PyTorchKR Papers

최신 논문

총 5,576편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1102.0183

High-Performance Neural Networks for Visual Object Classification

We present a fast, fully parameterizable GPU implementation of Convolutional Neural Network variants. Our feature extractors are neither carefully designed nor pre-wired, but rather learned in a supervised way. Our deep hierarchical architectures achieve the best published results on benchmarks for object classification (NORB, CIFAR10) and handwritten digit recognition (MNIST), with error rates of 2.53%, 19.51%, 0.35%, respectively. Deep nets trained by simple back-propagation perform better than more shallow ones. Learning is surprisingly rapid. NORB is completely trained within five epochs. Test error rates on MNIST drop to 2.42%, 0.97% and 0.48% after 1, 3 and 17 epochs, respectively.

arXiv:2109.12804

Fast-MD: Fast Multi-Decoder End-to-End Speech Translation with Non-Autoregressive Hidden Intermediates

The multi-decoder (MD) end-to-end speech translation model has demonstrated high translation quality by searching for better intermediate automatic speech recognition (ASR) decoder states as hidden intermediates (HI). It is a two-pass decoding model decomposing the overall task into ASR and machine translation sub-tasks. However, the decoding speed is not fast enough for real-world applications because it conducts beam search for both sub-tasks during inference. We propose Fast-MD, a fast MD model that generates HI by non-autoregressive (NAR) decoding based on connectionist temporal classification (CTC) outputs followed by an ASR decoder. We investigated two types of NAR HI: (1) parallel HI by using an autoregressive Transformer ASR decoder and (2) masked HI by using Mask-CTC, which combines CTC and the conditional masked language model. To reduce a mismatch in the ASR decoder between teacher-forcing during training and conditioning on CTC outputs during testing, we also propose sampling CTC outputs during training. Experimental evaluations on three corpora show that Fast-MD achieved about 2x and 4x faster decoding speed than that of the na\"ive MD model on GPU and CPU with comparable translation quality. Adopting the Conformer encoder and intermediate CTC loss further boosts its quality without sacrificing decoding speed.

arXiv:2109.12688

Nesterov Accelerated ADMM for Fast Diffeomorphic Image Registration

Deterministic approaches using iterative optimisation have been historically successful in diffeomorphic image registration (DiffIR). Although these approaches are highly accurate, they typically carry a significant computational burden. Recent developments in stochastic approaches based on deep learning have achieved sub-second runtimes for DiffIR with competitive registration accuracy, offering a fast alternative to conventional iterative methods. In this paper, we attempt to reduce this difference in speed whilst retaining the performance advantage of iterative approaches in DiffIR. We first propose a simple iterative scheme that functionally composes intermediate non-stationary velocity fields to handle large deformations in images whilst guaranteeing diffeomorphisms in the resultant deformation. We then propose a convex optimisation model that uses a regularisation term of arbitrary order to impose smoothness on these velocity fields and solve this model with a fast algorithm that combines Nesterov gradient descent and the alternating direction method of multipliers (ADMM). Finally, we leverage the computational power of GPU to implement this accelerated ADMM solver on a 3D cardiac MRI dataset, further reducing runtime to less than 2 seconds. In addition to producing strictly diffeomorphic deformations, our methods outperform both state-of-the-art deep learning-based and iterative DiffIR approaches in terms of dice and Hausdorff scores, with speed approaching the inference time of deep learning-based methods.

arXiv:2109.13076공식 코드 · Papers with Code

Using neural networks to solve the 2D Poisson equation for electric field computation in plasma fluid simulations

The Poisson equation is critical to get a self-consistent solution in plasma fluid simulations used for Hall effect thrusters and streamer discharges, since the Poisson solution appears as a source term of the unsteady nonlinear flow equations. As a first step, solving the 2D Poisson equation with zero Dirichlet boundary conditions using a deep neural network is investigated using multiple-scale architectures, defined in terms of number of branches, depth and receptive field. One key objective is to better understand how neural networks learn the Poisson solutions and provide guidelines to achieve optimal network configurations, especially when coupled to the time-varying Euler equations with plasma source terms. Here, the Receptive Field is found critical to correctly capture large topological structures of the field. The investigation of multiple architectures, losses, and hyperparameters provides an optimal network to solve accurately the steady Poisson problem. The performance of the optimal neural network solver, called PlasmaNet, is then monitored on meshes with increasing number of nodes, and compared with classical parallel linear solvers. Next, PlasmaNet is coupled with an unsteady Euler plasma fluid equations solver in the context of the electron plasma oscillation test case. In this time-evolving problem, a physical loss is necessary to produce a stable simulation. PlasmaNet is finally tested on a more complex case of discharge propagation involving chemistry and advection. The guidelines established in previous sections are applied to build the CNN to solve the same Poisson equation in cylindrical coordinates with different boundary conditions. Results reveal good CNN predictions and pave the way to new computational strategies using modern GPU-based hardware to predict unsteady problems involving a Poisson equation.

