PyTorchKR Papers

최신 논문

총 5,576편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2205.09332공식 코드 · Papers with Code

Accelerated Training of Physics-Informed Neural Networks (PINNs) using Meshless Discretizations

We present a new technique for the accelerated training of physics-informed neural networks (PINNs): discretely-trained PINNs (DT-PINNs). The repeated computation of partial derivative terms in the PINN loss functions via automatic differentiation during training is known to be computationally expensive, especially for higher-order derivatives. DT-PINNs are trained by replacing these exact spatial derivatives with high-order accurate numerical discretizations computed using meshless radial basis function-finite differences (RBF-FD) and applied via sparse-matrix vector multiplication. The use of RBF-FD allows for DT-PINNs to be trained even on point cloud samples placed on irregular domain geometries. Additionally, though traditional PINNs (vanilla-PINNs) are typically stored and trained in 32-bit floating-point (fp32) on the GPU, we show that for DT-PINNs, using fp64 on the GPU leads to significantly faster training times than fp32 vanilla-PINNs with comparable accuracy. We demonstrate the efficiency and accuracy of DT-PINNs via a series of experiments. First, we explore the effect of network depth on both numerical and automatic differentiation of a neural network with random weights and show that RBF-FD approximations of third-order accuracy and above are more efficient while being sufficiently accurate. We then compare the DT-PINNs to vanilla-PINNs on both linear and nonlinear Poisson equations and show that DT-PINNs achieve similar losses with 2-4x faster training times on a consumer GPU. Finally, we also demonstrate that similar results can be obtained for the PINN solution to the heat equation (a space-time problem) by discretizing the spatial derivatives using RBF-FD and using automatic differentiation for the temporal derivative. Our results show that fp64 DT-PINNs offer a superior cost-accuracy profile to fp32 vanilla-PINNs.

arXiv:2205.09646

Great Power, Great Responsibility: Recommendations for Reducing Energy for Training Language Models

The energy requirements of current natural language processing models continue to grow at a rapid, unsustainable pace. Recent works highlighting this problem conclude there is an urgent need for methods that reduce the energy needs of NLP and machine learning more broadly. In this article, we investigate techniques that can be used to reduce the energy consumption of common NLP applications. In particular, we focus on techniques to measure energy usage and different hardware and datacenter-oriented settings that can be tuned to reduce energy consumption for training and inference for language models. We characterize the impact of these settings on metrics such as computational performance and energy consumption through experiments conducted on a high performance computing system as well as popular cloud computing platforms. These techniques can lead to significant reduction in energy consumption when training language models or their use for inference. For example, power-capping, which limits the maximum power a GPU can consume, can enable a 15\% decrease in energy usage with marginal increase in overall computation time when training a transformer-based language model.

arXiv:2205.09470공식 코드 · Papers with Code

Nebula-I: A General Framework for Collaboratively Training Deep Learning Models on Low-Bandwidth Cloud Clusters

The ever-growing model size and scale of compute have attracted increasing interests in training deep learning models over multiple nodes. However, when it comes to training on cloud clusters, especially across remote clusters, huge challenges are faced. In this work, we introduce a general framework, Nebula-I, for collaboratively training deep learning models over remote heterogeneous clusters, the connections between which are low-bandwidth wide area networks (WANs). We took natural language processing (NLP) as an example to show how Nebula-I works in different training phases that include: a) pre-training a multilingual language model using two remote clusters; and b) fine-tuning a machine translation model using knowledge distilled from pre-trained models, which run through the most popular paradigm of recent deep learning. To balance the accuracy and communication efficiency, in Nebula-I, parameter-efficient training strategies, hybrid parallel computing methods and adaptive communication acceleration techniques are jointly applied. Meanwhile, security strategies are employed to guarantee the safety, reliability and privacy in intra-cluster computation and inter-cluster communication. Nebula-I is implemented with the PaddlePaddle deep learning framework, which can support collaborative training over heterogeneous hardware, e.g. GPU and NPU. Experiments demonstrate that the proposed framework could substantially maximize the training efficiency while preserving satisfactory NLP performance. By using Nebula-I, users can run large-scale training tasks over cloud clusters with minimum developments, and the utility of existed large pre-trained models could be further promoted. We also introduced new state-of-the-art results on cross-lingual natural language inference tasks, which are generated based upon a novel learning framework and Nebula-I.

