PyTorchKR Papers

최신 논문

총 10,163편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2101.12704

Federated Learning over Wireless Device-to-Device Networks: Algorithms and Convergence Analysis

The proliferation of Internet-of-Things (IoT) devices and cloud-computing applications over siloed data centers is motivating renewed interest in the collaborative training of a shared model by multiple individual clients via federated learning (FL). To improve the communication efficiency of FL implementations in wireless systems, recent works have proposed compression and dimension reduction mechanisms, along with digital and analog transmission schemes that account for channel noise, fading, and interference. The prior art has mainly focused on star topologies consisting of distributed clients and a central server. In contrast, this paper studies FL over wireless device-to-device (D2D) networks by providing theoretical insights into the performance of digital and analog implementations of decentralized stochastic gradient descent (DSGD). First, we introduce generic digital and analog wireless implementations of communication-efficient DSGD algorithms, leveraging random linear coding (RLC) for compression and over-the-air computation (AirComp) for simultaneous analog transmissions. Next, under the assumptions of convexity and connectivity, we provide convergence bounds for both implementations. The results demonstrate the dependence of the optimality gap on the connectivity and on the signal-to-noise ratio (SNR) levels in the network. The analysis is corroborated by experiments on an image-classification task.

arXiv:2101.12632

The Deep Radial Basis Function Data Descriptor (D-RBFDD) Network: A One-Class Neural Network for Anomaly Detection

Anomaly detection is a challenging problem in machine learning, and is even more so when dealing with instances that are captured in low-level, raw data representations without a well-behaved set of engineered features. The Radial Basis Function Data Descriptor (RBFDD) network is an effective solution for anomaly detection, however, it is a shallow model that does not deal effectively with raw data representations. This paper investigates approaches to modifying the RBFDD network to transform it into a deep one-class classifier suitable for anomaly detection problems with low-level raw data representations. We show that approaches based on transfer learning are not effective and our results suggest that this is because the latent representations learned by generic classification models are not suitable for anomaly detection. Instead we show that an approach that adds multiple convolutional layers before the RBF layer, to form a Deep Radial Basis Function Data Descriptor (D-RBFDD) network, is very effective. This is shown in a set of evaluation experiments using multiple anomaly detection scenarios created from publicly available image classification datasets, and a real-world anomaly detection dataset in which different types of arrhythmia are detected in electrocardiogram (ECG) data. Our experiments show that the D-RBFDD network out-performs state-of-the-art anomaly detection methods including the Deep Support Vector Data Descriptor (Deep SVDD), One-Class SVM, and Isolation Forest on the image datasets, and produces competitive results for the ECG dataset.

arXiv:2101.12491공식 코드 · Papers with Code

Efficient-CapsNet: Capsule Network with Self-Attention Routing

Deep convolutional neural networks, assisted by architectural design strategies, make extensive use of data augmentation techniques and layers with a high number of feature maps to embed object transformations. That is highly inefficient and for large datasets implies a massive redundancy of features detectors. Even though capsules networks are still in their infancy, they constitute a promising solution to extend current convolutional networks and endow artificial visual perception with a process to encode more efficiently all feature affine transformations. Indeed, a properly working capsule network should theoretically achieve higher results with a considerably lower number of parameters count due to intrinsic capability to generalize to novel viewpoints. Nevertheless, little attention has been given to this relevant aspect. In this paper, we investigate the efficiency of capsule networks and, pushing their capacity to the limits with an extreme architecture with barely 160K parameters, we prove that the proposed architecture is still able to achieve state-of-the-art results on three different datasets with only 2% of the original CapsNet parameters. Moreover, we replace dynamic routing with a novel non-iterative, highly parallelizable routing algorithm that can easily cope with a reduced number of capsules. Extensive experimentation with other capsule implementations has proved the effectiveness of our methodology and the capability of capsule networks to efficiently embed visual representations more prone to generalization.

