PyTorchKR Papers

최신 논문

총 5,575편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1806.10040

Crowd Counting with Density Adaption Networks

Crowd counting is one of the core tasks in various surveillance applications. A practical system involves estimating accurate head counts in dynamic scenarios under different lightning, camera perspective and occlusion states. Previous approaches estimate head counts despite that they can vary dramatically in different density settings; the crowd is often unevenly distributed and the results are therefore unsatisfactory. In this paper, we propose a lightweight deep learning framework that can automatically estimate the crowd density level and adaptively choose between different counter networks that are explicitly trained for different density domains. Experiments on two recent crowd counting datasets, UCF_CC_50 and ShanghaiTech, show that the proposed mechanism achieves promising improvements over state-of-the-art methods. Moreover, runtime speed is 20 FPS on a single GPU.

arXiv:1806.09907공식 코드 · Papers with Code

AirLab: Autograd Image Registration Laboratory

Medical image registration is an active research topic and forms a basis for many medical image analysis tasks. Although image registration is a rather general concept specialized methods are usually required to target a specific registration problem. The development and implementation of such methods has been tough so far as the gradient of the objective has to be computed. Also, its evaluation has to be performed preferably on a GPU for larger images and for more complex transformation models and regularization terms. This hinders researchers from rapid prototyping and poses hurdles to reproduce research results. There is a clear need for an environment which hides this complexity to put the modeling and the experimental exploration of registration methods into the foreground. With the "Autograd Image Registration Laboratory" (AIRLab), we introduce an open laboratory for image registration tasks, where the analytic gradients of the objective function are computed automatically and the device where the computations are performed, on a CPU or a GPU, is transparent. It is meant as a laboratory for researchers and developers enabling them to rapidly try out new ideas for registering images and to reproduce registration results which have already been published. AIRLab is implemented in Python using PyTorch as tensor and optimization library and SimpleITK for basic image IO. Therefore, it profits from recent advances made by the machine learning community concerning optimization and deep neural network models. The presented draft of this paper outlines AIRLab with first code snippets and performance analyses. A more exhaustive introduction will follow as a final version soon.

arXiv:1806.09819공식 코드 · Papers with Code

Limited Evaluation Evolutionary Optimization of Large Neural Networks

Stochastic gradient descent is the most prevalent algorithm to train neural networks. However, other approaches such as evolutionary algorithms are also applicable to this task. Evolutionary algorithms bring unique trade-offs that are worth exploring, but computational demands have so far restricted exploration to small networks with few parameters. We implement an evolutionary algorithm that executes entirely on the GPU, which allows to efficiently batch-evaluate a whole population of networks. Within this framework, we explore the limited evaluation evolutionary algorithm for neural network training and find that its batch evaluation idea comes with a large accuracy trade-off. In further experiments, we explore crossover operators and find that unprincipled random uniform crossover performs extremely well. Finally, we train a network with 92k parameters on MNIST using an EA and achieve 97.6 % test accuracy compared to 98 % test accuracy on the same network trained with Adam. Code is available at https://github.com/jprellberg/gpuea.

arXiv:1806.08593공식 코드 · Papers with Code

Tensor Monte Carlo: particle methods for the GPU era

Multi-sample, importance-weighted variational autoencoders (IWAE) give tighter bounds and more accurate uncertainty estimates than variational autoencoders (VAE) trained with a standard single-sample objective. However, IWAEs scale poorly: as the latent dimensionality grows, they require exponentially many samples to retain the benefits of importance weighting. While sequential Monte-Carlo (SMC) can address this problem, it is prohibitively slow because the resampling step imposes sequential structure which cannot be parallelised, and moreover, resampling is non-differentiable which is problematic when learning approximate posteriors. To address these issues, we developed tensor Monte-Carlo (TMC) which gives exponentially many importance samples by separately drawing $K$ samples for each of the $n$ latent variables, then averaging over all $K^n$ possible combinations. While the sum over exponentially many terms might seem to be intractable, in many cases it can be computed efficiently as a series of tensor inner-products. We show that TMC is superior to IWAE on a generative model with multiple stochastic layers trained on the MNIST handwritten digit database, and we show that TMC can be combined with standard variance reduction techniques.