arXiv:2109.11978

Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement Learning

In this work, we present and study a training set-up that achieves fast policy generation for real-world robotic tasks by using massive parallelism on a single workstation GPU. We analyze and discuss the impact of different training algorithm components in the massively parallel regime on the final policy performance and training times. In addition, we present a novel game-inspired curriculum that is well suited for training with thousands of simulated robots in parallel. We evaluate the approach by training the quadrupedal robot ANYmal to walk on challenging terrain. The parallel approach allows training policies for flat terrain in under four minutes, and in twenty minutes for uneven terrain. This represents a speedup of multiple orders of magnitude compared to previous work. Finally, we transfer the policies to the real robot to validate the approach. We open-source our training code to help accelerate further research in the field of learned legged locomotion.

arXiv:2109.11818

MODNet-V: Improving Portrait Video Matting via Background Restoration

To address the challenging portrait video matting problem more precisely, existing works typically apply some matting priors that require additional user efforts to obtain, such as annotated trimaps or background images. In this work, we observe that instead of asking the user to explicitly provide a background image, we may recover it from the input video itself. To this end, we first propose a novel background restoration module (BRM) to recover the background image dynamically from the input video. BRM is extremely lightweight and can be easily integrated into existing matting models. By combining BRM with a recent image matting model, MODNet, we then present MODNet-V for portrait video matting. Benefited from the strong background prior provided by BRM, MODNet-V has only 1/3 of the parameters of MODNet but achieves comparable or even better performances. Our design allows MODNet-V to be trained in an end-to-end manner on a single NVIDIA 3090 GPU. Finally, we introduce a new patch refinement module (PRM) to adapt MODNet-V for high-resolution videos while keeping MODNet-V lightweight and fast.

arXiv:2109.11094

PredictionNet: Real-Time Joint Probabilistic Traffic Prediction for Planning, Control, and Simulation

Predicting the future motion of traffic agents is crucial for safe and efficient autonomous driving. To this end, we present PredictionNet, a deep neural network (DNN) that predicts the motion of all surrounding traffic agents together with the ego-vehicle's motion. All predictions are probabilistic and are represented in a simple top-down rasterization that allows an arbitrary number of agents. Conditioned on a multi-layer map with lane information, the network outputs future positions, velocities, and backtrace vectors jointly for all agents including the ego-vehicle in a single pass. Trajectories are then extracted from the output. The network can be used to simulate realistic traffic, and it produces competitive results on popular benchmarks. More importantly, it has been used to successfully control a real-world vehicle for hundreds of kilometers, by combining it with a motion planning/control subsystem. The network runs faster than real-time on an embedded GPU, and the system shows good generalization (across sensory modalities and locations) due to the choice of input representation. Furthermore, we demonstrate that by extending the DNN with reinforcement learning (RL), it can better handle rare or unsafe events like aggressive maneuvers and crashes.

arXiv:2109.11067

Serving DNN Models with Multi-Instance GPUs: A Case of the Reconfigurable Machine Scheduling Problem

Multi-Instance GPU (MIG) is a new feature introduced by NVIDIA A100 GPUs that partitions one physical GPU into multiple GPU instances. With MIG, A100 can be the most cost-efficient GPU ever for serving Deep Neural Networks (DNNs). However, discovering the most efficient GPU partitions is challenging. The underlying problem is NP-hard; moreover, it is a new abstract problem, which we define as the Reconfigurable Machine Scheduling Problem (RMS). This paper studies serving DNNs with MIG, a new case of RMS. We further propose a solution, MIG-serving. MIG- serving is an algorithm pipeline that blends a variety of newly designed algorithms and customized classic algorithms, including a heuristic greedy algorithm, Genetic Algorithm (GA), and Monte Carlo Tree Search algorithm (MCTS). We implement MIG-serving on Kubernetes. Our experiments show that compared to using A100 as-is, MIG-serving can save up to 40% of GPUs while providing the same throughput.

arXiv:2109.10847공식 코드 · Papers with Code

Small-Bench NLP: Benchmark for small single GPU trained models in Natural Language Processing

