PyTorchKR Papers

최신 논문

총 8,922편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2104.12668

Appearance-based Gaze Estimation With Deep Learning: A Review and Benchmark

Human gaze provides valuable information on human focus and intentions, making it a crucial area of research. Recently, deep learning has revolutionized appearance-based gaze estimation. However, due to the unique features of gaze estimation research, such as the unfair comparison between 2D gaze positions and 3D gaze vectors and the different pre-processing and post-processing methods, there is a lack of a definitive guideline for developing deep learning-based gaze estimation algorithms. In this paper, we present a systematic review of the appearance-based gaze estimation methods using deep learning. Firstly, we survey the existing gaze estimation algorithms along the typical gaze estimation pipeline: deep feature extraction, deep learning model design, personal calibration and platforms. Secondly, to fairly compare the performance of different approaches, we summarize the data pre-processing and post-processing methods, including face/eye detection, data rectification, 2D/3D gaze conversion and gaze origin conversion. Finally, we set up a comprehensive benchmark for deep learning-based gaze estimation. We characterize all the public datasets and provide the source code of typical gaze estimation algorithms. This paper serves not only as a reference to develop deep learning-based gaze estimation methods, but also a guideline for future gaze estimation research. The project web page can be found at https://phi-ai.buaa.edu.cn/Gazehub.

arXiv:2104.12564공식 코드 · Papers with Code

Synthetic 3D Data Generation Pipeline for Geometric Deep Learning in Architecture

With the growing interest in deep learning algorithms and computational design in the architectural field, the need for large, accessible and diverse architectural datasets increases. We decided to tackle this problem by constructing a field-specific synthetic data generation pipeline that generates an arbitrary amount of 3D data along with the associated 2D and 3D annotations. The variety of annotations, the flexibility to customize the generated building and dataset parameters make this framework suitable for multiple deep learning tasks, including geometric deep learning that requires direct 3D supervision. Creating our building data generation pipeline we leveraged architectural knowledge from experts in order to construct a framework that would be modular, extendable and would provide a sufficient amount of class-balanced data samples. Moreover, we purposefully involve the researcher in the dataset customization allowing the introduction of additional building components, material textures, building classes, number and type of annotations as well as the number of views per 3D model sample. In this way, the framework would satisfy different research requirements and would be adaptable to a large variety of tasks. All code and data are made publicly available.

arXiv:2104.11892

A Survey of Modern Deep Learning based Object Detection Models

Object Detection is the task of classification and localization of objects in an image or video. It has gained prominence in recent years due to its widespread applications. This article surveys recent developments in deep learning based object detectors. Concise overview of benchmark datasets and evaluation metrics used in detection is also provided along with some of the prominent backbone architectures used in recognition tasks. It also covers contemporary lightweight classification models used on edge devices. Lastly, we compare the performances of these architectures on multiple metrics.

arXiv:2104.11556

Deep learning for sentence clustering in essay grading support

Essays as a form of assessment test student knowledge on a deeper level than short answer and multiple-choice questions. However, the manual evaluation of essays is time- and labor-consuming. Automatic clustering of essays, or their fragments, prior to manual evaluation presents a possible solution to reducing the effort required in the evaluation process. Such clustering presents numerous challenges due to the variability and ambiguity of natural language. In this paper, we introduce two datasets of undergraduate student essays in Finnish, manually annotated for salient arguments on the sentence level. Using these datasets, we evaluate several deep-learning embedding methods for their suitability to sentence clustering in support of essay grading. We find that the choice of the most suitable method depends on the nature of the exam question and the answers, with deep-learning methods being capable of, but not guaranteeing better performance over simpler methods based on lexical overlap.

arXiv:2104.11541

Deep Multi-Stage CSI Acquisition for Reconfigurable Intelligent Surface Aided MIMO Systems

This article aims to reduce huge pilot overhead when estimating the reconfigurable intelligent surface (RIS) relayed wireless channel. Motivated by the compelling grasp of deep learning in tackling nonlinear mapping problems, the proposed approach only activates a part of RIS elements and utilizes the corresponding cascaded channel estimate to predict another part. Through a synthetic deep neural network (DNN), the direct channel and active cascaded channel are first estimated sequentially, followed by the channel prediction for the inactive RIS elements. A three-stage training strategy is developed for this synthetic DNN. From simulation results, the proposed deep learning based approach is effective in reducing the pilot overhead and guaranteeing the reliable estimation accuracy.