arXiv:2101.11986공식 코드 · Papers with Code

Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet

Transformers, which are popular for language modeling, have been explored for solving vision tasks recently, e.g., the Vision Transformer (ViT) for image classification. The ViT model splits each image into a sequence of tokens with fixed length and then applies multiple Transformer layers to model their global relation for classification. However, ViT achieves inferior performance to CNNs when trained from scratch on a midsize dataset like ImageNet. We find it is because: 1) the simple tokenization of input images fails to model the important local structure such as edges and lines among neighboring pixels, leading to low training sample efficiency; 2) the redundant attention backbone design of ViT leads to limited feature richness for fixed computation budgets and limited training samples. To overcome such limitations, we propose a new Tokens-To-Token Vision Transformer (T2T-ViT), which incorporates 1) a layer-wise Tokens-to-Token (T2T) transformation to progressively structurize the image to tokens by recursively aggregating neighboring Tokens into one Token (Tokens-to-Token), such that local structure represented by surrounding tokens can be modeled and tokens length can be reduced; 2) an efficient backbone with a deep-narrow structure for vision transformer motivated by CNN architecture design after empirical study. Notably, T2T-ViT reduces the parameter count and MACs of vanilla ViT by half, while achieving more than 3.0\% improvement when trained from scratch on ImageNet. It also outperforms ResNets and achieves comparable performance with MobileNets by directly training on ImageNet. For example, T2T-ViT with comparable size to ResNet50 (21.5M parameters) can achieve 83.3\% top1 accuracy in image resolution 384$\times$384 on ImageNet. (Code: https://github.com/yitu-opensource/T2T-ViT)

arXiv:2101.11878

CORL: Compositional Representation Learning for Few-Shot Classification

Few-shot image classification consists of two consecutive learning processes: 1) In the meta-learning stage, the model acquires a knowledge base from a set of training classes. 2) During meta-testing, the acquired knowledge is used to recognize unseen classes from very few examples. Inspired by the compositional representation of objects in humans, we train a neural network architecture that explicitly represents objects as a dictionary of shared components and their spatial composition. In particular, during meta-learning, we train a knowledge base that consists of a dictionary of component representations and a dictionary of component activation maps that encode common spatial activation patterns of components. The elements of both dictionaries are shared among the training classes. During meta-testing, the representation of unseen classes is learned using the component representations and the component activation maps from the knowledge base. Finally, an attention mechanism is used to strengthen those components that are most important for each category. We demonstrate the value of our interpretable compositional learning framework for a few-shot classification using miniImageNet, tieredImageNet, CIFAR-FS, and FC100, where we achieve comparable performance.

arXiv:2101.11750

Information contraction in noisy binary neural networks and its implications

Neural networks have gained importance as the machine learning models that achieve state-of-the-art performance on large-scale image classification, object detection and natural language processing tasks. In this paper, we consider noisy binary neural networks, where each neuron has a non-zero probability of producing an incorrect output. These noisy models may arise from biological, physical and electronic contexts and constitute an important class of models that are relevant to the physical world. Intuitively, the number of neurons in such systems has to grow to compensate for the noise while maintaining the same level of expressive power and computation reliability. Our key finding is a lower bound for the required number of neurons in noisy neural networks, which is first of its kind. To prove this lower bound, we take an information theoretic approach and obtain a novel strong data processing inequality (SDPI), which not only generalizes the Evans-Schulman results for binary symmetric channels to general channels, but also improves the tightness drastically when applied to estimate end-to-end information contraction in networks. Our SDPI can be applied to various information processing systems, including neural networks and cellular automata. Applying the SDPI in noisy binary neural networks, we obtain our key lower bound and investigate its implications on network depth-width trade-offs, our results suggest a depth-width trade-off for noisy neural networks that is very different from the established understanding regarding noiseless neural networks. Furthermore, we apply the SDPI to study fault-tolerant cellular automata and obtain bounds on the error correction overheads and the relaxation time. This paper offers new understanding of noisy information processing systems through the lens of information theory.

