arXiv subjects

Hui Guan

Publications and source records attributed to Hui Guan.

46 records · Page 3Linked to original sources

arXiv2022

A Tree-Structured Multi-Task Model Recommender

Tree-structured multi-task architectures have been employed to jointly tackle multiple vision tasks in the context of multi-task learning (MTL). The major challenge is to determine where to branch out for each task given a backbone model to optimize for both task accuracy and computation efficiency. To address the challenge, this paper proposes a recommender that, given a set of tasks and a convolutional neural network-based backbone model, automatically suggests tree-structured multi-task architectures that could achieve a high task performance while meeting a user-specified computation budget without performing model training. Extensive evaluations on popular MTL benchmarks show that the recommended architectures could achieve competitive task accuracy and computation efficiency compared with state-of-the-art MTL methods. Our tree-structured multi-task model recommender is open-sourced and available at https://github.com/zhanglijun95/TreeMTL.

cs.LG↗

arXiv2022

Rethinking Hard-Parameter Sharing in Multi-Domain Learning

Hard parameter sharing in multi-domain learning (MDL) allows domains to share some of the model parameters to reduce storage cost while improving prediction accuracy. One common sharing practice is to share the bottom layers of a deep neural network among domains while using separate top layers for each domain. In this work, we revisit this common practice via an empirical study on image classification tasks from a diverse set of visual domains and make two surprising observations. (1) Using separate bottom-layer parameters could achieve significantly better performance than the common practice and this phenomenon holds with different experimental settings. (2) A multi-domain model with a small proportion of domain-specific parameters from bottom layers can achieve competitive performance with independent models trained on each domain separately. Our observations suggest that people adopt the new strategy of using separate bottom-layer parameters as a stronger baseline for model design in MDL.

cs.LG↗

arXiv2021

Toward Compact Parameter Representations for Architecture-Agnostic Neural Network Compression

This paper investigates deep neural network (DNN) compression from the perspective of compactly representing and storing trained parameters. We explore the previously overlooked opportunity of cross-layer architecture-agnostic representation sharing for DNN parameters. To do this, we decouple feedforward parameters from DNN architectures and leverage additive quantization, an extreme lossy compression method invented for image descriptors, to compactly represent the parameters. The representations are then finetuned on task objectives to improve task accuracy. We conduct extensive experiments on MobileNet-v2, VGG-11, ResNet-50, Feature Pyramid Networks, and pruned DNNs trained for classification, detection, and segmentation tasks. The conceptually simple scheme consistently outperforms iterative unstructured pruning. Applied to ResNet-50 with 76.1% top-1 accuracy on the ILSVRC12 classification challenge, it achieves a $7.2\times$ compression ratio with no accuracy loss and a $15.3\times$ compression ratio at 74.79% accuracy. Further analyses suggest that representation sharing can frequently happen across network layers and that learning shared representations for an entire DNN can achieve better accuracy at the same compression ratio than compressing the model as multiple separate parts. We release PyTorch code to facilitate DNN deployment on resource-constrained devices and spur future research on efficient representations and storage of DNN parameters.

cs.CV↗

arXiv2021

COMET: A Novel Memory-Efficient Deep Learning Training Framework by Using Error-Bounded Lossy Compression

Training wide and deep neural networks (DNNs) require large amounts of storage resources such as memory because the intermediate activation data must be saved in the memory during forward propagation and then restored for backward propagation. However, state-of-the-art accelerators such as GPUs are only equipped with very limited memory capacities due to hardware design constraints, which significantly limits the maximum batch size and hence performance speedup when training large-scale DNNs. Traditional memory saving techniques either suffer from performance overhead or are constrained by limited interconnect bandwidth or specific interconnect technology. In this paper, we propose a novel memory-efficient CNN training framework (called COMET) that leverages error-bounded lossy compression to significantly reduce the memory requirement for training, to allow training larger models or to accelerate training. Different from the state-of-the-art solutions that adopt image-based lossy compressors (such as JPEG) to compress the activation data, our framework purposely adopts error-bounded lossy compression with a strict error-controlling mechanism. Specifically, we perform a theoretical analysis on the compression error propagation from the altered activation data to the gradients, and empirically investigate the impact of altered gradients over the training process. Based on these analyses, we optimize the error-bounded lossy compression and propose an adaptive error-bound control scheme for activation data compression. We evaluate our design against state-of-the-art solutions with five widely-adopted CNNs and ImageNet dataset. Experiments demonstrate that our proposed framework can significantly reduce the training memory consumption by up to 13.5X over the baseline training and 1.8X over another state-of-the-art compression-based framework, respectively, with little or no accuracy loss.