arXiv:2104.11673공식 코드 · Papers with Code

Deep Learning Based Assessment of Synthetic Speech Naturalness

In this paper, we present a new objective prediction model for synthetic speech naturalness. It can be used to evaluate Text-To-Speech or Voice Conversion systems and works language independently. The model is trained end-to-end and based on a CNN-LSTM network that previously showed to give good results for speech quality estimation. We trained and tested the model on 16 different datasets, such as from the Blizzard Challenge and the Voice Conversion Challenge. Further, we show that the reliability of deep learning-based naturalness prediction can be improved by transfer learning from speech quality prediction models that are trained on objective POLQA scores. The proposed model is made publicly available and can, for example, be used to evaluate different TTS system configurations.

arXiv:2104.10931

Semiotic Aggregation in Deep Learning

Convolutional neural networks utilize a hierarchy of neural network layers. The statistical aspects of information concentration in successive layers can bring an insight into the feature abstraction process. We analyze the saliency maps of these layers from the perspective of semiotics, also known as the study of signs and sign-using behavior. In computational semiotics, this aggregation operation (known as superization) is accompanied by a decrease of spatial entropy: signs are aggregated into supersign. Using spatial entropy, we compute the information content of the saliency maps and study the superization processes which take place between successive layers of the network. In our experiments, we visualize the superization process and show how the obtained knowledge can be used to explain the neural decision model. In addition, we attempt to optimize the architecture of the neural model employing a semiotic greedy technique. To the extent of our knowledge, this is the first application of computational semiotics in the analysis and interpretation of deep neural networks.

arXiv:2104.11208공식 코드 · Papers with Code

Deep Video Matting via Spatio-Temporal Alignment and Aggregation

Despite the significant progress made by deep learning in natural image matting, there has been so far no representative work on deep learning for video matting due to the inherent technical challenges in reasoning temporal domain and lack of large-scale video matting datasets. In this paper, we propose a deep learning-based video matting framework which employs a novel and effective spatio-temporal feature aggregation module (ST-FAM). As optical flow estimation can be very unreliable within matting regions, ST-FAM is designed to effectively align and aggregate information across different spatial scales and temporal frames within the network decoder. To eliminate frame-by-frame trimap annotations, a lightweight interactive trimap propagation network is also introduced. The other contribution consists of a large-scale video matting dataset with groundtruth alpha mattes for quantitative evaluation and real-world high-resolution videos with trimaps for qualitative evaluation. Quantitative and qualitative experimental results show that our framework significantly outperforms conventional video matting and deep image matting methods applied to video in presence of multi-frame temporal information.

arXiv:2104.10775

Meta-learning for skin cancer detection using Deep Learning Techniques

This study focuses on automatic skin cancer detection using a Meta-learning approach for dermoscopic images. The aim of this study is to explore the benefits of the generalization of the knowledge extracted from non-medical data in the classification performance of medical data and the impact of the distribution shift problem within limited data by using a simple class and distribution balancer algorithm. In this study, a small sample of a combined dataset from 3 different sources was used to fine-tune a ResNet model pre-trained on non-medical data. The results show an increase in performance on detecting melanoma, malignant (skin cancer), and benign moles with the prior knowledge obtained from images of everyday objects from the ImageNet dataset by 20 points. These findings suggest that features from non-medical images can be used towards the classification of skin moles and that the distribution of the data affects the performance of the model.

arXiv:2104.10745공식 코드 · Papers with Code

PocketNet: A Smaller Neural Network for Medical Image Analysis

Medical imaging deep learning models are often large and complex, requiring specialized hardware to train and evaluate these models. To address such issues, we propose the PocketNet paradigm to reduce the size of deep learning models by throttling the growth of the number of channels in convolutional neural networks. We demonstrate that, for a range of segmentation and classification tasks, PocketNet architectures produce results comparable to that of conventional neural networks while reducing the number of parameters by multiple orders of magnitude, using up to 90% less GPU memory, and speeding up training times by up to 40%, thereby allowing such models to be trained and deployed in resource-constrained settings.

arXiv:2104.10584

Deep Learning for Click-Through Rate Estimation

Click-through rate (CTR) estimation plays as a core function module in various personalized online services, including online advertising, recommender systems, and web search etc. From 2015, the success of deep learning started to benefit CTR estimation performance and now deep CTR models have been widely applied in many industrial platforms. In this survey, we provide a comprehensive review of deep learning models for CTR estimation tasks. First, we take a review of the transfer from shallow to deep CTR models and explain why going deep is a necessary trend of development. Second, we concentrate on explicit feature interaction learning modules of deep CTR models. Then, as an important perspective on large platforms with abundant user histories, deep behavior models are discussed. Moreover, the recently emerged automated methods for deep CTR architecture design are presented. Finally, we summarize the survey and discuss the future prospects of this field.