arXiv:2205.09248공식 코드 · Papers with Code

MESH2IR: Neural Acoustic Impulse Response Generator for Complex 3D Scenes

We propose a mesh-based neural network (MESH2IR) to generate acoustic impulse responses (IRs) for indoor 3D scenes represented using a mesh. The IRs are used to create a high-quality sound experience in interactive applications and audio processing. Our method can handle input triangular meshes with arbitrary topologies (2K - 3M triangles). We present a novel training technique to train MESH2IR using energy decay relief and highlight its benefits. We also show that training MESH2IR on IRs preprocessed using our proposed technique significantly improves the accuracy of IR generation. We reduce the non-linearity in the mesh space by transforming 3D scene meshes to latent space using a graph convolution network. Our MESH2IR is more than 200 times faster than a geometric acoustic algorithm on a CPU and can generate more than 10,000 IRs per second on an NVIDIA GeForce RTX 2080 Ti GPU for a given furnished indoor 3D scene. The acoustic metrics are used to characterize the acoustic environment. We show that the acoustic metrics of the IRs predicted from our MESH2IR match the ground truth with less than 10% error. We also highlight the benefits of MESH2IR on audio and speech processing applications such as speech dereverberation and speech separation. To the best of our knowledge, ours is the first neural-network-based approach to predict IRs from a given 3D scene mesh in real-time.

arXiv:2003.03074

Real-time Tone Mapping: A State of the Art Report

The rising demand for high quality display has ensued active research in high dynamic range (HDR) imaging, which has the potential to replace the standard dynamic range imaging. This is due to HDR's features like accurate reproducibility of a scene with its entire spectrum of visible lighting and color depth. But this capability comes with expensive capture, display, storage and distribution resource requirements. Also, display of HDR images/video content on an ordinary display device with limited dynamic range requires some form of adaptation. Many adaptation algorithms, widely known as tone mapping operators, have been studied and proposed in the last few decades. In this state of the art report, we present a comprehensive survey of 50+ tone mapping algorithms that have been implemented on hardware for acceleration and real-time performance. These algorithms have been adapted or redesigned to make them hardware-friendly. All real-time application poses strict timing constraints which requires time exact processing of the algorithm. This design challenge require novel solution, and in this report we focus on these issues. In this we survey will discuss those tonemap algorithms which have been implemented on GPU [1-10], FPGA [11-41], and ASIC [42-53] in terms of their hardware specifications and performance. Output image quality is an important metric for tonemap algorithms. From our literature survey we found that, various objective quality metrics have been used to demonstrate the functionality of adapting the algorithm on hardware platform. We have compiled and studied all the metrics used in this survey [54-67]. Finally, in this report we demonstrate the link between hardware cost and image quality thereby illustrating the underlying trade-off which will be useful for the research community.

arXiv:2205.07141

BackLink: Supervised Local Training with Backward Links

Empowered by the backpropagation (BP) algorithm, deep neural networks have dominated the race in solving various cognitive tasks. The restricted training pattern in the standard BP requires end-to-end error propagation, causing large memory cost and prohibiting model parallelization. Existing local training methods aim to resolve the training obstacle by completely cutting off the backward path between modules and isolating their gradients to reduce memory cost and accelerate the training process. These methods prevent errors from flowing between modules and hence information exchange, resulting in inferior performance. This work proposes a novel local training algorithm, BackLink, which introduces inter-module backward dependency and allows errors to flow between modules. The algorithm facilitates information to flow backward along with the network. To preserve the computational advantage of local training, BackLink restricts the error propagation length within the module. Extensive experiments performed in various deep convolutional neural networks demonstrate that our method consistently improves the classification performance of local training algorithms over other methods. For example, in ResNet32 with 16 local modules, our method surpasses the conventional greedy local training method by 4.00\% and a recent work by 1.83\% in accuracy on CIFAR10, respectively. Analysis of computational costs reveals that small overheads are incurred in GPU memory costs and runtime on multiple GPUs. Our method can lead up to a 79\% reduction in memory cost and 52\% in simulation runtime in ResNet110 compared to the standard BP. Therefore, our method could create new opportunities for improving training algorithms towards better efficiency and biological plausibility.

