arXiv subjects

Yue Wang

Publications and source records attributed to Yue Wang.

At least 793 records · Page 44Linked to original sources

arXiv2020

Multi-Frame to Single-Frame: Knowledge Distillation for 3D Object Detection

A common dilemma in 3D object detection for autonomous driving is that high-quality, dense point clouds are only available during training, but not testing. We use knowledge distillation to bridge the gap between a model trained on high-quality inputs at training time and another tested on low-quality inputs at inference time. In particular, we design a two-stage training pipeline for point cloud object detection. First, we train an object detection model on dense point clouds, which are generated from multiple frames using extra information only available at training time. Then, we train the model's identical counterpart on sparse single-frame point clouds with consistency regularization on features from both models. We show that this procedure improves performance on low-quality data during testing, without additional overhead.

cs.CV↗

arXiv2020

Multimodal Joint Attribute Prediction and Value Extraction for E-commerce Product

Product attribute values are essential in many e-commerce scenarios, such as customer service robots, product recommendations, and product retrieval. While in the real world, the attribute values of a product are usually incomplete and vary over time, which greatly hinders the practical applications. In this paper, we propose a multimodal method to jointly predict product attributes and extract values from textual product descriptions with the help of the product images. We argue that product attributes and values are highly correlated, e.g., it will be easier to extract the values on condition that the product attributes are given. Thus, we jointly model the attribute prediction and value extraction tasks from multiple aspects towards the interactions between attributes and values. Moreover, product images have distinct effects on our tasks for different product attributes and values. Thus, we selectively draw useful visual information from product images to enhance our model. We annotate a multimodal product attribute value dataset that contains 87,194 instances, and the experimental results on this dataset demonstrate that explicitly modeling the relationship between attributes and values facilitates our method to establish the correspondence between them, and selectively utilizing visual product information is necessary for the task. Our code and dataset will be released to the public.

cs.CL↗

arXiv2020

NPENAS: Neural Predictor Guided Evolution for Neural Architecture Search

Neural architecture search (NAS) is a promising method for automatically design neural architectures. NAS adopts a search strategy to explore the predefined search space to find outstanding performance architecture with the minimum searching costs. Bayesian optimization and evolutionary algorithms are two commonly used search strategies, but they suffer from computationally expensive, challenge to implement or inefficient exploration ability. In this paper, we propose a neural predictor guided evolutionary algorithm to enhance the exploration ability of EA for NAS (NPENAS) and design two kinds of neural predictors. The first predictor is defined from Bayesian optimization and we propose a graph-based uncertainty estimation network as a surrogate model that is easy to implement and computationally efficient. The second predictor is a graph-based neural network that directly outputs the performance prediction of the input neural architecture. The NPENAS using the two neural predictors are denoted as NPENAS-BO and NPENAS-NP respectively. In addition, we introduce a new random architecture sampling method to overcome the drawbacks of the existing sampling method. Extensive experiments demonstrate the superiority of NPENAS. Quantitative results on three NAS search spaces indicate that both NPENAS-BO and NPENAS-NP outperform most existing NAS algorithms, with NPENAS-BO achieving state-of-the-art performance on NASBench-201 and NPENAS-NP on NASBench-101 and DARTS, respectively.

cs.LG↗

arXiv2020

Attachment and antibiotic response of early-stage biofilms studied using resonant hyperspectral imaging

Many bacterial species readily develop biofilms that act as a protective matrix against external challenge, e.g. from antimicrobial treatment. Therefore, biofilms are often responsible for persistent and recurring infections. Established methods for studying biofilms are either destructive or they focus on the biofilm surface. A non-destructive method that is sensitive to the underside of the biofilm is highly desirable, as it allows studying the penetration of antibiotics through the film. Here, we demonstrate that the high surface sensitivity of resonant hyperspectral imaging provides this capability. The method allows us to monitor the early stages of Escherichia coli biofilm formation, cell attachment and microcolony formation, in-situ and in real time. We study the response of the biofilm to a number of different antibiotics and verify our observations using confocal microscopy. Based on this ability to closely monitor the surface-bound cells, resonant hyperspectral imaging gives new insights into the antimicrobial resistance of biofilms.