Recent progress in the Natural Language Processing domain has given us several State-of-the-Art (SOTA) pretrained models which can be finetuned for specific tasks. These large models with billions of parameters trained on numerous GPUs/TPUs over weeks are leading in the benchmark leaderboards. In this paper, we discuss the need for a benchmark for cost and time effective smaller models trained on a single GPU. This will enable researchers with resource constraints experiment with novel and innovative ideas on tokenization, pretraining tasks, architecture, fine tuning methods etc. We set up Small-Bench NLP, a benchmark for small efficient neural language models trained on a single GPU. Small-Bench NLP benchmark comprises of eight NLP tasks on the publicly available GLUE datasets and a leaderboard to track the progress of the community. Our ELECTRA-DeBERTa (15M parameters) small model architecture achieves an average score of 81.53 which is comparable to that of BERT-Base's 82.20 (110M parameters). Our models, code and leaderboard are available at https://github.com/smallbenchnlp

arXiv:2109.10739

Predicting Efficiency/Effectiveness Trade-offs for Dense vs. Sparse Retrieval Strategy Selection

Over the last few years, contextualized pre-trained transformer models such as BERT have provided substantial improvements on information retrieval tasks. Recent approaches based on pre-trained transformer models such as BERT, fine-tune dense low-dimensional contextualized representations of queries and documents in embedding space. While these dense retrievers enjoy substantial retrieval effectiveness improvements compared to sparse retrievers, they are computationally intensive, requiring substantial GPU resources, and dense retrievers are known to be more expensive from both time and resource perspectives. In addition, sparse retrievers have been shown to retrieve complementary information with respect to dense retrievers, leading to proposals for hybrid retrievers. These hybrid retrievers leverage low-cost, exact-matching based sparse retrievers along with dense retrievers to bridge the semantic gaps between query and documents. In this work, we address this trade-off between the cost and utility of sparse vs dense retrievers by proposing a classifier to select a suitable retrieval strategy (i.e., sparse vs. dense vs. hybrid) for individual queries. Leveraging sparse retrievers for queries which can be answered with sparse retrievers decreases the number of calls to GPUs. Consequently, while utility is maintained, query latency decreases. Although we use less computational resources and spend less time, we still achieve improved performance. Our classifier can select between sparse and dense retrieval strategies based on the query alone. We conduct experiments on the MS MARCO passage dataset demonstrating an improved range of efficiency/effectiveness trade-offs between purely sparse, purely dense or hybrid retrieval strategies, allowing an appropriate strategy to be selected based on a target latency and resource budget.

arXiv:2109.10593

Emulating Aerosol Microphysics with Machine Learning

Aerosol particles play an important role in the climate system by absorbing and scattering radiation and influencing cloud properties. They are also one of the biggest sources of uncertainty for climate modeling. Many climate models do not include aerosols in sufficient detail. In order to achieve higher accuracy, aerosol microphysical properties and processes have to be accounted for. This is done in the ECHAM-HAM global climate aerosol model using the M7 microphysics model, but increased computational costs make it very expensive to run at higher resolutions or for a longer time. We aim to use machine learning to approximate the microphysics model at sufficient accuracy and reduce the computational cost by being fast at inference time. The original M7 model is used to generate data of input-output pairs to train a neural network on it. By using a special logarithmic transform we are able to learn the variables tendencies achieving an average $R^2$ score of $89\%$. On a GPU we achieve a speed-up of 120 compared to the original model.

arXiv:2109.09854

Object Detection in Thermal Spectrum for Advanced Driver-Assistance Systems (ADAS)

Object detection in thermal infrared spectrum provides more reliable data source in low-lighting conditions and different weather conditions, as it is useful both in-cabin and outside for pedestrian, animal, and vehicular detection as well as for detecting street-signs & lighting poles. This paper is about exploring and adapting state-of-the-art object detection and classifier framework on thermal vision with seven distinct classes for advanced driver-assistance systems (ADAS). The trained network variants on public datasets are validated on test data with three different test approaches which include test-time with no augmentation, test-time augmentation, and test-time with model ensembling. Additionally, the efficacy of trained networks is tested on locally gathered novel test-data captured with an uncooled LWIR prototype thermal camera in challenging weather and environmental scenarios. The performance analysis of trained models is investigated by computing precision, recall, and mean average precision scores (mAP). Furthermore, the trained model architecture is optimized using TensorRT inference accelerator and deployed on resource-constrained edge hardware Nvidia Jetson Nano to explicitly reduce the inference time on GPU as well as edge devices for further real-time onboard installations.