arXiv:1806.08198

DPP-Net: Device-aware Progressive Search for Pareto-optimal Neural Architectures

Recent breakthroughs in Neural Architectural Search (NAS) have achieved state-of-the-art performances in applications such as image classification and language modeling. However, these techniques typically ignore device-related objectives such as inference time, memory usage, and power consumption. Optimizing neural architecture for device-related objectives is immensely crucial for deploying deep networks on portable devices with limited computing resources. We propose DPP-Net: Device-aware Progressive Search for Pareto-optimal Neural Architectures, optimizing for both device-related (e.g., inference time and memory usage) and device-agnostic (e.g., accuracy and model size) objectives. DPP-Net employs a compact search space inspired by current state-of-the-art mobile CNNs, and further improves search efficiency by adopting progressive search (Liu et al. 2017). Experimental results on CIFAR-10 are poised to demonstrate the effectiveness of Pareto-optimal networks found by DPP-Net, for three different devices: (1) a workstation with Titan X GPU, (2) NVIDIA Jetson TX1 embedded system, and (3) mobile phone with ARM Cortex-A53. Compared to CondenseNet and NASNet (Mobile), DPP-Net achieves better performances: higher accuracy and shorter inference time on various devices. Additional experimental results show that models found by DPP-Net also achieve considerably-good performance on ImageNet as well.

arXiv:1806.07987

A Hierarchical Deep Architecture and Mini-Batch Selection Method For Joint Traffic Sign and Light Detection

Traffic light and sign detectors on autonomous cars are integral for road scene perception. The literature is abundant with deep learning networks that detect either lights or signs, not both, which makes them unsuitable for real-life deployment due to the limited graphics processing unit (GPU) memory and power available on embedded systems. The root cause of this issue is that no public dataset contains both traffic light and sign labels, which leads to difficulties in developing a joint detection framework. We present a deep hierarchical architecture in conjunction with a mini-batch proposal selection mechanism that allows a network to detect both traffic lights and signs from training on separate traffic light and sign datasets. Our method solves the overlapping issue where instances from one dataset are not labelled in the other dataset. We are the first to present a network that performs joint detection on traffic lights and signs. We measure our network on the Tsinghua-Tencent 100K benchmark for traffic sign detection and the Bosch Small Traffic Lights benchmark for traffic light detection and show it outperforms the existing Bosch Small Traffic light state-of-the-art method. We focus on autonomous car deployment and show our network is more suitable than others because of its low memory footprint and real-time image processing time. Qualitative results can be viewed at https://youtu.be/_YmogPzBXOw

arXiv:1806.07078

Learning to Update for Object Tracking with Recurrent Meta-learner

Model update lies at the heart of object tracking. Generally, model update is formulated as an online learning problem where a target model is learned over the online training set. Our key innovation is to \emph{formulate the model update problem in the meta-learning framework and learn the online learning algorithm itself using large numbers of offline videos}, i.e., \emph{learning to update}. The learned updater takes as input the online training set and outputs an updated target model. As a first attempt, we design the learned updater based on recurrent neural networks (RNNs) and demonstrate its application in a template-based tracker and a correlation filter-based tracker. Our learned updater consistently improves the base trackers and runs faster than realtime on GPU while requiring small memory footprint during testing. Experiments on standard benchmarks demonstrate that our learned updater outperforms commonly used update baselines including the efficient exponential moving average (EMA)-based update and the well-designed stochastic gradient descent (SGD)-based update. Equipped with our learned updater, the template-based tracker achieves state-of-the-art performance among realtime trackers on GPU.