physics.bio-ph↗

arXiv2020

DDSL: Efficient Subgraph Listing on Distributed and Dynamic Graphs

Subgraph listing is a fundamental problem in graph theory and has wide applications in areas like sociology, chemistry, and social networks. Modern graphs can usually be large-scale as well as highly dynamic, which challenges the efficiency of existing subgraph listing algorithms. Recent works have shown the benefits of partitioning and processing big graphs in a distributed system, however, there is only few work targets subgraph listing on dynamic graphs in a distributed environment. In this paper, we propose an efficient approach, called Distributed and Dynamic Subgraph Listing (DDSL), which can incrementally update the results instead of running from scratch. DDSL follows a general distributed join framework. In this framework, we use a Neighbor-Preserved storage for data graphs, which takes bounded extra space and supports dynamic updating. After that, we propose a comprehensive cost model to estimate the I/O cost of listing subgraphs. Then based on this cost model, we develop an algorithm to find the optimal join tree for a given pattern. To handle dynamic graphs, we propose an efficient left-deep join algorithm to incrementally update the join results. Extensive experiments are conducted on real-world datasets. The results show that DDSL outperforms existing methods in dealing with both static dynamic graphs in terms of the responding time.

cs.DB↗

arXiv2020

Low Complexity Trellis-Coded Quantization in Versatile Video Coding

The forthcoming Versatile Video Coding (VVC) standard adopts the trellis-coded quantization, which leverages the delicate trellis graph to map the quantization candidates within one block into the optimal path. Despite the high compression efficiency, the complex trellis search with soft decision quantization may hinder the applications due to high complexity and low throughput capacity. To reduce the complexity, in this paper, we propose a low complexity trellis-coded quantization scheme in a scientifically sound way with theoretical modeling of the rate and distortion. As such, the trellis departure point can be adaptively adjusted, and unnecessarily visited branches are accordingly pruned, leading to the shrink of total trellis stages and simplification of transition branches. Extensive experimental results on the VVC test model show that the proposed scheme is effective in reducing the encoding complexity by 11% and 5% with all intra and random access configurations, respectively, at the cost of only 0.11% and 0.05% BD-Rate increase. Meanwhile, on average 24% and 27% quantization time savings can be achieved under all intra and random access configurations. Due to the excellent performance, the VVC test model has adopted one implementation of the proposed scheme.

cs.MM↗

arXiv2020

Enhancing the Erdős-Lovász Tihany Conjecture for graphs with independence number two

Let $s\ge2$ and $t\ge2$ be integers. A graph $G$ is $(s,t)$-\emph{splittable} if $V(G)$ can be partitioned into two sets $S$ and $T$ such that $χ(G[S])\geq s$ and $χ(G[T])\geq t$. The well-known Erdős-Lovász Tihany Conjecture from 1968 states that every graph $G$ whose chromatic number $χ(G)=s+t-1$ is more than its clique number $ω(G)$ is $(s,t)$-splittable. In this paper, we prove an enhanced version of the Erdős-Lovász Tihany Conjecture for graphs with independence number two. That is, for every graph $G$ with $χ(G)=s+t-1>ω(G)+1$ is $(s,t+1)$-splittable. There are examples showing that this result is best possible.

math.CO↗

arXiv2020

Three solutions for a new Kirchhoff-type problem

This article concerns on the existence of multiple solutions for a new Kirchhoff-type problem with negative modulus. We prove that there exist three nontrivial solutions when the parameter is enough small via the variational methods and algebraic analysis. Moreover, our fundamental technique is one of the Mountain Pass Lemma, Ekeland variational principle, and Minimax principle.