arXiv:2205.07134공식 코드 · Papers with Code

ETAD: Training Action Detection End to End on a Laptop

Temporal action detection (TAD) with end-to-end training often suffers from the pain of huge demand for computing resources due to long video duration. In this work, we propose an efficient temporal action detector (ETAD) that can train directly from video frames with extremely low GPU memory consumption. Our main idea is to minimize and balance the heavy computation among features and gradients in each training iteration. We propose to sequentially forward the snippet frame through the video encoder, and backward only a small necessary portion of gradients to update the encoder. To further alleviate the computational redundancy in training, we propose to dynamically sample only a small subset of proposals during training. Moreover, various sampling strategies and ratios are studied for both the encoder and detector. ETAD achieves state-of-the-art performance on TAD benchmarks with remarkable efficiency. On ActivityNet-1.3, training ETAD in 18 hours can reach 38.25% average mAP with only 1.3 GB memory consumption per video under end-to-end training. Our code will be publicly released.

arXiv:2205.06688공식 코드 · Papers with Code

A Unified Framework for Implicit Sinkhorn Differentiation

The Sinkhorn operator has recently experienced a surge of popularity in computer vision and related fields. One major reason is its ease of integration into deep learning frameworks. To allow for an efficient training of respective neural networks, we propose an algorithm that obtains analytical gradients of a Sinkhorn layer via implicit differentiation. In comparison to prior work, our framework is based on the most general formulation of the Sinkhorn operator. It allows for any type of loss function, while both the target capacities and cost matrices are differentiated jointly. We further construct error bounds of the resulting algorithm for approximate inputs. Finally, we demonstrate that for a number of applications, simply replacing automatic differentiation with our algorithm directly improves the stability and accuracy of the obtained gradients. Moreover, we show that it is computationally more efficient, particularly when resources like GPU memory are scarce.

arXiv:2205.05535공식 코드 · Papers with Code

Clinical Prompt Learning with Frozen Language Models

Prompt learning is a new paradigm in the Natural Language Processing (NLP) field which has shown impressive performance on a number of natural language tasks with common benchmarking text datasets in full, few-shot, and zero-shot train-evaluation setups. Recently, it has even been observed that large but frozen pre-trained language models (PLMs) with prompt learning outperform smaller but fine-tuned models. However, as with many recent NLP trends, the performance of even the largest PLMs such as GPT-3 do not perform well on specialized domains (e.g. medical text), and the common practice to achieve State of the Art (SoTA) results still consists of pre-training and fine-tuning the PLMs on downstream tasks. The reliance on fine-tuning large PLMs is problematic in clinical settings where data is often held in non-GPU environments, and more resource efficient methods of training specialized domain models is crucial. We investigated the viability of prompt learning on clinically meaningful decision tasks and directly compared with more traditional fine-tuning methods. Results are partially in line with the prompt learning literature, with prompt learning able to match or improve on traditional fine-tuning with substantially fewer trainable parameters and requiring less training data. We argue that prompt learning therefore provides lower computational resource costs applicable to clinical settings, that can serve as an alternative to fine-tuning ever increasing in size PLMs. Complementary code to reproduce experiments presented in this work can be found at: https://github.com/NtaylorOX/Public_Clinical_Prompt.

arXiv:2205.05398

Scalable Stochastic Parametric Verification with Stochastic Variational Smoothed Model Checking

Parametric verification of linear temporal properties for stochastic models can be expressed as computing the satisfaction probability of a certain property as a function of the parameters of the model. Smoothed model checking (smMC) aims at inferring the satisfaction function over the entire parameter space from a limited set of observations obtained via simulation. As observations are costly and noisy, smMC is framed as a Bayesian inference problem so that the estimates have an additional quantification of the uncertainty. In smMC the authors use Gaussian Processes (GP), inferred by means of the Expectation Propagation algorithm. This approach provides accurate reconstructions with statistically sound quantification of the uncertainty. However, it inherits the well-known scalability issues of GP. In this paper, we exploit recent advances in probabilistic machine learning to push this limitation forward, making Bayesian inference of smMC scalable to larger datasets and enabling its application to models with high dimensional parameter spaces. We propose Stochastic Variational Smoothed Model Checking (SV-smMC), a solution that exploits stochastic variational inference (SVI) to approximate the posterior distribution of the smMC problem. The strength and flexibility of SVI make SV-smMC applicable to two alternative probabilistic models: Gaussian Processes (GP) and Bayesian Neural Networks (BNN). The core ingredient of SVI is a stochastic gradient-based optimization that makes inference easily parallelizable and that enables GPU acceleration. In this paper, we compare the performances of smMC against those of SV-smMC by looking at the scalability, the computational efficiency and the accuracy of the reconstructed satisfaction function.