cs.AI↗

arXiv2021

Scalable Graph Neural Network Training: The Case for Sampling

Graph Neural Networks (GNNs) are a new and increasingly popular family of deep neural network architectures to perform learning on graphs. Training them efficiently is challenging due to the irregular nature of graph data. The problem becomes even more challenging when scaling to large graphs that exceed the capacity of single devices. Standard approaches to distributed DNN training, such as data and model parallelism, do not directly apply to GNNs. Instead, two different approaches have emerged in the literature: whole-graph and sample-based training. In this paper, we review and compare the two approaches. Scalability is challenging with both approaches, but we make a case that research should focus on sample-based training since it is a more promising approach. Finally, we review recent systems supporting sample-based training.

cs.LG↗

arXiv2021

NumaPerf: Predictive and Full NUMA Profiling

Parallel applications are extremely challenging to achieve the optimal performance on the NUMA architecture, which necessitates the assistance of profiling tools. However, existing NUMA-profiling tools share some similar shortcomings, such as portability, effectiveness, and helpfulness issues. This paper proposes a novel profiling tool - NumaPerf - that overcomes these issues. NumaPerf aims to identify potential performance issues for any NUMA architecture, instead of only on the current hardware. To achieve this, NumaPerf focuses on memory sharing patterns between threads, instead of real remote accesses. NumaPerf further detects potential thread migrations and load imbalance issues that could significantly affect the performance but are omitted by existing profilers. NumaPerf also separates cache coherence issues that may require different fix strategies. Based on our extensive evaluation, NumaPerf is able to identify more performance issues than any existing tool, while fixing these bugs leads to up to 5.94x performance speedup.

cs.PF↗

arXiv2020

SID-NISM: A Self-supervised Low-light Image Enhancement Framework

When capturing images in low-light conditions, the images often suffer from low visibility, which not only degrades the visual aesthetics of images, but also significantly degenerates the performance of many computer vision algorithms. In this paper, we propose a self-supervised low-light image enhancement framework (SID-NISM), which consists of two components, a Self-supervised Image Decomposition Network (SID-Net) and a Nonlinear Illumination Saturation Mapping function (NISM). As a self-supervised network, SID-Net could decompose the given low-light image into its reflectance, illumination and noise directly without any prior training or reference image, which distinguishes it from existing supervised-learning methods greatly. Then, the decomposed illumination map will be enhanced by NISM. Having the restored illumination map, the enhancement can be achieved accordingly. Experiments on several public challenging low-light image datasets reveal that the images enhanced by SID-NISM are more natural and have less unexpected artifacts.

cs.CV↗

arXiv2019

Post-Training 4-bit Quantization on Embedding Tables

Continuous representations have been widely adopted in recommender systems where a large number of entities are represented using embedding vectors. As the cardinality of the entities increases, the embedding components can easily contain millions of parameters and become the bottleneck in both storage and inference due to large memory consumption. This work focuses on post-training 4-bit quantization on the continuous embeddings. We propose row-wise uniform quantization with greedy search and codebook-based quantization that consistently outperforms state-of-the-art quantization approaches on reducing accuracy degradation. We deploy our uniform quantization technique on a production model in Facebook and demonstrate that it can reduce the model size to only 13.89% of the single-precision version while the model quality stays neutral.

cs.LG↗

arXiv2019

In-Place Zero-Space Memory Protection for CNN

Convolutional Neural Networks (CNN) are being actively explored for safety-critical applications such as autonomous vehicles and aerospace, where it is essential to ensure the reliability of inference results in the presence of possible memory faults. Traditional methods such as error correction codes (ECC) and Triple Modular Redundancy (TMR) are CNN-oblivious and incur substantial memory overhead and energy cost. This paper introduces in-place zero-space ECC assisted with a new training scheme weight distribution-oriented training. The new method provides the first known zero space cost memory protection for CNNs without compromising the reliability offered by traditional ECC.

cs.LG↗

arXiv2017

First Study on Data Readiness Level

We introduce the idea of Data Readiness Level (DRL) to measure the relative richness of data to answer specific questions often encountered by data scientists. We first approach the problem in its full generality explaining its desired mathematical properties and applications and then we propose and study two DRL metrics. Specifically, we define DRL as a function of at least four properties of data: Noisiness, Believability, Relevance, and Coherence. The information-theoretic based metrics, Cosine Similarity and Document Disparity, are proposed as indicators of Relevance and Coherence for a piece of data. The proposed metrics are validated through a text-based experiment using Twitter data.

cs.IR↗