PyTorchKR Papers

최신 논문

총 2,153편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1908.08996

Justlookup: One Millisecond Deep Feature Extraction for Point Clouds By Lookup Tables

Deep models are capable of fitting complex high dimensional functions while usually yielding large computation load. There is no way to speed up the inference process by classical lookup tables due to the high-dimensional input and limited memory size. Recently, a novel architecture (PointNet) for point clouds has demonstrated that it is possible to obtain a complicated deep function from a set of 3-variable functions. In this paper, we exploit this property and apply a lookup table to encode these 3-variable functions. This method ensures that the inference time is only determined by the memory access no matter how complicated the deep function is. We conduct extensive experiments on ModelNet and ShapeNet datasets and demonstrate that we can complete the inference process in 1.5 ms on an Intel i7-8700 CPU (single core mode), 32x speedup over the PointNet architecture without any performance degradation.

arXiv:1908.11809

High Performance Scalable FPGA Accelerator for Deep Neural Networks

Low-precision is the first order knob for achieving higher Artificial Intelligence Operations (AI-TOPS). However the algorithmic space for sub-8-bit precision compute is diverse, with disruptive changes happening frequently, making FPGAs a natural choice for Deep Neural Network inference, In this work we present an FPGA-based accelerator for CNN inference acceleration. We use {\it INT-8-2} compute (with {\it 8 bit} activation and {2 bit} weights) which is recently showing promise in the literature, and which no known ASIC, CPU or GPU natively supports today. Using a novel Adaptive Logic Module (ALM) based design, as a departure from traditional DSP based designs, we are able to achieve high performance measurement of 5 AI-TOPS for {\it Arria10} and project a performance of 76 AI-TOPS at 0.7 TOPS/W for {\it Stratix10}. This exceeds known CPU, GPU performance and comes close to best known ASIC (TPU) numbers, while retaining the versatility of the FPGA platform for other applications.

arXiv:1908.11450

Neural Network Inference on Mobile SoCs

The ever-increasing demand from mobile Machine Learning (ML) applications calls for evermore powerful on-chip computing resources. Mobile devices are empowered with heterogeneous multi-processor Systems-on-Chips (SoCs) to process ML workloads such as Convolutional Neural Network (CNN) inference. Mobile SoCs house several different types of ML capable components on-die, such as CPU, GPU, and accelerators. These different components are capable of independently performing inference but with very different power-performance characteristics. In this article, we provide a quantitative evaluation of the inference capabilities of the different components on mobile SoCs. We also present insights behind their respective power-performance behavior. Finally, we explore the performance limit of the mobile SoCs by synergistically engaging all the components concurrently. We observe that a mobile SoC provides up to 2x improvement with parallel inference when all its components are engaged, as opposed to engaging only one component.

arXiv:1908.11338

TapirXLA: Embedding Fork-Join Parallelism into the XLA Compiler in TensorFlow Using Tapir

This work introduces TapirXLA, a replacement for TensorFlow's XLA compiler that embeds recursive fork-join parallelism into XLA's low-level representation of code. Machine-learning applications rely on efficient parallel processing to achieve performance, and they employ a variety of technologies to improve performance, including compiler technology. But compilers in machine-learning frameworks lack a deep understanding of parallelism, causing them to lose performance by missing optimizations on parallel computation. This work studies how Tapir, a compiler intermediate representation (IR) that embeds parallelism into a mainstream compiler IR, can be incorporated into a compiler for machine learning to remedy this problem. TapirXLA modifies the XLA compiler in TensorFlow to employ the Tapir/LLVM compiler to optimize low-level parallel computation. TapirXLA encodes the parallelism within high-level TensorFlow operations using Tapir's representation of fork-join parallelism. TapirXLA also exposes to the compiler implementations of linear-algebra library routines whose parallel operations are encoded using Tapir's representation. We compared the performance of TensorFlow using TapirXLA against TensorFlow using an unmodified XLA compiler. On four neural-network benchmarks, TapirXLA speeds up the parallel running time of the network by a geometric-mean multiplicative factor of 30% to 100%, across four CPU architectures.