arXiv:2109.08870

Fast query-by-example speech search using separable model

Traditional Query-by-Example (QbE) speech search approaches usually use methods based on frame-level features, while state-of-the-art approaches tend to use models based on acoustic word embeddings (AWEs) to transform variable length audio signals into fixed length feature vector representations. However, these approaches cannot meet the requirements of the search quality as well as speed at the same time. In this paper, we propose a novel fast QbE speech search method based on separable models to fix this problem. First, a QbE speech search training framework is introduced. Second, we design a novel model inference scheme based on RepVGG which can efficiently improve the QbE search quality. Third, we modify and improve our QbE speech search model according to the proposed model inference scheme. Experiments on keywords dataset shows that our proposed method can improve the GPU Real-time Factor (RTF) from 1/150 to 1/2300 by just applying separable model scheme and outperforms other state-of-the-art methods.

arXiv:2109.08815

Probabilistic Inference of Simulation Parameters via Parallel Differentiable Simulation

To accurately reproduce measurements from the real world, simulators need to have an adequate model of the physical system and require the parameters of the model be identified. We address the latter problem of estimating parameters through a Bayesian inference approach that approximates a posterior distribution over simulation parameters given real sensor measurements. By extending the commonly used Gaussian likelihood model for trajectories via the multiple-shooting formulation, our chosen particle-based inference algorithm Stein Variational Gradient Descent is able to identify highly nonlinear, underactuated systems. We leverage GPU code generation and differentiable simulation to evaluate the likelihood and its gradient for many particles in parallel. Our algorithm infers non-parametric distributions over simulation parameters more accurately than comparable baselines and handles constraints over parameters efficiently through gradient-based optimization. We evaluate estimation performance on several physical experiments. On an underactuated mechanism where a 7-DOF robot arm excites an object with an unknown mass configuration, we demonstrate how our inference technique can identify symmetries between the parameters and provide highly accurate predictions. Project website: https://uscresl.github.io/prob-diff-sim

arXiv:2109.08710

On-device neural speech synthesis

Recent advances in text-to-speech (TTS) synthesis, such as Tacotron and WaveRNN, have made it possible to construct a fully neural network based TTS system, by coupling the two components together. Such a system is conceptually simple as it only takes grapheme or phoneme input, uses Mel-spectrogram as an intermediate feature, and directly generates speech samples. The system achieves quality equal or close to natural speech. However, the high computational cost of the system and issues with robustness have limited their usage in real-world speech synthesis applications and products. In this paper, we present key modeling improvements and optimization strategies that enable deploying these models, not only on GPU servers, but also on mobile devices. The proposed system can generate high-quality 24 kHz speech at 5x faster than real time on server and 3x faster than real time on mobile devices.

arXiv:2109.09495공식 코드 · Papers with Code

GhostShiftAddNet: More Features from Energy-Efficient Operations

Deep convolutional neural networks (CNNs) are computationally and memory intensive. In CNNs, intensive multiplication can have resource implications that may challenge the ability for effective deployment of inference on resource-constrained edge devices. This paper proposes GhostShiftAddNet, where the motivation is to implement a hardware-efficient deep network: a multiplication-free CNN with fewer redundant features. We introduce a new bottleneck block, GhostSA, that converts all multiplications in the block to cheap operations. The bottleneck uses an appropriate number of bit-shift filters to process intrinsic feature maps, then applies a series of transformations that consist of bit-wise shifts with addition operations to generate more feature maps that fully learn to capture information underlying intrinsic features. We schedule the number of bit-shift and addition operations for different hardware platforms. We conduct extensive experiments and ablation studies with desktop and embedded (Jetson Nano) devices for implementation and measurements. We demonstrate the proposed GhostSA block can replace bottleneck blocks in the backbone of state-of-the-art networks architectures and gives improved performance on image classification benchmarks. Further, our GhostShiftAddNet can achieve higher classification accuracy with fewer FLOPs and parameters (reduced by up to 3x) than GhostNet. When compared to GhostNet, inference latency on the Jetson Nano is improved by 1.3x and 2x on the GPU and CPU respectively. Code is available open-source on \url{https://github.com/JIABI/GhostShiftAddNet}.

arXiv:2109.08937공식 코드 · Papers with Code

UNetFormer: A UNet-like Transformer for Efficient Semantic Segmentation of Remote Sensing Urban Scene Imagery