arXiv:2101.11685

CNN with large memory layers

This work is centred around the recently proposed product key memory structure \cite{large_memory}, implemented for a number of computer vision applications. The memory structure can be regarded as a simple computation primitive suitable to be augmented to nearly all neural network architectures. The memory block allows implementing sparse access to memory with square root complexity scaling with respect to the memory capacity. The latter scaling is possible due to the incorporation of Cartesian product space decomposition of the key space for the nearest neighbour search. We have tested the memory layer on the classification, image reconstruction and relocalization problems and found that for some of those, the memory layers can provide significant speed/accuracy improvement with the high utilization of the key-value elements, while others require more careful fine-tuning and suffer from dying keys. To tackle the later problem we have introduced a simple technique of memory re-initialization which helps us to eliminate unused key-value pairs from the memory and engage them in training again. We have conducted various experiments and got improvements in speed and accuracy for classification and PoseNet relocalization models. We showed that the re-initialization has a huge impact on a toy example of randomly labeled data and observed some gains in performance on the image classification task. We have also demonstrated the generalization property perseverance of the large memory layers on the relocalization problem, while observing the spatial correlations between the images and the selected memory cells.

arXiv:2101.11461

Machine learning with limited data

Thanks to the availability of powerful computing resources, big data and deep learning algorithms, we have made great progress on computer vision in the last few years. Computer vision systems begin to surpass humans in some tasks, such as object recognition, object detection, face recognition and pose estimation. Lots of computer vision algorithms have been deployed to real world applications and started to improve our life quality. However, big data and labels are not always available. Sometimes we only have very limited labeled data, such as medical images which requires experts to label them. In this paper, we study few shot image classification, in which we only have very few labeled data. Machine learning with little data is a big challenge. To tackle this challenge, we propose two methods and test their effectiveness thoroughly. One method is to augment image features by mixing the style of these images. The second method is applying spatial attention to explore the relations between patches of images. We also find that domain shift is a critical issue in few shot learning when the training domain and testing domain are different. So we propose a more realistic cross-domain few-shot learning with unlabeled data setting, in which some unlabeled data is available in the target domain. We propose two methods in this setting. Our first method transfers the style information of the unlabeled target dataset to the samples in the source dataset and trains a model with stylized images and original images. Our second method proposes a unified framework to fully utilize all the data. Both of our methods surpass the baseline method by a large margin.

arXiv:2101.11221

Learning task-agnostic representation via toddler-inspired learning

One of the inherent limitations of current AI systems, stemming from the passive learning mechanisms (e.g., supervised learning), is that they perform well on labeled datasets but cannot deduce knowledge on their own. To tackle this problem, we derive inspiration from a highly intentional learning system via action: the toddler. Inspired by the toddler's learning procedure, we design an interactive agent that can learn and store task-agnostic visual representation while exploring and interacting with objects in the virtual environment. Experimental results show that such obtained representation was expandable to various vision tasks such as image classification, object localization, and distance estimation tasks. In specific, the proposed model achieved 100%, 75.1% accuracy and 1.62% relative error, respectively, which is noticeably better than autoencoder-based model (99.7%, 66.1%, 1.95%), and also comparable with those of supervised models (100%, 87.3%, 0.71%).

arXiv:2101.11606공식 코드 · Papers with Code

Generative Multi-Label Zero-Shot Learning

Multi-label zero-shot learning strives to classify images into multiple unseen categories for which no data is available during training. The test samples can additionally contain seen categories in the generalized variant. Existing approaches rely on learning either shared or label-specific attention from the seen classes. Nevertheless, computing reliable attention maps for unseen classes during inference in a multi-label setting is still a challenge. In contrast, state-of-the-art single-label generative adversarial network (GAN) based approaches learn to directly synthesize the class-specific visual features from the corresponding class attribute embeddings. However, synthesizing multi-label features from GANs is still unexplored in the context of zero-shot setting. In this work, we introduce different fusion approaches at the attribute-level, feature-level and cross-level (across attribute and feature-levels) for synthesizing multi-label features from their corresponding multi-label class embedding. To the best of our knowledge, our work is the first to tackle the problem of multi-label feature synthesis in the (generalized) zero-shot setting. Comprehensive experiments are performed on three zero-shot image classification benchmarks: NUS-WIDE, Open Images and MS COCO. Our cross-level fusion-based generative approach outperforms the state-of-the-art on all three datasets. Furthermore, we show the generalization capabilities of our fusion approach in the zero-shot detection task on MS COCO, achieving favorable performance against existing methods. The source code is available at https://github.com/akshitac8/Generative_MLZSL.