arXiv:1908.09296

Exploring the Performance of Deep Residual Networks in Crazyhouse Chess

Crazyhouse is a chess variant that incorporates all of the classical chess rules, but allows users to drop pieces captured from the opponent as a normal move. Until 2018, all competitive computer engines for this board game made use of an alpha-beta pruning algorithm with a hand-crafted evaluation function for each position. Previous machine learning-based algorithms for just regular chess, such as NeuroChess and Giraffe, took hand-crafted evaluation features as input rather than a raw board representation. More recent projects, such as AlphaZero, reached massive success but required massive computational resources in order to reach its final strength. This paper describes the development of SixtyFour, an engine designed to compete in the chess variant of Crazyhouse with limited hardware. This specific variant poses a multitude of significant challenges due to its large branching factor, state-space complexity, and the multiple move types a player can make. We propose the novel creation of a neural network-based evaluation function for Crazyhouse. More importantly, we evaluate the effectiveness of an ensemble model, which allows the training time and datasets to be easily distributed on regular CPU hardware commodity. Early versions of the network have attained a playing level comparable to a strong amateur on online servers.

arXiv:1908.09207

Demystifying the MLPerf Benchmark Suite

MLPerf, an emerging machine learning benchmark suite strives to cover a broad range of applications of machine learning. We present a study on its characteristics and how the MLPerf benchmarks differ from some of the previous deep learning benchmarks like DAWNBench and DeepBench. We find that application benchmarks such as MLPerf (although rich in kernels) exhibit different features compared to kernel benchmarks such as DeepBench. MLPerf benchmark suite contains a diverse set of models which allows unveiling various bottlenecks in the system. Based on our findings, dedicated low latency interconnect between GPUs in multi-GPU systems is required for optimal distributed deep learning training. We also observe variation in scaling efficiency across the MLPerf models. The variation exhibited by the different models highlight the importance of smart scheduling strategies for multi-GPU training. Another observation is that CPU utilization increases with increase in number of GPUs used for training. Corroborating prior work we also observe and quantify improvements possible by compiler optimizations, mixed-precision training and use of Tensor Cores.

arXiv:1908.07847

CUDA optimized Neural Network predicts blood glucose control from quantified joint mobility and anthropometrics

Neural network training entails heavy computation with obvious bottlenecks. The Compute Unified Device Architecture (CUDA) programming model allows us to accelerate computation by passing the processing workload from the CPU to the graphics processing unit (GPU). In this paper, we leveraged the power of Nvidia GPUs to parallelize all of the computation involved in training, to accelerate a backpropagation feed-forward neural network with one hidden layer using CUDA and C++. This optimized neural network was tasked with predicting the level of glycated hemoglobin (HbA1c) from non-invasive markers. The rate of increase in the prevalence of Diabetes Mellitus has resulted in an urgent need for early detection and accurate diagnosis. However, due to the invasiveness and limitations of conventional tests, alternate means are being considered. Limited Joint Mobility (LJM) has been reported as an indicator for poor glycemic control. LJM of the fingers is quantified and its link to HbA1c is investigated along with other potential non-invasive markers of HbA1c. We collected readings of 33 potential markers from 120 participants at a clinic in south Trinidad. Our neural network achieved 95.65% accuracy on the training and 86.67% accuracy on the testing set for male participants and 97.73% and 66.67% accuracy on the training and testing sets for female participants. Using 960 CUDA cores from a Nvidia GeForce GTX 660, our parallelized neural network was trained 50 times faster on both subsets, than its corresponding CPU implementation on an Intel Core (TM) i7-3630QM 2.40 GHz CPU.

arXiv:1908.06972

PrivFT: Private and Fast Text Classification with Homomorphic Encryption

The need for privacy-preserving analytics is higher than ever due to the severity of privacy risks and to comply with new privacy regulations leading to an amplified interest in privacy-preserving techniques that try to balance between privacy and utility. In this work, we present an efficient method for Text Classification while preserving the privacy of the content using Fully Homomorphic Encryption (FHE). Our system (named \textbf{Priv}ate \textbf{F}ast \textbf{T}ext (PrivFT)) performs two tasks: 1) making inference of encrypted user inputs using a plaintext model and 2) training an effective model using an encrypted dataset. For inference, we train a supervised model and outline a system for homomorphic inference on encrypted user inputs with zero loss to prediction accuracy. In the second part, we show how to train a model using fully encrypted data to generate an encrypted model. We provide a GPU implementation of the Cheon-Kim-Kim-Song (CKKS) FHE scheme and compare it with existing CPU implementations to achieve 1 to 2 orders of magnitude speedup at various parameter settings. We implement PrivFT in GPUs to achieve a run time per inference of less than 0.66 seconds. Training on a relatively large encrypted dataset is more computationally intensive requiring 5.04 days.