math.AP↗

arXiv2020

Pillar-based Object Detection for Autonomous Driving

We present a simple and flexible object detection framework optimized for autonomous driving. Building on the observation that point clouds in this application are extremely sparse, we propose a practical pillar-based approach to fix the imbalance issue caused by anchors. In particular, our algorithm incorporates a cylindrical projection into multi-view feature learning, predicts bounding box parameters per pillar rather than per point or per anchor, and includes an aligned pillar-to-point projection module to improve the final prediction. Our anchor-free approach avoids hyperparameter search associated with past methods, simplifying 3D object detection while significantly improving upon state-of-the-art.

cs.CV↗

arXiv2020

Miniaturization of the Superconducting Memory Cell via a Three-Dimensional Nb Nano-Superconducting Quantum Interference Device

Scalable memories that can match the speeds of superconducting logic circuits have long been desired to enable a superconducting computer. A superconducting loop that includes a Josephson junction can store a flux quantum state in picoseconds. However, the requirement for the loop inductance to create a bi-state hysteresis sets a limit on the minimal area occupied by a single memory cell. Here, we present a miniaturized superconducting memory cell based on a Three-Dimensional (3D) Nb nano-Superconducting QUantum Interference Device (nano-SQUID). The major cell area here fits within an 8*9 μm^2 rectangle with a cross-selected function for memory implementation. The cell shows periodic tunable hysteresis between two neighbouring flux quantum states produced by bias current sweeping because of the large modulation depth of the 3D nano-SQUID (~66%). Furthermore, the measured Current-Phase Relations (CPRs) of nano-SQUIDs are shown to be skewed from a sine function, as predicted by theoretical modelling. The skewness and the critical current of 3D nano-SQUIDs are linearly correlated. It is also found that the hysteresis loop size is in a linear scaling relationship with the CPR skewness using the statistics from characterisation of 26 devices. We show that the CPR skewness range of π/4~3π/4 is equivalent to a large loop inductance in creating a stable bi-state hysteresis for memory implementation. Therefore, the skewed CPR of 3D nano-SQUID enables further superconducting memory cell miniaturization by overcoming the inductance limitation of the loop area.

cond-mat.mes-hall↗

arXiv2020

Constructing Basis Path Set by Eliminating Path Dependency

The way the basis path set works in neural network remains mysterious, and the generalization of newly appeared G-SGD algorithm to more practical network is hindered. The Basis Path Set Searching problem is formulated from the perspective of graph theory, to find the basis path set in a regular complicated neural network. Our paper aims to discover the underlying cause of path dependency between two independent substructures. Algorithm DEAH is designed to solve the Basis Path Set Searching problem by eliminating such path dependency. The path subdivision chain is proposed to effectively eliminate the path dependency inside the chain and between chains. The theoretical proofs and analysis of polynomial time complexity are presented. The paper therefore provides one methodology to find the basis path set in a more general neural network, which offers theoretical and algorithmic support for the application of G-SGD algorithm in more practical scenarios.

cs.DS↗

arXiv2020

A Novel Cascade Binary Tagging Framework for Relational Triple Extraction

Extracting relational triples from unstructured text is crucial for large-scale knowledge graph construction. However, few existing works excel in solving the overlapping triple problem where multiple relational triples in the same sentence share the same entities. In this work, we introduce a fresh perspective to revisit the relational triple extraction task and propose a novel cascade binary tagging framework (CasRel) derived from a principled problem formulation. Instead of treating relations as discrete labels as in previous works, our new framework models relations as functions that map subjects to objects in a sentence, which naturally handles the overlapping problem. Experiments show that the CasRel framework already outperforms state-of-the-art methods even when its encoder module uses a randomly initialized BERT encoder, showing the power of the new tagging framework. It enjoys further performance boost when employing a pre-trained BERT encoder, outperforming the strongest baseline by 17.5 and 30.2 absolute gain in F1-score on two public datasets NYT and WebNLG, respectively. In-depth analysis on different scenarios of overlapping triples shows that the method delivers consistent performance gain across all these scenarios. The source code and data are released online.

cs.CL↗

arXiv2020

Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need?

