becomes challenging to determine which strategy should be used for previously unseen datasets. We argue that existing active learning algorithms are heavily influenced by the phase transition because the empirical risk over the entire active learning pool estimated by these algorithms is inaccurate and influenced by the number of labeled samples. To address this issue, we propose a novel active learning strategy, neural tangent kernel clustering-pseudo-labels (NTKCPL). It estimates empirical risk based on pseudo-labels and the model prediction with NTK approximation. We analyze the factors affecting this approximation error and design a pseudo-label clustering generation method to reduce the approximation error. We validate our method on five datasets, empirically demonstrating that it outperforms the baseline methods in most cases and is valid over a wider range of training budgets.

arXiv:2306.04539공식 코드 · Papers with Code

Multimodal Learning Without Labeled Multimodal Data: Guarantees and Applications

In many machine learning systems that jointly learn from multiple modalities, a core research question is to understand the nature of multimodal interactions: how modalities combine to provide new task-relevant information that was not present in either alone. We study this challenge of interaction quantification in a semi-supervised setting with only labeled unimodal data and naturally co-occurring multimodal data (e.g., unlabeled images and captions, video and corresponding audio) but when labeling them is time-consuming. Using a precise information-theoretic definition of interactions, our key contribution is the derivation of lower and upper bounds to quantify the amount of multimodal interactions in this semi-supervised setting. We propose two lower bounds: one based on the shared information between modalities and the other based on disagreement between separately trained unimodal classifiers, and derive an upper bound through connections to approximate algorithms for min-entropy couplings. We validate these estimated bounds and show how they accurately track true interactions. Finally, we show how these theoretical results can be used to estimate multimodal model performance, guide data collection, and select appropriate multimodal models for various tasks.

arXiv:2306.04337

A study on the impact of Self-Supervised Learning on automatic dysarthric speech assessment

Automating dysarthria assessments offers the opportunity to develop practical, low-cost tools that address the current limitations of manual and subjective assessments. Nonetheless, the small size of most dysarthria datasets makes it challenging to develop automated assessment. Recent research showed that speech representations from models pre-trained on large unlabelled data can enhance Automatic Speech Recognition (ASR) performance for dysarthric speech. We are the first to evaluate the representations from pre-trained state-of-the-art Self-Supervised models across three downstream tasks on dysarthric speech: disease classification, word recognition and intelligibility classification, and under three noise scenarios on the UA-Speech dataset. We show that HuBERT is the most versatile feature extractor across dysarthria classification, word recognition, and intelligibility classification, achieving respectively $+24.7\%, +61\%, \text{and} +7.2\%$ accuracy compared to classical acoustic features.

arXiv:2306.03789

On the Robustness of Arabic Speech Dialect Identification

Arabic dialect identification (ADI) tools are an important part of the large-scale data collection pipelines necessary for training speech recognition models. As these pipelines require application of ADI tools to potentially out-of-domain data, we aim to investigate how vulnerable the tools may be to this domain shift. With self-supervised learning (SSL) models as a starting point, we evaluate transfer learning and direct classification from SSL features. We undertake our evaluation under rich conditions, with a goal to develop ADI systems from pretrained models and ultimately evaluate performance on newly collected data. In order to understand what factors contribute to model decisions, we carry out a careful human study of a subset of our data. Our analysis confirms that domain shift is a major challenge for ADI models. We also find that while self-training does alleviate this challenges, it may be insufficient for realistic conditions.

arXiv:2306.03773

Some voices are too common: Building fair speech recognition systems using the Common Voice dataset

Automatic speech recognition (ASR) systems become increasingly efficient thanks to new advances in neural network training like self-supervised learning. However, they are known to be unfair toward certain groups, for instance, people speaking with an accent. In this work, we use the French Common Voice dataset to quantify the biases of a pre-trained wav2vec~2.0 model toward several demographic groups. By fine-tuning the pre-trained model on a variety of fixed-size, carefully crafted training sets, we demonstrate the importance of speaker diversity. We also run an in-depth analysis of the Common Voice corpus and identify important shortcomings that should be taken into account by users of this dataset.