arXiv:2205.05069

Accelerating the Training of Video Super-Resolution Models

Despite that convolution neural networks (CNN) have recently demonstrated high-quality reconstruction for video super-resolution (VSR), efficiently training competitive VSR models remains a challenging problem. It usually takes an order of magnitude more time than training their counterpart image models, leading to long research cycles. Existing VSR methods typically train models with fixed spatial and temporal sizes from beginning to end. The fixed sizes are usually set to large values for good performance, resulting to slow training. However, is such a rigid training strategy necessary for VSR? In this work, we show that it is possible to gradually train video models from small to large spatial/temporal sizes, i.e., in an easy-to-hard manner. In particular, the whole training is divided into several stages and the earlier stage has smaller training spatial shape. Inside each stage, the temporal size also varies from short to long while the spatial size remains unchanged. Training is accelerated by such a multigrid training strategy, as most of computation is performed on smaller spatial and shorter temporal shapes. For further acceleration with GPU parallelization, we also investigate the large minibatch training without the loss in accuracy. Extensive experiments demonstrate that our method is capable of largely speeding up training (up to $6.2\times$ speedup in wall-clock training time) without performance drop for various VSR models. The code is available at https://github.com/TencentARC/Efficient-VSR-Training.

arXiv:2205.04702

Training Personalized Recommendation Systems from (GPU) Scratch: Look Forward not Backwards

Personalized recommendation models (RecSys) are one of the most popular machine learning workload serviced by hyperscalers. A critical challenge of training RecSys is its high memory capacity requirements, reaching hundreds of GBs to TBs of model size. In RecSys, the so-called embedding layers account for the majority of memory usage so current systems employ a hybrid CPU-GPU design to have the large CPU memory store the memory hungry embedding layers. Unfortunately, training embeddings involve several memory bandwidth intensive operations which is at odds with the slow CPU memory, causing performance overheads. Prior work proposed to cache frequently accessed embeddings inside GPU memory as means to filter down the embedding layer traffic to CPU memory, but this paper observes several limitations with such cache design. In this work, we present a fundamentally different approach in designing embedding caches for RecSys. Our proposed ScratchPipe architecture utilizes unique properties of RecSys training to develop an embedding cache that not only sees the past but also the "future" cache accesses. ScratchPipe exploits such property to guarantee that the active working set of embedding layers can "always" be captured inside our proposed cache design, enabling embedding layer training to be conducted at GPU memory speed.

arXiv:2205.04067공식 코드 · Papers with Code

Sub-Word Alignment Is Still Useful: A Vest-Pocket Method for Enhancing Low-Resource Machine Translation

We leverage embedding duplication between aligned sub-words to extend the Parent-Child transfer learning method, so as to improve low-resource machine translation. We conduct experiments on benchmark datasets of My-En, Id-En and Tr-En translation scenarios. The test results show that our method produces substantial improvements, achieving the BLEU scores of 22.5, 28.0 and 18.1 respectively. In addition, the method is computationally efficient which reduces the consumption of training time by 63.8%, reaching the duration of 1.6 hours when training on a Tesla 16GB P100 GPU. All the models and source codes in the experiments will be made publicly available to support reproducible research.

arXiv:2205.04295

A modular software framework for the design and implementation of ptychography algorithms

Computational methods are driving high impact microscopy techniques such as ptychography. However, the design and implementation of new algorithms is often a laborious process, as many parts of the code are written in close-to-the-hardware programming constructs to speed up the reconstruction. In this paper, we present SciComPty, a new ptychography software framework aiming at simulating ptychography datasets and testing state-of-the-art and new reconstruction algorithms. Despite its simplicity, the software leverages GPU accelerated processing through the PyTorch CUDA interface. This is essential to design new methods that can readily be employed. As an example, we present an improved position refinement method based on Adam and a new version of the rPIE algorithm, adapted for partial coherence setups. Results are shown on both synthetic and real datasets. The software is released as open-source.