arXiv:2101.11605

Bottleneck Transformers for Visual Recognition

We present BoTNet, a conceptually simple yet powerful backbone architecture that incorporates self-attention for multiple computer vision tasks including image classification, object detection and instance segmentation. By just replacing the spatial convolutions with global self-attention in the final three bottleneck blocks of a ResNet and no other changes, our approach improves upon the baselines significantly on instance segmentation and object detection while also reducing the parameters, with minimal overhead in latency. Through the design of BoTNet, we also point out how ResNet bottleneck blocks with self-attention can be viewed as Transformer blocks. Without any bells and whistles, BoTNet achieves 44.4% Mask AP and 49.7% Box AP on the COCO Instance Segmentation benchmark using the Mask R-CNN framework; surpassing the previous best published single model and single scale results of ResNeSt evaluated on the COCO validation set. Finally, we present a simple adaptation of the BoTNet design for image classification, resulting in models that achieve a strong performance of 84.7% top-1 accuracy on the ImageNet benchmark while being up to 1.64x faster in compute time than the popular EfficientNet models on TPU-v3 hardware. We hope our simple and effective approach will serve as a strong baseline for future research in self-attention models for vision

arXiv:2101.11453공식 코드 · Papers with Code

Meta Adversarial Training against Universal Patches

Recently demonstrated physical-world adversarial attacks have exposed vulnerabilities in perception systems that pose severe risks for safety-critical applications such as autonomous driving. These attacks place adversarial artifacts in the physical world that indirectly cause the addition of a universal patch to inputs of a model that can fool it in a variety of contexts. Adversarial training is the most effective defense against image-dependent adversarial attacks. However, tailoring adversarial training to universal patches is computationally expensive since the optimal universal patch depends on the model weights which change during training. We propose meta adversarial training (MAT), a novel combination of adversarial training with meta-learning, which overcomes this challenge by meta-learning universal patches along with model training. MAT requires little extra computation while continuously adapting a large set of patches to the current model. MAT considerably increases robustness against universal patch attacks on image classification and traffic-light detection.

arXiv:2101.10657공식 코드 · Papers with Code

Advantages and Bottlenecks of Quantum Machine Learning for Remote Sensing

This concept paper aims to provide a brief outline of quantum computers, explore existing methods of quantum image classification techniques, so focusing on remote sensing applications, and discuss the bottlenecks of performing these algorithms on currently available open source platforms. Initial results demonstrate feasibility. Next steps include expanding the size of the quantum hidden layer and increasing the variety of output image options.

arXiv:2101.10578공식 코드 · Papers with Code

Malware Detection Using Frequency Domain-Based Image Visualization and Deep Learning

We propose a novel method to detect and visualize malware through image classification. The executable binaries are represented as grayscale images obtained from the count of N-grams (N=2) of bytes in the Discrete Cosine Transform (DCT) domain and a neural network is trained for malware detection. A shallow neural network is trained for classification, and its accuracy is compared with deep-network architectures such as ResNet that are trained using transfer learning. Neither dis-assembly nor behavioral analysis of malware is required for these methods. Motivated by the visual similarity of these images for different malware families, we compare our deep neural network models with standard image features like GIST descriptors to evaluate the performance. A joint feature measure is proposed to combine different features using error analysis to get an accurate ensemble model for improved classification performance. A new dataset called MaleX which contains around 1 million malware and benign Windows executable samples is created for large-scale malware detection and classification experiments. Experimental results are quite promising with 96% binary classification accuracy on MaleX. The proposed model is also able to generalize well on larger unseen malware samples and the results compare favorably with state-of-the-art static analysis-based malware detection algorithms.