arXiv:2004.02068

Ghost Handwritten Digit Recognition based on Deep Learning

We present a ghost handwritten digit recognition method for the unknown handwritten digits based on ghost imaging (GI) with deep neural network, where a few detection signals from the bucket detector, generated by the Cosine Transform speckle, are used as the characteristic information and the input of the designed deep neural network (DNN), and the classification is designed as the output of the DNN. The results show that the proposed scheme has a higher recognition accuracy (as high as 98.14% for the simulations, and 92.9% for the experiments ) with a smaller sampling ratio (say 12.76%). With the increase of the sampling ratio, the recognition accuracy is enhanced greatly. Compared with the traditional recognition scheme using the same DNN structure, the proposed scheme has a little better performance with a lower complexity and non-locality property. The proposed scheme provides a promising way for remote sensing.

arXiv:2104.09852

Adversarial Training for Deep Learning-based Intrusion Detection Systems

Nowadays, Deep Neural Networks (DNNs) report state-of-the-art results in many machine learning areas, including intrusion detection. Nevertheless, recent studies in computer vision have shown that DNNs can be vulnerable to adversarial attacks that are capable of deceiving them into misclassification by injecting specially crafted data. In security-critical areas, such attacks can cause serious damage; therefore, in this paper, we examine the effect of adversarial attacks on deep learning-based intrusion detection. In addition, we investigate the effectiveness of adversarial training as a defense against such attacks. Experimental results show that with sufficient distortion, adversarial examples are able to mislead the detector and that the use of adversarial training can improve the robustness of intrusion detection.

arXiv:2104.10047

Geometric Deep Learning on Anatomical Meshes for the Prediction of Alzheimer's Disease

Geometric deep learning can find representations that are optimal for a given task and therefore improve the performance over pre-defined representations. While current work has mainly focused on point representations, meshes also contain connectivity information and are therefore a more comprehensive characterization of the underlying anatomical surface. In this work, we evaluate four recent geometric deep learning approaches that operate on mesh representations. These approaches can be grouped into template-free and template-based approaches, where the template-based methods need a more elaborate pre-processing step with the definition of a common reference template and correspondences. We compare the different networks for the prediction of Alzheimer's disease based on the meshes of the hippocampus. Our results show advantages for template-based methods in terms of accuracy, number of learnable parameters, and training speed. While the template creation may be limiting for some applications, neuroimaging has a long history of building templates with automated tools readily available. Overall, working with meshes is more involved than working with simplistic point clouds, but they also offer new avenues for designing geometric deep learning architectures.

arXiv:2104.09839공식 코드 · Papers with Code

Deep learning with transfer functions: new applications in system identification

This paper presents a linear dynamical operator described in terms of a rational transfer function, endowed with a well-defined and efficient back-propagation behavior for automatic derivatives computation. The operator enables end-to-end training of structured networks containing linear transfer functions and other differentiable units {by} exploiting standard deep learning software. Two relevant applications of the operator in system identification are presented. The first one consists in the integration of {prediction error methods} in deep learning. The dynamical operator is included as {the} last layer of a neural network in order to obtain the optimal one-step-ahead prediction error. The second one considers identification of general block-oriented models from quantized data. These block-oriented models are constructed by combining linear dynamical operators with static nonlinearities described as standard feed-forward neural networks. A custom loss function corresponding to the log-likelihood of quantized output observations is defined. For gradient-based optimization, the derivatives of the log-likelihood are computed by applying the back-propagation algorithm through the whole network. Two system identification benchmarks are used to show the effectiveness of the proposed methodologies.

arXiv:2104.08438공식 코드 · Papers with Code

Bayesian graph convolutional neural networks via tempered MCMC

Deep learning models, such as convolutional neural networks, have long been applied to image and multi-media tasks, particularly those with structured data. More recently, there has been more attention to unstructured data that can be represented via graphs. These types of data are often found in health and medicine, social networks, and research data repositories. Graph convolutional neural networks have recently gained attention in the field of deep learning that takes advantage of graph-based data representation with automatic feature extraction via convolutions. Given the popularity of these methods in a wide range of applications, robust uncertainty quantification is vital. This remains a challenge for large models and unstructured datasets. Bayesian inference provides a principled approach to uncertainty quantification of model parameters for deep learning models. Although Bayesian inference has been used extensively elsewhere, its application to deep learning remains limited due to the computational requirements of the Markov Chain Monte Carlo (MCMC) methods. Recent advances in parallel computing and advanced proposal schemes in MCMC sampling methods has opened the path for Bayesian deep learning. In this paper, we present Bayesian graph convolutional neural networks that employ tempered MCMC sampling with Langevin-gradient proposal distribution implemented via parallel computing. Our results show that the proposed method can provide accuracy similar to advanced optimisers while providing uncertainty quantification for key benchmark problems.