arXiv:1806.07353

Faster SGD training by minibatch persistency

It is well known that, for most datasets, the use of large-size minibatches for Stochastic Gradient Descent (SGD) typically leads to slow convergence and poor generalization. On the other hand, large minibatches are of great practical interest as they allow for a better exploitation of modern GPUs. Previous literature on the subject concentrated on how to adjust the main SGD parameters (in particular, the learning rate) when using large minibatches. In this work we introduce an additional feature, that we call minibatch persistency, that consists in reusing the same minibatch for K consecutive SGD iterations. The computational conjecture here is that a large minibatch contains a significant sample of the training set, so one can afford to slightly overfitting it without worsening generalization too much. The approach is intended to speedup SGD convergence, and also has the advantage of reducing the overhead related to data loading on the internal GPU memory. We present computational results on CIFAR-10 with an AlexNet architecture, showing that even small persistency values (K=2 or 5) already lead to a significantly faster convergence and to a comparable (or even better) generalization than the standard "disposable minibatch" approach (K=1), in particular when large minibatches are used. The lesson learned is that minibatch persistency can be a simple yet effective way to deal with large minibatches.

arXiv:1304.3992

GPU Acclerated Automated Feature Extraction from Satellite Images

The availability of large volumes of remote sensing data insists on higher degree of automation in feature extraction, making it a need of the hour.The huge quantum of data that needs to be processed entails accelerated processing to be enabled.GPUs, which were originally designed to provide efficient visualization, are being massively employed for computation intensive parallel processing environments. Image processing in general and hence automated feature extraction, is highly computation intensive, where performance improvements have a direct impact on societal needs. In this context, an algorithm has been formulated for automated feature extraction from a panchromatic or multispectral image based on image processing techniques. Two Laplacian of Guassian (LoG) masks were applied on the image individually followed by detection of zero crossing points and extracting the pixels based on their standard deviation with the surrounding pixels. The two extracted images with different LoG masks were combined together which resulted in an image with the extracted features and edges. Finally the user is at liberty to apply the image smoothing step depending on the noise content in the extracted image. The image is passed through a hybrid median filter to remove the salt and pepper noise from the image. This paper discusses the aforesaid algorithm for automated feature extraction, necessity of deployment of GPUs for the same; system-level challenges and quantifies the benefits of integrating GPUs in such environment. The results demonstrate that substantial enhancement in performance margin can be achieved with the best utilization of GPU resources and an efficient parallelization strategy. Performance results in comparison with the conventional computing scenario have provided a speedup of 20x, on realization of this parallelizing strategy.

arXiv:1303.7032

A Massively Parallel Associative Memory Based on Sparse Neural Networks

Associative memories store content in such a way that the content can be later retrieved by presenting the memory with a small portion of the content, rather than presenting the memory with an address as in more traditional memories. Associative memories are used as building blocks for algorithms within database engines, anomaly detection systems, compression algorithms, and face recognition systems. A classical example of an associative memory is the Hopfield neural network. Recently, Gripon and Berrou have introduced an alternative construction which builds on ideas from the theory of error correcting codes and which greatly outperforms the Hopfield network in capacity, diversity, and efficiency. In this paper we implement a variation of the Gripon-Berrou associative memory on a general purpose graphical processing unit (GPU). The work of Gripon and Berrou proposes two retrieval rules, sum-of-sum and sum-of-max. The sum-of-sum rule uses only matrix-vector multiplication and is easily implemented on the GPU. The sum-of-max rule is much less straightforward to implement because it involves non-linear operations. However, the sum-of-max rule gives significantly better retrieval error rates. We propose a hybrid rule tailored for implementation on a GPU which achieves a 880-fold speedup without sacrificing any accuracy.

arXiv:1311.7194

Real-time High Resolution Fusion of Depth Maps on GPU

A system for live high quality surface reconstruction using a single moving depth camera on a commodity hardware is presented. High accuracy and real-time frame rate is achieved by utilizing graphics hardware computing capabilities via OpenCL and by using sparse data structure for volumetric surface representation. Depth sensor pose is estimated by combining serial texture registration algorithm with iterative closest points algorithm (ICP) aligning obtained depth map to the estimated scene model. Aligned surface is then fused into the scene. Kalman filter is used to improve fusion quality. Truncated signed distance function (TSDF) stored as block-based sparse buffer is used to represent surface. Use of sparse data structure greatly increases accuracy of scanned surfaces and maximum scanning area. Traditional GPU implementation of volumetric rendering and fusion algorithms were modified to exploit sparsity to achieve desired performance. Incorporation of texture registration for sensor pose estimation and Kalman filter for measurement integration improved accuracy and robustness of scanning process.