arXiv:1908.06660공식 코드 · Papers with Code

Learning to play the Chess Variant Crazyhouse above World Champion Level with Deep Neural Networks and Human Data

Deep neural networks have been successfully applied in learning the board games Go, chess and shogi without prior knowledge by making use of reinforcement learning. Although starting from zero knowledge has been shown to yield impressive results, it is associated with high computationally costs especially for complex games. With this paper, we present CrazyAra which is a neural network based engine solely trained in supervised manner for the chess variant crazyhouse. Crazyhouse is a game with a higher branching factor than chess and there is only limited data of lower quality available compared to AlphaGo. Therefore, we focus on improving efficiency in multiple aspects while relying on low computational resources. These improvements include modifications in the neural network design and training configuration, the introduction of a data normalization step and a more sample efficient Monte-Carlo tree search which has a lower chance to blunder. After training on 569,537 human games for 1.5 days we achieve a move prediction accuracy of 60.4%. During development, versions of CrazyAra played professional human players. Most notably, CrazyAra achieved a four to one win over 2017 crazyhouse world champion Justin Tan (aka LM Jann Lee) who is more than 400 Elo higher rated compared to the average player in our training set. Furthermore, we test the playing strength of CrazyAra on CPU against all participants of the second Crazyhouse Computer Championships 2017, winning against twelve of the thirteen participants. Finally, for CrazyAraFish we continue training our model on generated engine games. In ten long-time control matches playing Stockfish 10, CrazyAraFish wins three games and draws one out of ten matches.

arXiv:1908.01314공식 코드 · Papers with Code

MoGA: Searching Beyond MobileNetV3

The evolution of MobileNets has laid a solid foundation for neural network applications on mobile end. With the latest MobileNetV3, neural architecture search again claimed its supremacy in network design. Unfortunately, till today all mobile methods mainly focus on CPU latencies instead of GPU, the latter, however, is much preferred in practice for it has faster speed, lower overhead and less interference. Bearing the target hardware in mind, we propose the first Mobile GPU-Aware (MoGA) neural architecture search in order to be precisely tailored for real-world applications. Further, the ultimate objective to devise a mobile network lies in achieving better performance by maximizing the utilization of bounded resources. Urging higher capability while restraining time consumption is not reconcilable. We alleviate the tension by weighted evolution techniques. Moreover, we encourage increasing the number of parameters for higher representational power. With 200x fewer GPU days than MnasNet, we obtain a series of models that outperform MobileNetV3 under the similar latency constraints, i.e., MoGA-A achieves 75.9% top-1 accuracy on ImageNet, MoGA-B meets 75.5% which costs only 0.5 ms more on mobile GPU. MoGA-C best attests GPU-awareness by reaching 75.3% and being slower on CPU but faster on GPU.The models and test code is made available here https://github.com/xiaomi-automl/MoGA.

arXiv:1907.10701공식 코드 · Papers with Code

Benchmarking TPU, GPU, and CPU Platforms for Deep Learning

Training deep learning models is compute-intensive and there is an industry-wide trend towards hardware specialization to improve performance. To systematically benchmark deep learning platforms, we introduce ParaDnn, a parameterized benchmark suite for deep learning that generates end-to-end models for fully connected (FC), convolutional (CNN), and recurrent (RNN) neural networks. Along with six real-world models, we benchmark Google's Cloud TPU v2/v3, NVIDIA's V100 GPU, and an Intel Skylake CPU platform. We take a deep dive into TPU architecture, reveal its bottlenecks, and highlight valuable lessons learned for future specialized system design. We also provide a thorough comparison of the platforms and find that each has unique strengths for some types of models. Finally, we quantify the rapid performance improvements that specialized software stacks provide for the TPU and GPU platforms.

arXiv:1907.10219

Efficient Circle-Based Camera Pose Tracking Free of PnP

Camera pose tracking attracts much interest both from academic and industrial communities, of which the methods based on planar markers are easy to be implemented. However, most of the existing methods need to identify multiple points in the marker images for matching to space points. Then, PnP and RANSAC are used to compute the camera pose. If cameras move fast or are far away from markers, matching is easy to generate errors and even RANSAC cannot remove incorrect matching. Then, the result by PnP cannot have good performance. To solve this problem, we design circular markers and represent 6D camera pose analytically and unifiedly as very concise forms from each of the marker by projective invariance. Afterwards, the pose is further optimized by a proposed nonlinear cost function based on a polar-n-direction geometric distance. The method is from imaged circle edges and without PnP/RANSAC, making pose tracking robust and accurate. Experimental results show that the proposed method outperforms the state of the arts in terms of noise, blur, and distance from camera to marker. Simultaneously, it can still run at about 100 FPS on a consumer computer with only CPU.