arXiv:2101.10423공식 코드 · Papers with Code

Online Continual Learning in Image Classification: An Empirical Survey

Online continual learning for image classification studies the problem of learning to classify images from an online stream of data and tasks, where tasks may include new classes (class incremental) or data nonstationarity (domain incremental). One of the key challenges of continual learning is to avoid catastrophic forgetting (CF), i.e., forgetting old tasks in the presence of more recent tasks. Over the past few years, many methods and tricks have been introduced to address this problem, but many have not been fairly and systematically compared under a variety of realistic and practical settings. To better understand the relative advantages of various approaches and the settings where they work best, this survey aims to (1) compare state-of-the-art methods such as MIR, iCARL, and GDumb and determine which works best at different experimental settings; (2) determine if the best class incremental methods are also competitive in domain incremental setting; (3) evaluate the performance of 7 simple but effective trick such as "review" trick and nearest class mean (NCM) classifier to assess their relative impact. Regarding (1), we observe iCaRL remains competitive when the memory buffer is small; GDumb outperforms many recently proposed methods in medium-size datasets and MIR performs the best in larger-scale datasets. For (2), we note that GDumb performs quite poorly while MIR -- already competitive for (1) -- is also strongly competitive in this very different but important setting. Overall, this allows us to conclude that MIR is overall a strong and versatile method across a wide variety of settings. For (3), we find that all 7 tricks are beneficial, and when augmented with the "review" trick and NCM classifier, MIR produces performance levels that bring online continual learning much closer to its ultimate goal of matching offline training.

arXiv:2101.10845공식 코드 · Papers with Code

Analysis and evaluation of Deep Learning based Super-Resolution algorithms to improve performance in Low-Resolution Face Recognition

Surveillance scenarios are prone to several problems since they usually involve low-resolution footage, and there is no control of how far the subjects may be from the camera in the first place. This situation is suitable for the application of upsampling (super-resolution) algorithms since they may be able to recover the discriminant properties of the subjects involved. While general super-resolution approaches were proposed to enhance image quality for human-level perception, biometrics super-resolution methods seek the best "computer perception" version of the image since their focus is on improving automatic recognition performance. Convolutional neural networks and deep learning algorithms, in general, have been applied to computer vision tasks and are now state-of-the-art for several sub-domains, including image classification, restoration, and super-resolution. However, no work has evaluated the effects that the latest proposed super-resolution methods may have upon the accuracy and face verification performance in low-resolution "in-the-wild" data. This project aimed at evaluating and adapting different deep neural network architectures for the task of face super-resolution driven by face recognition performance in real-world low-resolution images. The experimental results in a real-world surveillance and attendance datasets showed that general super-resolution architectures might enhance face verification performance of deep neural networks trained on high-resolution faces. Also, since neural networks are function approximators and can be trained based on specific objective functions, the use of a customized loss function optimized for feature extraction showed promising results for recovering discriminant features in low-resolution face images.

arXiv:2101.10532공식 코드 · Papers with Code

Hyperspectral Image Classification: Artifacts of Dimension Reduction on Hybrid CNN

Convolutional Neural Networks (CNN) has been extensively studied for Hyperspectral Image Classification (HSIC) more specifically, 2D and 3D CNN models have proved highly efficient in exploiting the spatial and spectral information of Hyperspectral Images. However, 2D CNN only considers the spatial information and ignores the spectral information whereas 3D CNN jointly exploits spatial-spectral information at a high computational cost. Therefore, this work proposed a lightweight CNN (3D followed by 2D-CNN) model which significantly reduces the computational cost by distributing spatial-spectral feature extraction across a lighter model alongside a preprocessing that has been carried out to improve the classification results. Five benchmark Hyperspectral datasets (i.e., SalinasA, Salinas, Indian Pines, Pavia University, Pavia Center, and Botswana) are used for experimental evaluation. The experimental results show that the proposed pipeline outperformed in terms of generalization performance, statistical significance, and computational complexity, as compared to the state-of-the-art 2D/3D CNN models except commonly used computationally expensive design choices.