arXiv:2306.03494

Structurally Different Neural Network Blocks for the Segmentation of Atrial and Aortic Perivascular Adipose Tissue in Multi-centre CT Angiography Scans

Since the emergence of convolutional neural networks (CNNs) and, later, vision transformers (ViTs), deep learning architectures have predominantly relied on identical block types with varying hyperparameters. We propose a novel block alternation strategy to leverage the complementary strengths of different architectural designs, assembling structurally distinct components similar to Lego blocks. We introduce LegoNet, a deep learning framework that alternates CNN-based and SwinViT-based blocks to enhance feature learning for medical image segmentation. We investigate three variations of LegoNet and apply this concept to a previously unexplored clinical problem: the segmentation of the internal mammary artery (IMA), aorta, and perivascular adipose tissue (PVAT) from computed tomography angiography (CTA) scans. These PVAT regions have been shown to possess prognostic value in assessing cardiovascular risk and primary clinical outcomes. We evaluate LegoNet on large datasets, achieving superior performance to other leading architectures. Furthermore, we assess the model's generalizability on external testing cohorts, where an expert clinician corrects the model's segmentations, achieving DSC > 0.90 across various external, international, and public cohorts. To further validate the model's clinical reliability, we perform intra- and inter-observer variability analysis, demonstrating strong agreement with human annotations. The proposed methodology has significant implications for diagnostic cardiovascular management and early prognosis, offering a robust, automated solution for vascular and perivascular segmentation and risk assessment in clinical practice, paving the way for personalised medicine.

arXiv:2306.03407

LESS: Label-efficient Multi-scale Learning for Cytological Whole Slide Image Screening

In computational pathology, multiple instance learning (MIL) is widely used to circumvent the computational impasse in giga-pixel whole slide image (WSI) analysis. It usually consists of two stages: patch-level feature extraction and slide-level aggregation. Recently, pretrained models or self-supervised learning have been used to extract patch features, but they suffer from low effectiveness or inefficiency due to overlooking the task-specific supervision provided by slide labels. Here we propose a weakly-supervised Label-Efficient WSI Screening method, dubbed LESS, for cytological WSI analysis with only slide-level labels, which can be effectively applied to small datasets. First, we suggest using variational positive-unlabeled (VPU) learning to uncover hidden labels of both benign and malignant patches. We provide appropriate supervision by using slide-level labels to improve the learning of patch-level features. Next, we take into account the sparse and random arrangement of cells in cytological WSIs. To address this, we propose a strategy to crop patches at multiple scales and utilize a cross-attention vision transformer (CrossViT) to combine information from different scales for WSI classification. The combination of our two steps achieves task-alignment, improving effectiveness and efficiency. We validate the proposed label-efficient method on a urine cytology WSI dataset encompassing 130 samples (13,000 patches) and FNAC 2019 dataset with 212 samples (21,200 patches). The experiment shows that the proposed LESS reaches 84.79%, 85.43%, 91.79% and 78.30% on a urine cytology WSI dataset, and 96.88%, 96.86%, 98.95%, 97.06% on FNAC 2019 dataset in terms of accuracy, AUC, sensitivity and specificity. It outperforms state-of-the-art MIL methods on pathology WSIs and realizes automatic cytological WSI cancer screening.

arXiv:2306.03346공식 코드 · Papers with Code

Stabilizing Contrastive RL: Techniques for Robotic Goal Reaching from Offline Data

Robotic systems that rely primarily on self-supervised learning have the potential to decrease the amount of human annotation and engineering effort required to learn control strategies. In the same way that prior robotic systems have leveraged self-supervised techniques from computer vision (CV) and natural language processing (NLP), our work builds on prior work showing that the reinforcement learning (RL) itself can be cast as a self-supervised problem: learning to reach any goal without human-specified rewards or labels. Despite the seeming appeal, little (if any) prior work has demonstrated how self-supervised RL methods can be practically deployed on robotic systems. By first studying a challenging simulated version of this task, we discover design decisions about architectures and hyperparameters that increase the success rate by $2 \times$. These findings lay the groundwork for our main result: we demonstrate that a self-supervised RL algorithm based on contrastive learning can solve real-world, image-based robotic manipulation tasks, with tasks being specified by a single goal image provided after training.