arXiv:1907.10147

A Neural Network-Based On-device Learning Anomaly Detector for Edge Devices

Semi-supervised anomaly detection is an approach to identify anomalies by learning the distribution of normal data. Backpropagation neural networks (i.e., BP-NNs) based approaches have recently drawn attention because of their good generalization capability. In a typical situation, BP-NN-based models are iteratively optimized in server machines with input data gathered from edge devices. However, (1) the iterative optimization often requires significant efforts to follow changes in the distribution of normal data (i.e., concept drift), and (2) data transfers between edge and server impose additional latency and energy consumption. To address these issues, we propose ONLAD and its IP core, named ONLAD Core. ONLAD is highly optimized to perform fast sequential learning to follow concept drift in less than one millisecond. ONLAD Core realizes on-device learning for edge devices at low power consumption, which realizes standalone execution where data transfers between edge and server are not required. Experiments show that ONLAD has favorable anomaly detection capability in an environment that simulates concept drift. Evaluations of ONLAD Core confirm that the training latency is 1.95x~6.58x faster than the other software implementations. Also, the runtime power consumption of ONLAD Core implemented on PYNQ-Z1 board, a small FPGA/CPU SoC platform, is 5.0x~25.4x lower than them.

arXiv:1907.09658공식 코드 · Papers with Code

Make Skeleton-based Action Recognition Model Smaller, Faster and Better

Although skeleton-based action recognition has achieved great success in recent years, most of the existing methods may suffer from a large model size and slow execution speed. To alleviate this issue, we analyze skeleton sequence properties to propose a Double-feature Double-motion Network (DD-Net) for skeleton-based action recognition. By using a lightweight network structure (i.e., 0.15 million parameters), DD-Net can reach a super fast speed, as 3,500 FPS on one GPU, or, 2,000 FPS on one CPU. By employing robust features, DD-Net achieves the state-of-the-art performance on our experimental datasets: SHREC (i.e., hand actions) and JHMDB (i.e., body actions). Our code will be released with this paper later.

arXiv:1907.09500공식 코드 · Papers with Code

deepCR: Cosmic Ray Rejection with Deep Learning

Cosmic ray (CR) identification and replacement are critical components of imaging and spectroscopic reduction pipelines involving solid-state detectors. We present deepCR, a deep learning based framework for CR identification and subsequent image inpainting based on the predicted CR mask. To demonstrate the effectiveness of this framework, we train and evaluate models on Hubble Space Telescope ACS/WFC images of sparse extragalactic fields, globular clusters, and resolved galaxies. We demonstrate that at a false positive rate of 0.5%, deepCR achieves close to 100% detection rates in both extragalactic and globular cluster fields, and 91% in resolved galaxy fields, which is a significant improvement over the current state-of-the-art method LACosmic. Compared to a multicore CPU implementation of LACosmic, deepCR CR mask predictions run up to 6.5 times faster on CPU and 90 times faster on a single GPU. For image inpainting, the mean squared errors of deepCR predictions are 20 times lower in globular cluster fields, 5 times lower in resolved galaxy fields, and 2.5 times lower in extragalactic fields, compared to the best performing non-neural technique tested. We present our framework and the trained models as an open-source Python project, with a simple-to-use API. To facilitate reproducibility of the results we also provide a benchmarking codebase.

arXiv:1907.09916

A Hardware-Efficient ADMM-Based SVM Training Algorithm for Edge Computing

This work demonstrates a hardware-efficient support vector machine (SVM) training algorithm via the alternative direction method of multipliers (ADMM) optimizer. Low-rank approximation is exploited to reduce the dimension of the kernel matrix by employing the Nystr\"{o}m method. Verified in four datasets, the proposed ADMM-based training algorithm with rank approximation reduces 32$\times$ of matrix dimension with only 2% drop in inference accuracy. Compared to the conventional sequential minimal optimization (SMO) algorithm, the ADMM-based training algorithm is able to achieve a 9.8$\times$10$^7$ shorter latency for training 2048 samples. Hardware design techniques, including pre-computation and memory sharing, are proposed to reduce the computational complexity by 62% and the memory usage by 60%. As a proof of concept, an epileptic seizure detector chip is designed to demonstrate the effectiveness of the proposed hardware-efficient training algorithm. The chip achieves a 153,310$\times$ higher energy efficiency and a 364$\times$ higher throughput-to-area ratio for SVM training than a high-end CPU. This work provides a promising solution for edge devices which require low-power and real-time training.