The focus of recent meta-learning research has been on the development of learning algorithms that can quickly adapt to test time tasks with limited data and low computational cost. Few-shot learning is widely used as one of the standard benchmarks in meta-learning. In this work, we show that a simple baseline: learning a supervised or self-supervised representation on the meta-training set, followed by training a linear classifier on top of this representation, outperforms state-of-the-art few-shot learning methods. An additional boost can be achieved through the use of self-distillation. This demonstrates that using a good learned embedding model can be more effective than sophisticated meta-learning algorithms. We believe that our findings motivate a rethinking of few-shot image classification benchmarks and the associated role of meta-learning algorithms. Code is available at: http://github.com/WangYueFt/rfs/.

cs.CV↗

arXiv2020

AutoDNNchip: An Automated DNN Chip Predictor and Builder for Both FPGAs and ASICs

Recent breakthroughs in Deep Neural Networks (DNNs) have fueled a growing demand for DNN chips. However, designing DNN chips is non-trivial because: (1) mainstream DNNs have millions of parameters and operations; (2) the large design space due to the numerous design choices of dataflows, processing elements, memory hierarchy, etc.; and (3) an algorithm/hardware co-design is needed to allow the same DNN functionality to have a different decomposition, which would require different hardware IPs to meet the application specifications. Therefore, DNN chips take a long time to design and require cross-disciplinary experts. To enable fast and effective DNN chip design, we propose AutoDNNchip - a DNN chip generator that can automatically generate both FPGA- and ASIC-based DNN chip implementation given DNNs from machine learning frameworks (e.g., PyTorch) for a designated application and dataset. Specifically, AutoDNNchip consists of two integrated enablers: (1) a Chip Predictor, built on top of a graph-based accelerator representation, which can accurately and efficiently predict a DNN accelerator's energy, throughput, and area based on the DNN model parameters, hardware configuration, technology-based IPs, and platform constraints; and (2) a Chip Builder, which can automatically explore the design space of DNN chips (including IP selection, block configuration, resource balancing, etc.), optimize chip design via the Chip Predictor, and then generate optimized synthesizable RTL to achieve the target design metrics. Experimental results show that our Chip Predictor's predicted performance differs from real-measured ones by < 10% when validated using 15 DNN models and 4 platforms (edge-FPGA/TPU/GPU and ASIC). Furthermore, accelerators generated by our AutoDNNchip can achieve better (up to 3.86X improvement) performance than that of expert-crafted state-of-the-art accelerators.

cs.DC↗

arXiv2020

Finite-sample Analysis of Greedy-GQ with Linear Function Approximation under Markovian Noise

Greedy-GQ is an off-policy two timescale algorithm for optimal control in reinforcement learning. This paper develops the first finite-sample analysis for the Greedy-GQ algorithm with linear function approximation under Markovian noise. Our finite-sample analysis provides theoretical justification for choosing stepsizes for this two timescale algorithm for faster convergence in practice, and suggests a trade-off between the convergence rate and the quality of the obtained policy. Our paper extends the finite-sample analyses of two timescale reinforcement learning algorithms from policy evaluation to optimal control, which is of more practical interest. Specifically, in contrast to existing finite-sample analyses for two timescale methods, e.g., GTD, GTD2 and TDC, where their objective functions are convex, the objective function of the Greedy-GQ algorithm is non-convex. Moreover, the Greedy-GQ algorithm is also not a linear two-timescale stochastic approximation algorithm. Our techniques in this paper provide a general framework for finite-sample analysis of non-convex value-based reinforcement learning algorithms for optimal control.