arXiv:2306.03506

Subgraph Networks Based Contrastive Learning

Graph contrastive learning (GCL), as a self-supervised learning method, can solve the problem of annotated data scarcity. It mines explicit features in unannotated graphs to generate favorable graph representations for downstream tasks. Most existing GCL methods focus on the design of graph augmentation strategies and mutual information estimation operations. Graph augmentation produces augmented views by graph perturbations. These views preserve a locally similar structure and exploit explicit features. However, these methods have not considered the interaction existing in subgraphs. To explore the impact of substructure interactions on graph representations, we propose a novel framework called subgraph network-based contrastive learning (SGNCL). SGNCL applies a subgraph network generation strategy to produce augmented views. This strategy converts the original graph into an Edge-to-Node mapping network with both topological and attribute features. The single-shot augmented view is a first-order subgraph network that mines the interaction between nodes, node-edge, and edges. In addition, we also investigate the impact of the second-order subgraph augmentation on mining graph structure interactions, and further, propose a contrastive objective that fuses the first-order and second-order subgraph information. We compare SGNCL with classical and state-of-the-art graph contrastive learning methods on multiple benchmark datasets of different domains. Extensive experiments show that SGNCL achieves competitive or better performance (top three) on all datasets in unsupervised learning settings. Furthermore, SGNCL achieves the best average gain of 6.9\% in transfer learning compared to the best method. Finally, experiments also demonstrate that mining substructure interactions have positive implications for graph contrastive learning.

arXiv:2306.02541

OTF: Optimal Transport based Fusion of Supervised and Self-Supervised Learning Models for Automatic Speech Recognition

Self-Supervised Learning (SSL) Automatic Speech Recognition (ASR) models have shown great promise over Supervised Learning (SL) ones in low-resource settings. However, the advantages of SSL are gradually weakened when the amount of labeled data increases in many industrial applications. To further improve the ASR performance when abundant labels are available, we first explore the potential of combining SL and SSL ASR models via analyzing their complementarity in recognition accuracy and optimization property. Then, we propose a novel Optimal Transport based Fusion (OTF) method for SL and SSL models without incurring extra computation cost in inference. Specifically, optimal transport is adopted to softly align the layer-wise weights to unify the two different networks into a single one. Experimental results on the public 1k-hour English LibriSpeech dataset and our in-house 2.6k-hour Chinese dataset show that OTF largely outperforms the individual models with lower error rates.

arXiv:2306.02330공식 코드 · Papers with Code

Graph Transformer for Recommendation

This paper presents a novel approach to representation learning in recommender systems by integrating generative self-supervised learning with graph transformer architecture. We highlight the importance of high-quality data augmentation with relevant self-supervised pretext tasks for improving performance. Towards this end, we propose a new approach that automates the self-supervision augmentation process through a rationale-aware generative SSL that distills informative user-item interaction patterns. The proposed recommender with Graph TransFormer (GFormer) that offers parameterized collaborative rationale discovery for selective augmentation while preserving global-aware user-item relationships. In GFormer, we allow the rationale-aware SSL to inspire graph collaborative filtering with task-adaptive invariant rationalization in graph transformer. The experimental results reveal that our GFormer has the capability to consistently improve the performance over baselines on different datasets. Several in-depth experiments further investigate the invariant rationale-aware augmentation from various aspects. The source code for this work is publicly available at: https://github.com/HKUDS/GFormer.

arXiv:2306.02972

Simultaneous or Sequential Training? How Speech Representations Cooperate in a Multi-Task Self-Supervised Learning System