arXiv:1907.09831공식 코드 · Papers with Code

Real-Time Correlation Tracking via Joint Model Compression and Transfer

Correlation filters (CF) have received considerable attention in visual tracking because of their computational efficiency. Leveraging deep features via off-the-shelf CNN models (e.g., VGG), CF trackers achieve state-of-the-art performance while consuming a large number of computing resources. This limits deep CF trackers to be deployed to many mobile platforms on which only a single-core CPU is available. In this paper, we propose to jointly compress and transfer off-the-shelf CNN models within a knowledge distillation framework. We formulate a CNN model pretrained from the image classification task as a teacher network, and distill this teacher network into a lightweight student network as the feature extractor to speed up CF trackers. In the distillation process, we propose a fidelity loss to enable the student network to maintain the representation capability of the teacher network. Meanwhile, we design a tracking loss to adapt the objective of the student network from object recognition to visual tracking. The distillation process is performed offline on multiple layers and adaptively updates the student network using a background-aware online learning scheme. Extensive experiments on five challenging datasets demonstrate that the lightweight student network accelerates the speed of state-of-the-art deep CF trackers to real-time on a single-core CPU while maintaining almost the same tracking accuracy.

arXiv:1907.08467공식 코드 · Papers with Code

Accelerating Reinforcement Learning through GPU Atari Emulation

We introduce CuLE (CUDA Learning Environment), a CUDA port of the Atari Learning Environment (ALE) which is used for the development of deep reinforcement algorithms. CuLE overcomes many limitations of existing CPU-based emulators and scales naturally to multiple GPUs. It leverages GPU parallelization to run thousands of games simultaneously and it renders frames directly on the GPU, to avoid the bottleneck arising from the limited CPU-GPU communication bandwidth. CuLE generates up to 155M frames per hour on a single GPU, a finding previously achieved only through a cluster of CPUs. Beyond highlighting the differences between CPU and GPU emulators in the context of reinforcement learning, we show how to leverage the high throughput of CuLE by effective batching of the training data, and show accelerated convergence for A2C+V-trace. CuLE is available at https://github.com/NVLabs/cule .

arXiv:1907.07745

Real-Time Highly Accurate Dense Depth on a Power Budget using an FPGA-CPU Hybrid SoC

Obtaining highly accurate depth from stereo images in real time has many applications across computer vision and robotics, but in some contexts, upper bounds on power consumption constrain the feasible hardware to embedded platforms such as FPGAs. Whilst various stereo algorithms have been deployed on these platforms, usually cut down to better match the embedded architecture, certain key parts of the more advanced algorithms, e.g. those that rely on unpredictable access to memory or are highly iterative in nature, are difficult to deploy efficiently on FPGAs, and thus the depth quality that can be achieved is limited. In this paper, we leverage a FPGA-CPU chip to propose a novel, sophisticated, stereo approach that combines the best features of SGM and ELAS-based methods to compute highly accurate dense depth in real time. Our approach achieves an 8.7% error rate on the challenging KITTI 2015 dataset at over 50 FPS, with a power consumption of only 5W.

arXiv:1907.07323공식 코드 · Papers with Code

STRASS: A Light and Effective Method for Extractive Summarization Based on Sentence Embeddings

This paper introduces STRASS: Summarization by TRAnsformation Selection and Scoring. It is an extractive text summarization method which leverages the semantic information in existing sentence embedding spaces. Our method creates an extractive summary by selecting the sentences with the closest embeddings to the document embedding. The model learns a transformation of the document embedding to minimize the similarity between the extractive summary and the ground truth summary. As the transformation is only composed of a dense layer, the training can be done on CPU, therefore, inexpensive. Moreover, inference time is short and linear according to the number of sentences. As a second contribution, we introduce the French CASS dataset, composed of judgments from the French Court of cassation and their corresponding summaries. On this dataset, our results show that our method performs similarly to the state of the art extractive methods with effective training and inferring time.