Semantic segmentation of remotely sensed urban scene images is required in a wide range of practical applications, such as land cover mapping, urban change detection, environmental protection, and economic assessment.Driven by rapid developments in deep learning technologies, the convolutional neural network (CNN) has dominated semantic segmentation for many years. CNN adopts hierarchical feature representation, demonstrating strong capabilities for local information extraction. However, the local property of the convolution layer limits the network from capturing the global context. Recently, as a hot topic in the domain of computer vision, Transformer has demonstrated its great potential in global information modelling, boosting many vision-related tasks such as image classification, object detection, and particularly semantic segmentation. In this paper, we propose a Transformer-based decoder and construct a UNet-like Transformer (UNetFormer) for real-time urban scene segmentation. For efficient segmentation, the UNetFormer selects the lightweight ResNet18 as the encoder and develops an efficient global-local attention mechanism to model both global and local information in the decoder. Extensive experiments reveal that our method not only runs faster but also produces higher accuracy compared with state-of-the-art lightweight models. Specifically, the proposed UNetFormer achieved 67.8% and 52.4% mIoU on the UAVid and LoveDA datasets, respectively, while the inference speed can achieve up to 322.4 FPS with a 512x512 input on a single NVIDIA GTX 3090 GPU. In further exploration, the proposed Transformer-based decoder combined with a Swin Transformer encoder also achieves the state-of-the-art result (91.3% F1 and 84.1% mIoU) on the Vaihingen dataset. The source code will be freely available at https://github.com/WangLibo1995/GeoSeg.

arXiv:2109.08250

Are we ready for beyond-application high-volume data? The Reeds robot perception benchmark dataset

This paper presents a dataset, called Reeds, for research on robot perception algorithms. The dataset aims to provide demanding benchmark opportunities for algorithms, rather than providing an environment for testing application-specific solutions. A boat was selected as a logging platform in order to provide highly dynamic kinematics. The sensor package includes six high-performance vision sensors, two long-range lidars, radar, as well as GNSS and an IMU. The spatiotemporal resolution of sensors were maximized in order to provide large variations and flexibility in the data, offering evaluation at a large number of different resolution presets based on the resolution found in other datasets. Reeds also provides means of a fair and reproducible comparison of algorithms, by running all evaluations on a common server backend. As the dataset contains massive-scale data, the evaluation principle also serves as a way to avoid moving data unnecessarily. It was also found that naive evaluation of algorithms, where each evaluation is computed sequentially, was not practical as the fetch and decode task of each frame would not scale well. Instead, each frame is only decoded once and then fed to all algorithms in parallel, including for GPU-based algorithms.

arXiv:2109.08580공식 코드 · Papers with Code

Self-Supervised Neural Architecture Search for Imbalanced Datasets

Neural Architecture Search (NAS) provides state-of-the-art results when trained on well-curated datasets with annotated labels. However, annotating data or even having balanced number of samples can be a luxury for practitioners from different scientific fields, e.g., in the medical domain. To that end, we propose a NAS-based framework that bears the threefold contributions: (a) we focus on the self-supervised scenario, i.e., where no labels are required to determine the architecture, and (b) we assume the datasets are imbalanced, (c) we design each component to be able to run on a resource constrained setup, i.e., on a single GPU (e.g. Google Colab). Our components build on top of recent developments in self-supervised learning~\citep{zbontar2021barlow}, self-supervised NAS~\citep{kaplan2020self} and extend them for the case of imbalanced datasets. We conduct experiments on an (artificially) imbalanced version of CIFAR-10 and we demonstrate our proposed method outperforms standard neural networks, while using $27\times$ less parameters. To validate our assumption on a naturally imbalanced dataset, we also conduct experiments on ChestMNIST and COVID-19 X-ray. The results demonstrate how the proposed method can be used in imbalanced datasets, while it can be fully run on a single GPU. Code is available \href{https://github.com/TimofeevAlex/ssnas_imbalanced}{here}.

arXiv:2109.08219공식 코드 · Papers with Code

Dr. Top-k: Delegate-Centric Top-k on GPUs

Recent top-$k$ computation efforts explore the possibility of revising various sorting algorithms to answer top-$k$ queries on GPUs. These endeavors, unfortunately, perform significantly more work than needed. This paper introduces Dr. Top-k, a Delegate-centric top-$k$ system on GPUs that can reduce the top-$k$ workloads significantly. Particularly, it contains three major contributions: First, we introduce a comprehensive design of the delegate-centric concept, including maximum delegate, delegate-based filtering, and $\beta$ delegate mechanisms to help reduce the workload for top-$k$ up to more than 99%. Second, due to the difficulty and importance of deriving a proper subrange size, we perform a rigorous theoretical analysis, coupled with thorough experimental validations to identify the desirable subrange size. Third, we introduce four key system optimizations to enable fast multi-GPU top-$k$ computation. Taken together, this work constantly outperforms the state-of-the-art.