Speech representation learning with self-supervised algorithms has resulted in notable performance boosts in many downstream tasks. Recent work combined self-supervised learning (SSL) and visually grounded speech (VGS) processing mechanisms for representation learning. The joint training with SSL and VGS mechanisms provides the opportunity to utilize both unlabeled speech and speech-related visual information based on data availability. This has shown to enhance the quality of learned representations, especially at encoding semantic- and lexical-level knowledge. In this work, we further study the joint optimization of wav2vec 2.0-based SSL and transformer-based VGS as a multi-task learning system. We explore a set of training scenarios to understand how speech representations are shared or transferred between the two tasks, and what is the optimal training strategy for cross-modal semantic retrieval and phoneme discrimination performance. As a result, we find that sequential training with wav2vec 2.0 first and VGS next provides higher performance on audio-visual retrieval compared to simultaneous optimization of both learning mechanisms. However, the parallel SSL-VGS training reduces the effects of catastrophic forgetting when switching between optimization criteria. Moreover, the results suggest that phonemic representations learned through the VGS mechanism may generalize better across datasets compared to those learned with SSL.

arXiv:2306.01929

Recent Advances of Local Mechanisms in Computer Vision: A Survey and Outlook of Recent Work

Inspired by the fact that human brains can emphasize discriminative parts of the input and suppress irrelevant ones, substantial local mechanisms have been designed to boost the development of computer vision. They can not only focus on target parts to learn discriminative local representations, but also process information selectively to improve the efficiency. In terms of application scenarios and paradigms, local mechanisms have different characteristics. In this survey, we provide a systematic review of local mechanisms for various computer vision tasks and approaches, including fine-grained visual recognition, person re-identification, few-/zero-shot learning, multi-modal learning, self-supervised learning, Vision Transformers, and so on. Categorization of local mechanisms in each field is summarized. Then, advantages and disadvantages for every category are analyzed deeply, leaving room for exploration. Finally, future research directions about local mechanisms have also been discussed that may benefit future works. To the best our knowledge, this is the first survey about local mechanisms on computer vision. We hope that this survey can shed light on future research in the computer vision field.

arXiv:2306.02507

Deep learning powered real-time identification of insects using citizen science data

Insect-pests significantly impact global agricultural productivity and quality. Effective management involves identifying the full insect community, including beneficial insects and harmful pests, to develop and implement integrated pest management strategies. Automated identification of insects under real-world conditions presents several challenges, including differentiating similar-looking species, intra-species dissimilarity and inter-species similarity, several life cycle stages, camouflage, diverse imaging conditions, and variability in insect orientation. A deep-learning model, InsectNet, is proposed to address these challenges. InsectNet is endowed with five key features: (a) utilization of a large dataset of insect images collected through citizen science; (b) label-free self-supervised learning for large models; (c) improving prediction accuracy for species with a small sample size; (d) enhancing model trustworthiness; and (e) democratizing access through streamlined MLOps. This approach allows accurate identification (>96% accuracy) of over 2500 insect species, including pollinator (e.g., butterflies, bees), parasitoid (e.g., some wasps and flies), predator species (e.g., lady beetles, mantises, dragonflies) and harmful pest species (e.g., armyworms, cutworms, grasshoppers, stink bugs). InsectNet can identify invasive species, provide fine-grained insect species identification, and work effectively in challenging backgrounds. It also can abstain from making predictions when uncertain, facilitating seamless human intervention and making it a practical and trustworthy tool. InsectNet can guide citizen science data collection, especially for invasive species where early detection is crucial. Similar approaches may transform other agricultural challenges like disease detection and underscore the importance of data collection, particularly through citizen science efforts..

arXiv:2306.01623공식 코드 · Papers with Code

HomE: Homography-Equivariant Video Representation Learning

Recent advances in self-supervised representation learning have enabled more efficient and robust model performance without relying on extensive labeled data. However, most works are still focused on images, with few working on videos and even fewer on multi-view videos, where more powerful inductive biases can be leveraged for self-supervision. In this work, we propose a novel method for representation learning of multi-view videos, where we explicitly model the representation space to maintain Homography Equivariance (HomE). Our method learns an implicit mapping between different views, culminating in a representation space that maintains the homography relationship between neighboring views. We evaluate our HomE representation via action recognition and pedestrian intent prediction as downstream tasks. On action classification, our method obtains 96.4% 3-fold accuracy on the UCF101 dataset, better than most state-of-the-art self-supervised learning methods. Similarly, on the STIP dataset, we outperform the state-of-the-art by 6% for pedestrian intent prediction one second into the future while also obtaining an accuracy of 91.2% for pedestrian action (cross vs. not-cross) classification. Code is available at https://github.com/anirudhs123/HomE.