arXiv:2101.10452

Black-box Adversarial Attacks on Monocular Depth Estimation Using Evolutionary Multi-objective Optimization

This paper proposes an adversarial attack method to deep neural networks (DNNs) for monocular depth estimation, i.e., estimating the depth from a single image. Single image depth estimation has improved drastically in recent years due to the development of DNNs. However, vulnerabilities of DNNs for image classification have been revealed by adversarial attacks, and DNNs for monocular depth estimation could contain similar vulnerabilities. Therefore, research on vulnerabilities of DNNs for monocular depth estimation has spread rapidly, but many of them assume white-box conditions where inside information of DNNs is available, or are transferability-based black-box attacks that require a substitute DNN model and a training dataset. Utilizing Evolutionary Multi-objective Optimization, the proposed method in this paper analyzes DNNs under the black-box condition where only output depth maps are available. In addition, the proposed method does not require a substitute DNN that has a similar architecture to the target DNN nor any knowledge about training data used to train the target model. Experimental results showed that the proposed method succeeded in attacking two DNN-based methods that were trained with indoor and outdoor scenes respectively.

arXiv:2101.09336

A Comprehensive Survey on Hardware-Aware Neural Architecture Search

Neural Architecture Search (NAS) methods have been growing in popularity. These techniques have been fundamental to automate and speed up the time consuming and error-prone process of synthesizing novel Deep Learning (DL) architectures. NAS has been extensively studied in the past few years. Arguably their most significant impact has been in image classification and object detection tasks where the state of the art results have been obtained. Despite the significant success achieved to date, applying NAS to real-world problems still poses significant challenges and is not widely practical. In general, the synthesized Convolution Neural Network (CNN) architectures are too complex to be deployed in resource-limited platforms, such as IoT, mobile, and embedded systems. One solution growing in popularity is to use multi-objective optimization algorithms in the NAS search strategy by taking into account execution latency, energy consumption, memory footprint, etc. This kind of NAS, called hardware-aware NAS (HW-NAS), makes searching the most efficient architecture more complicated and opens several questions. In this survey, we provide a detailed review of existing HW-NAS research and categorize them according to four key dimensions: the search space, the search strategy, the acceleration technique, and the hardware cost estimation strategies. We further discuss the challenges and limitations of existing approaches and potential future directions. This is the first survey paper focusing on hardware-aware NAS. We hope it serves as a valuable reference for the various techniques and algorithms discussed and paves the road for future research towards hardware-aware NAS.

arXiv:2101.09895

Spatio-temporal Data Augmentation for Visual Surveillance

Visual surveillance aims to stably detect a foreground object using a continuous image acquired from a fixed camera. Recent deep learning methods based on supervised learning show superior performance compared to classical background subtraction algorithms. However, there is still a room for improvement in static foreground, dynamic background, hard shadow, illumination changes, camouflage, etc. In addition, most of the deep learning-based methods operates well on environments similar to training. If the testing environments are different from training ones, their performance degrades. As a result, additional training on those operating environments is required to ensure a good performance. Our previous work which uses spatio-temporal input data consisted of a number of past images, background images and current image showed promising results in different environments from training, although it uses a simple U-NET structure. In this paper, we propose a data augmentation technique suitable for visual surveillance for additional performance improvement using the same network used in our previous work. In deep learning, most data augmentation techniques deal with spatial-level data augmentation techniques for use in image classification and object detection. In this paper, we propose a new method of data augmentation in the spatio-temporal dimension suitable for our previous work. Two data augmentation methods of adjusting background model images and past images are proposed. Through this, it is shown that performance can be improved in difficult areas such as static foreground and ghost objects, compared to previous studies. Through quantitative and qualitative evaluation using SBI, LASIESTA, and our own dataset, we show that it gives superior performance compared to deep learning-based algorithms and background subtraction algorithms.