cs.LG↗

arXiv2020

User-generated Video Quality Assessment: A Subjective and Objective Study

Recently, we have observed an exponential increase of user-generated content (UGC) videos. The distinguished characteristic of UGC videos originates from the video production and delivery chain, as they are usually acquired and processed by non-professional users before uploading to the hosting platforms for sharing. As such, these videos usually undergo multiple distortion stages that may affect visual quality before ultimately being viewed. Inspired by the increasing consensus that the optimization of the video coding and processing shall be fully driven by the perceptual quality, in this paper, we propose to study the quality of the UGC videos from both objective and subjective perspectives. We first construct a UGC video quality assessment (VQA) database, aiming to provide useful guidance for the UGC video coding and processing in the hosting platform. The database contains source UGC videos uploaded to the platform and their transcoded versions that are ultimately enjoyed by end-users, along with their subjective scores. Furthermore, we develop an objective quality assessment algorithm that automatically evaluates the quality of the transcoded videos based on the corrupted reference, which is in accordance with the application scenarios of UGC video sharing in the hosting platforms. The information from the corrupted reference is well leveraged and the quality is predicted based on the inferred quality maps with deep neural networks (DNN). Experimental results show that the proposed method yields superior performance. Both subjective and objective evaluations of the UGC videos also shed lights on the design of perceptual UGC video coding.

cs.MM↗

arXiv2020

A New MRAM-based Process In-Memory Accelerator for Efficient Neural Network Training with Floating Point Precision

The excellent performance of modern deep neural networks (DNNs) comes at an often prohibitive training cost, limiting the rapid development of DNN innovations and raising various environmental concerns. To reduce the dominant data movement cost of training, process in-memory (PIM) has emerged as a promising solution as it alleviates the need to access DNN weights. However, state-of-the-art PIM DNN training accelerators employ either analog/mixed signal computing which has limited precision or digital computing based on a memory technology that supports limited logic functions and thus requires complicated procedure to realize floating point computation. In this paper, we propose a spin orbit torque magnetic random access memory (SOT-MRAM) based digital PIM accelerator that supports floating point precision. Specifically, this new accelerator features an innovative (1) SOT-MRAM cell, (2) full addition design, and (3) floating point computation. Experiment results show that the proposed SOT-MRAM PIM based DNN training accelerator can achieve 3.3$\times$, 1.8$\times$, and 2.5$\times$ improvement in terms of energy, latency, and area, respectively, compared with a state-of-the-art PIM based DNN training accelerator.

cs.DC↗

arXiv2020

SmartExchange: Trading Higher-cost Memory Storage/Access for Lower-cost Computation

We present SmartExchange, an algorithm-hardware co-design framework to trade higher-cost memory storage/access for lower-cost computation, for energy-efficient inference of deep neural networks (DNNs). We develop a novel algorithm to enforce a specially favorable DNN weight structure, where each layerwise weight matrix can be stored as the product of a small basis matrix and a large sparse coefficient matrix whose non-zero elements are all power-of-2. To our best knowledge, this algorithm is the first formulation that integrates three mainstream model compression ideas: sparsification or pruning, decomposition, and quantization, into one unified framework. The resulting sparse and readily-quantized DNN thus enjoys greatly reduced energy consumption in data movement as well as weight storage. On top of that, we further design a dedicated accelerator to fully utilize the SmartExchange-enforced weights to improve both energy efficiency and latency performance. Extensive experiments show that 1) on the algorithm level, SmartExchange outperforms state-of-the-art compression techniques, including merely sparsification or pruning, decomposition, and quantization, in various ablation studies based on nine DNN models and four datasets; and 2) on the hardware level, the proposed SmartExchange based accelerator can improve the energy efficiency by up to 6.7$\times$ and the speedup by up to 19.2$\times$ over four state-of-the-art DNN accelerators, when benchmarked on seven DNN models (including four standard DNNs, two compact DNN models, and one segmentation model) and three datasets.

cs.LG↗