arXiv:1312.6186

GPU Asynchronous Stochastic Gradient Descent to Speed Up Neural Network Training

The ability to train large-scale neural networks has resulted in state-of-the-art performance in many areas of computer vision. These results have largely come from computational break throughs of two forms: model parallelism, e.g. GPU accelerated training, which has seen quick adoption in computer vision circles, and data parallelism, e.g. A-SGD, whose large scale has been used mostly in industry. We report early experiments with a system that makes use of both model parallelism and data parallelism, we call GPU A-SGD. We show using GPU A-SGD it is possible to speed up training of large convolutional neural networks useful for computer vision. We believe GPU A-SGD will make it possible to train larger networks on larger training sets in a reasonable amount of time.

arXiv:1312.6182

Large-Scale Paralleled Sparse Principal Component Analysis

Principal component analysis (PCA) is a statistical technique commonly used in multivariate data analysis. However, PCA can be difficult to interpret and explain since the principal components (PCs) are linear combinations of the original variables. Sparse PCA (SPCA) aims to balance statistical fidelity and interpretability by approximating sparse PCs whose projections capture the maximal variance of original data. In this paper we present an efficient and paralleled method of SPCA using graphics processing units (GPUs), which can process large blocks of data in parallel. Specifically, we construct parallel implementations of the four optimization formulations of the generalized power method of SPCA (GP-SPCA), one of the most efficient and effective SPCA approaches, on a GPU. The parallel GPU implementation of GP-SPCA (using CUBLAS) is up to eleven times faster than the corresponding CPU implementation (using CBLAS), and up to 107 times faster than a MatLab implementation. Extensive comparative experiments in several real-world datasets confirm that SPCA offers a practical advantage.

arXiv:1401.4107

Revisiting loss-specific training of filter-based MRFs for image restoration

It is now well known that Markov random fields (MRFs) are particularly effective for modeling image priors in low-level vision. Recent years have seen the emergence of two main approaches for learning the parameters in MRFs: (1) probabilistic learning using sampling-based algorithms and (2) loss-specific training based on MAP estimate. After investigating existing training approaches, it turns out that the performance of the loss-specific training has been significantly underestimated in existing work. In this paper, we revisit this approach and use techniques from bi-level optimization to solve it. We show that we can get a substantial gain in the final performance by solving the lower-level problem in the bi-level framework with high accuracy using our newly proposed algorithm. As a result, our trained model is on par with highly specialized image denoising algorithms and clearly outperforms probabilistically trained MRF models. Our findings suggest that for the loss-specific training scheme, solving the lower-level problem with higher accuracy is beneficial. Our trained model comes along with the additional advantage, that inference is extremely efficient. Our GPU-based implementation takes less than 1s to produce state-of-the-art performance.

arXiv:1312.5851

Fast Training of Convolutional Networks through FFTs

Convolutional networks are one of the most widely employed architectures in computer vision and machine learning. In order to leverage their ability to learn complex functions, large amounts of data are required for training. Training a large convolutional network to produce state-of-the-art results can take weeks, even when using modern GPUs. Producing labels using a trained network can also be costly when dealing with web-scale datasets. In this work, we present a simple algorithm which accelerates training and inference by a significant factor, and can yield improvements of over an order of magnitude compared to existing state-of-the-art implementations. This is done by computing convolutions as pointwise products in the Fourier domain while reusing the same transformed feature map many times. The algorithm is implemented on a GPU architecture and addresses a number of related challenges.

arXiv:1403.4238

Computer Vision Accelerators for Mobile Systems based on OpenCL GPGPU Co-Processing