arXiv:2205.03133공식 코드 · Papers with Code

BDIS: Bayesian Dense Inverse Searching Method for Real-Time Stereo Surgical Image Matching

In stereoscope-based Minimally Invasive Surgeries (MIS), dense stereo matching plays an indispensable role in 3D shape recovery, AR, VR, and navigation tasks. Although numerous Deep Neural Network (DNN) approaches are proposed, the conventional prior-free approaches are still popular in the industry because of the lack of open-source annotated data set and the limitation of the task-specific pre-trained DNNs. Among the prior-free stereo matching algorithms, there is no successful real-time algorithm in none GPU environment for MIS. This paper proposes the first CPU-level real-time prior-free stereo matching algorithm for general MIS tasks. We achieve an average 17 Hz on 640*480 images with a single-core CPU (i5-9400) for surgical images. Meanwhile, it achieves slightly better accuracy than the popular ELAS. The patch-based fast disparity searching algorithm is adopted for the rectified stereo images. A coarse-to-fine Bayesian probability and a spatial Gaussian mixed model were proposed to evaluate the patch probability at different scales. An optional probability density function estimation algorithm was adopted to quantify the prediction variance. Extensive experiments demonstrated the proposed method's capability to handle ambiguities introduced by the textureless surfaces and the photometric inconsistency from the non-Lambertian reflectance and dark illumination. The estimated probability managed to balance the confidences of the patches for stereo images at different scales. It has similar or higher accuracy and fewer outliers than the baseline ELAS in MIS, while it is 4-5 times faster. The code and the synthetic data sets are available at https://github.com/JingweiSong/BDIS-v2.

arXiv:2205.02646공식 코드 · Papers with Code

Fast Reconstruction of Three-Quarter Sampling Measurements Using Recurrent Local Joint Sparse Deconvolution and Extrapolation

Recently, non-regular three-quarter sampling has shown to deliver an increased image quality of image sensors by using differently oriented L-shaped pixels compared to the same number of square pixels. A three-quarter sampling sensor can be understood as a conventional low-resolution sensor where one quadrant of each square pixel is opaque. Subsequent to the measurement, the data can be reconstructed on a regular grid with twice the resolution in both spatial dimensions using an appropriate reconstruction algorithm. For this reconstruction, local joint sparse deconvolution and extrapolation (L-JSDE) has shown to perform very well. As a disadvantage, L-JSDE requires long computation times of several dozen minutes per megapixel. In this paper, we propose a faster version of L-JSDE called recurrent L-JSDE (RL-JSDE) which is a reformulation of L-JSDE. For reasonable recurrent measurement patterns, RL-JSDE provides significant speedups on both CPU and GPU without sacrificing image quality. Compared to L-JSDE, 20-fold and 733-fold speedups are achieved on CPU and GPU, respectively.

arXiv:2205.01906

ASE: Large-Scale Reusable Adversarial Skill Embeddings for Physically Simulated Characters

The incredible feats of athleticism demonstrated by humans are made possible in part by a vast repertoire of general-purpose motor skills, acquired through years of practice and experience. These skills not only enable humans to perform complex tasks, but also provide powerful priors for guiding their behaviors when learning new tasks. This is in stark contrast to what is common practice in physics-based character animation, where control policies are most typically trained from scratch for each task. In this work, we present a large-scale data-driven framework for learning versatile and reusable skill embeddings for physically simulated characters. Our approach combines techniques from adversarial imitation learning and unsupervised reinforcement learning to develop skill embeddings that produce life-like behaviors, while also providing an easy to control representation for use on new downstream tasks. Our models can be trained using large datasets of unstructured motion clips, without requiring any task-specific annotation or segmentation of the motion data. By leveraging a massively parallel GPU-based simulator, we are able to train skill embeddings using over a decade of simulated experiences, enabling our model to learn a rich and versatile repertoire of skills. We show that a single pre-trained model can be effectively applied to perform a diverse set of new tasks. Our system also allows users to specify tasks through simple reward functions, and the skill embedding then enables the character to automatically synthesize complex and naturalistic strategies in order to achieve the task objectives.