arXiv:2306.01395

Masked Autoencoder for Unsupervised Video Summarization

Summarizing a video requires a diverse understanding of the video, ranging from recognizing scenes to evaluating how much each frame is essential enough to be selected as a summary. Self-supervised learning (SSL) is acknowledged for its robustness and flexibility to multiple downstream tasks, but the video SSL has not shown its value for dense understanding tasks like video summarization. We claim an unsupervised autoencoder with sufficient self-supervised learning does not need any extra downstream architecture design or fine-tuning weights to be utilized as a video summarization model. The proposed method to evaluate the importance score of each frame takes advantage of the reconstruction score of the autoencoder's decoder. We evaluate the method in major unsupervised video summarization benchmarks to show its effectiveness under various experimental settings.

arXiv:2306.00788

Understanding Augmentation-based Self-Supervised Representation Learning via RKHS Approximation and Regression

Data augmentation is critical to the empirical success of modern self-supervised representation learning, such as contrastive learning and masked language modeling. However, a theoretical understanding of the exact role of augmentation remains limited. Recent work has built the connection between self-supervised learning and the approximation of the top eigenspace of a graph Laplacian operator, suggesting that learning a linear probe atop such representation can be connected to RKHS regression. Building on this insight, this work delves into a statistical analysis of augmentation-based pretraining. Starting from the isometry property, a geometric characterization of the target function given by the augmentation, we disentangle the effects of the model and the augmentation, and prove two generalization bounds that are free of model complexity. Our first bound works for an arbitrary encoder, where the prediction error is decomposed as the sum of an estimation error incurred by fitting a linear probe with RKHS regression, and an approximation error entailed by RKHS approximation. Our second bound specifically addresses the case where the encoder is near-optimal, that is it approximates the top-d eigenspace of the RKHS induced by the augmentation. A key ingredient in our analysis is the augmentation complexity, which we use to quantitatively compare different augmentations and analyze their impact on downstream performance.

arXiv:2306.00481

Automatic Data Augmentation for Domain Adapted Fine-Tuning of Self-Supervised Speech Representations

Self-Supervised Learning (SSL) has allowed leveraging large amounts of unlabeled speech data to improve the performance of speech recognition models even with small annotated datasets. Despite this, speech SSL representations may fail while facing an acoustic mismatch between the pretraining and target datasets. To address this issue, we propose a novel supervised domain adaptation method, designed for cases exhibiting such a mismatch in acoustic domains. It consists in applying properly calibrated data augmentations on a large clean dataset, bringing it closer to the target domain, and using it as part of an initial fine-tuning stage. Augmentations are automatically selected through the minimization of a conditional-dependence estimator, based on the target dataset. The approach is validated during an oracle experiment with controlled distortions and on two amateur-collected low-resource domains, reaching better performances compared to the baselines in both cases.

arXiv:2306.00452공식 코드 · Papers with Code

Speech Self-Supervised Representation Benchmarking: Are We Doing it Right?

Self-supervised learning (SSL) has recently allowed leveraging large datasets of unlabeled speech signals to reach impressive performance on speech tasks using only small amounts of annotated data. The high number of proposed approaches fostered the need and rise of extended benchmarks that evaluate their performance on a set of downstream tasks exploring various aspects of the speech signal. However, and while the number of considered tasks has been growing, most rely upon a single decoding architecture that maps the frozen SSL representations to the downstream labels. This work investigates the robustness of such benchmarking results to changes in the decoder architecture. Interestingly, it appears that varying the architecture of the downstream decoder leads to significant variations in the leaderboards of most tasks. Concerningly, our study reveals that benchmarking using limited decoders may cause a counterproductive increase in the sizes of the developed SSL models.