In this paper, we present an OpenCL-based heterogeneous implementation of a computer vision algorithm -- image inpainting-based object removal algorithm -- on mobile devices. To take advantage of the computation power of the mobile processor, the algorithm workflow is partitioned between the CPU and the GPU based on the profiling results on mobile devices, so that the computationally-intensive kernels are accelerated by the mobile GPGPU (general-purpose computing using graphics processing units). By exploring the implementation trade-offs and utilizing the proposed optimization strategies at different levels including algorithm optimization, parallelism optimization, and memory access optimization, we significantly speed up the algorithm with the CPU-GPU heterogeneous implementation, while preserving the quality of the output images. Experimental results show that heterogeneous computing based on GPGPU co-processing can significantly speed up the computer vision algorithms and makes them practical on real-world mobile devices.

arXiv:1404.1066

Parallel Support Vector Machines in Practice

In this paper, we evaluate the performance of various parallel optimization methods for Kernel Support Vector Machines on multicore CPUs and GPUs. In particular, we provide the first comparison of algorithms with explicit and implicit parallelization. Most existing parallel implementations for multi-core or GPU architectures are based on explicit parallelization of Sequential Minimal Optimization (SMO)---the programmers identified parallelizable components and hand-parallelized them, specifically tuned for a particular architecture. We compare these approaches with each other and with implicitly parallelized algorithms---where the algorithm is expressed such that most of the work is done within few iterations with large dense linear algebra operations. These can be computed with highly-optimized libraries, that are carefully parallelized for a large variety of parallel platforms. We highlight the advantages and disadvantages of both approaches and compare them on various benchmark data sets. We find an approximate implicitly parallel algorithm which is surprisingly efficient, permits a much simpler implementation, and leads to unprecedented speedups in SVM training.

arXiv:1404.0774

GPU Accelerated Fractal Image Compression for Medical Imaging in Parallel Computing Platform

In this paper, we implemented both sequential and parallel version of fractal image compression algorithms using CUDA (Compute Unified Device Architecture) programming model for parallelizing the program in Graphics Processing Unit for medical images, as they are highly similar within the image itself. There are several improvement in the implementation of the algorithm as well. Fractal image compression is based on the self similarity of an image, meaning an image having similarity in majority of the regions. We take this opportunity to implement the compression algorithm and monitor the effect of it using both parallel and sequential implementation. Fractal compression has the property of high compression rate and the dimensionless scheme. Compression scheme for fractal image is of two kind, one is encoding and another is decoding. Encoding is very much computational expensive. On the other hand decoding is less computational. The application of fractal compression to medical images would allow obtaining much higher compression ratios. While the fractal magnification an inseparable feature of the fractal compression would be very useful in presenting the reconstructed image in a highly readable form. However, like all irreversible methods, the fractal compression is connected with the problem of information loss, which is especially troublesome in the medical imaging. A very time consuming encoding pro- cess, which can last even several hours, is another bothersome drawback of the fractal compression.

arXiv:1404.1521

Exploring the power of GPU's for training Polyglot language models

One of the major research trends currently is the evolution of heterogeneous parallel computing. GP-GPU computing is being widely used and several applications have been designed to exploit the massive parallelism that GP-GPU's have to offer. While GPU's have always been widely used in areas of computer vision for image processing, little has been done to investigate whether the massive parallelism provided by GP-GPU's can be utilized effectively for Natural Language Processing(NLP) tasks. In this work, we investigate and explore the power of GP-GPU's in the task of learning language models. More specifically, we investigate the performance of training Polyglot language models using deep belief neural networks. We evaluate the performance of training the model on the GPU and present optimizations that boost the performance on the GPU.One of the key optimizations, we propose increases the performance of a function involved in calculating and updating the gradient by approximately 50 times on the GPU for sufficiently large batch sizes. We show that with the above optimizations, the GP-GPU's performance on the task increases by factor of approximately 3-4. The optimizations we made are generic Theano optimizations and hence potentially boost the performance of other models which rely on these operations.We also show that these optimizations result in the GPU's performance at this task being now comparable to that on the CPU. We conclude by presenting a thorough evaluation of the applicability of GP-GPU's for this task and highlight the factors limiting the performance of training a Polyglot model on the GPU.