arXiv:2205.01721공식 코드 · Papers with Code

In Defense of Image Pre-Training for Spatiotemporal Recognition

Image pre-training, the current de-facto paradigm for a wide range of visual tasks, is generally less favored in the field of video recognition. By contrast, a common strategy is to directly train with spatiotemporal convolutional neural networks (CNNs) from scratch. Nonetheless, interestingly, by taking a closer look at these from-scratch learned CNNs, we note there exist certain 3D kernels that exhibit much stronger appearance modeling ability than others, arguably suggesting appearance information is already well disentangled in learning. Inspired by this observation, we hypothesize that the key to effectively leveraging image pre-training lies in the decomposition of learning spatial and temporal features, and revisiting image pre-training as the appearance prior to initializing 3D kernels. In addition, we propose Spatial-Temporal Separable (STS) convolution, which explicitly splits the feature channels into spatial and temporal groups, to further enable a more thorough decomposition of spatiotemporal features for fine-tuning 3D CNNs. Our experiments show that simply replacing 3D convolution with STS notably improves a wide range of 3D CNNs without increasing parameters and computation on both Kinetics-400 and Something-Something V2. Moreover, this new training pipeline consistently achieves better results on video recognition with significant speedup. For instance, we achieve +0.6% top-1 of Slowfast on Kinetics-400 over the strong 256-epoch 128-GPU baseline while fine-tuning for only 50 epochs with 4 GPUs. The code and models are available at https://github.com/UCSC-VLAA/Image-Pretraining-for-Video.

arXiv:2205.01679

Physics to the Rescue: Deep Non-line-of-sight Reconstruction for High-speed Imaging

Computational approach to imaging around the corner, or non-line-of-sight (NLOS) imaging, is becoming a reality thanks to major advances in imaging hardware and reconstruction algorithms. A recent development towards practical NLOS imaging, Nam et al. demonstrated a high-speed non-confocal imaging system that operates at 5Hz, 100x faster than the prior art. This enormous gain in acquisition rate, however, necessitates numerous approximations in light transport, breaking many existing NLOS reconstruction methods that assume an idealized image formation model. To bridge the gap, we present a novel deep model that incorporates the complementary physics priors of wave propagation and volume rendering into a neural network for high-quality and robust NLOS reconstruction. This orchestrated design regularizes the solution space by relaxing the image formation model, resulting in a deep model that generalizes well on real captures despite being exclusively trained on synthetic data. Further, we devise a unified learning framework that enables our model to be flexibly trained using diverse supervision signals, including target intensity images or even raw NLOS transient measurements. Once trained, our model renders both intensity and depth images at inference time in a single forward pass, capable of processing more than 5 captures per second on a high-end GPU. Through extensive qualitative and quantitative experiments, we show that our method outperforms prior physics and learning based approaches on both synthetic and real measurements. We anticipate that our method along with the fast capturing system will accelerate future development of NLOS imaging for real world applications that require high-speed imaging.

arXiv:2205.00263공식 코드 · Papers with Code

Complete Verification via Multi-Neuron Relaxation Guided Branch-and-Bound

State-of-the-art neural network verifiers are fundamentally based on one of two paradigms: either encoding the whole verification problem via tight multi-neuron convex relaxations or applying a Branch-and-Bound (BaB) procedure leveraging imprecise but fast bounding methods on a large number of easier subproblems. The former can capture complex multi-neuron dependencies but sacrifices completeness due to the inherent limitations of convex relaxations. The latter enables complete verification but becomes increasingly ineffective on larger and more challenging networks. In this work, we present a novel complete verifier which combines the strengths of both paradigms: it leverages multi-neuron relaxations to drastically reduce the number of subproblems generated during the BaB process and an efficient GPU-based dual optimizer to solve the remaining ones. An extensive evaluation demonstrates that our verifier achieves a new state-of-the-art on both established benchmarks as well as networks with significantly higher accuracy than previously considered. The latter result (up to 28% certification gains) indicates meaningful progress towards creating verifiers that can handle practically relevant networks.