arXiv:2104.08364공식 코드 · Papers with Code

Sync-Switch: Hybrid Parameter Synchronization for Distributed Deep Learning

Stochastic Gradient Descent (SGD) has become the de facto way to train deep neural networks in distributed clusters. A critical factor in determining the training throughput and model accuracy is the choice of the parameter synchronization protocol. For example, while Bulk Synchronous Parallel (BSP) often achieves better converged accuracy, the corresponding training throughput can be negatively impacted by stragglers. In contrast, Asynchronous Parallel (ASP) can have higher throughput, but its convergence and accuracy can be impacted by stale gradients. To improve the performance of synchronization protocol, recent work often focuses on designing new protocols with a heavy reliance on hard-to-tune hyper-parameters. In this paper, we design a hybrid synchronization approach that exploits the benefits of both BSP and ASP, i.e., reducing training time while simultaneously maintaining the converged accuracy. Based on extensive empirical profiling, we devise a collection of adaptive policies that determine how and when to switch between synchronization protocols. Our policies include both offline ones that target recurring jobs and online ones for handling transient stragglers. We implement the proposed policies in a prototype system, called Sync-Switch, on top of TensorFlow, and evaluate the training performance with popular deep learning models and datasets. Our experiments show that Sync-Switch achieves up to 5.13X throughput speedup and similar converged accuracy when comparing to BSP. Further, we observe that Sync-Switch achieves 3.8% higher converged accuracy with just 1.23X the training time compared to training with ASP. Moreover, Sync-Switch can be used in settings when training with ASP leads to divergence errors. Sync-Switch achieves all of these benefits with very low overhead, e.g., the framework overhead can be as low as 1.7% of the total training time.

arXiv:2104.09005

Scalable Bayesian Deep Learning with Kernel Seed Networks

This paper addresses the scalability problem of Bayesian deep neural networks. The performance of deep neural networks is undermined by the fact that these algorithms have poorly calibrated measures of uncertainty. This restricts their application in high risk domains such as computer aided diagnosis and autonomous vehicle navigation. Bayesian Deep Learning (BDL) offers a promising method for representing uncertainty in neural network. However, BDL requires a separate set of parameters to store the mean and standard deviation of model weights to learn a distribution. This results in a prohibitive 2-fold increase in the number of model parameters. To address this problem we present a method for performing BDL, namely Kernel Seed Networks (KSN), which does not require a 2-fold increase in the number of parameters. KSNs use 1x1 Convolution operations to learn a compressed latent space representation of the parameter distribution. In this paper we show how this allows KSNs to outperform conventional BDL methods while reducing the number of required parameters by up to a factor of 6.6.

arXiv:2104.09375공식 코드 · Papers with Code

A Multi-Task Deep Learning Framework for Building Footprint Segmentation

The task of building footprint segmentation has been well-studied in the context of remote sensing (RS) as it provides valuable information in many aspects, however, difficulties brought by the nature of RS images such as variations in the spatial arrangements and in-consistent constructional patterns require studying further, since it often causes poorly classified segmentation maps. We address this need by designing a joint optimization scheme for the task of building footprint delineation and introducing two auxiliary tasks; image reconstruction and building footprint boundary segmentation with the intent to reveal the common underlying structure to advance the classification accuracy of a single task model under the favor of auxiliary tasks. In particular, we propose a deep multi-task learning (MTL) based unified fully convolutional framework which operates in an end-to-end manner by making use of joint loss function with learnable loss weights considering the homoscedastic uncertainty of each task loss. Experimental results conducted on the SpaceNet6 dataset demonstrate the potential of the proposed MTL framework as it improves the classification accuracy considerably compared to single-task and lesser compounded tasks.

arXiv:2104.08894공식 코드 · Papers with Code

The Intrinsic Dimension of Images and Its Impact on Learning

It is widely believed that natural image data exhibits low-dimensional structure despite the high dimensionality of conventional pixel representations. This idea underlies a common intuition for the remarkable success of deep learning in computer vision. In this work, we apply dimension estimation tools to popular datasets and investigate the role of low-dimensional structure in deep learning. We find that common natural image datasets indeed have very low intrinsic dimension relative to the high number of pixels in the images. Additionally, we find that low dimensional datasets are easier for neural networks to learn, and models solving these tasks generalize better from training to test data. Along the way, we develop a technique for validating our dimension estimation tools on synthetic data generated by GANs allowing us to actively manipulate the intrinsic dimension by controlling the image generation process. Code for our experiments may be found here https://github.com/ppope/dimensions.