arXiv subjects

Yan Gao

Publications and source records attributed to Yan Gao.

At least 181 records · Page 10Linked to original sources

arXiv2022

Match to Win: Analysing Sequences Lengths for Efficient Self-supervised Learning in Speech and Audio

Self-supervised learning (SSL) has proven vital in speech and audio-related applications. The paradigm trains a general model on unlabeled data that can later be used to solve specific downstream tasks. This type of model is costly to train as it requires manipulating long input sequences that can only be handled by powerful centralised servers. Surprisingly, despite many attempts to increase training efficiency through model compression, the effects of truncating input sequence lengths to reduce computation have not been studied. In this paper, we provide the first empirical study of SSL pre-training for different specified sequence lengths and link this to various downstream tasks. We find that training on short sequences can dramatically reduce resource costs while retaining a satisfactory performance for all tasks. This simple one-line change would promote the migration of SSL training from data centres to user-end edge devices for more realistic and personalised applications.

cs.SD↗

arXiv2022

Efficient and Accurate Quantized Image Super-Resolution on Mobile NPUs, Mobile AI & AIM 2022 challenge: Report

Image super-resolution is a common task on mobile and IoT devices, where one often needs to upscale and enhance low-resolution images and video frames. While numerous solutions have been proposed for this problem in the past, they are usually not compatible with low-power mobile NPUs having many computational and memory constraints. In this Mobile AI challenge, we address this problem and propose the participants to design an efficient quantized image super-resolution solution that can demonstrate a real-time performance on mobile NPUs. The participants were provided with the DIV2K dataset and trained INT8 models to do a high-quality 3X image upscaling. The runtime of all models was evaluated on the Synaptics VS680 Smart Home board with a dedicated edge NPU capable of accelerating quantized neural networks. All proposed solutions are fully compatible with the above NPU, demonstrating an up to 60 FPS rate when reconstructing Full HD resolution images. A detailed description of all models developed in the challenge is provided in this paper.

eess.IV↗

arXiv2022

Reasoning Like Program Executors

Reasoning over natural language is a long-standing goal for the research community. However, studies have shown that existing language models are inadequate in reasoning. To address the issue, we present POET, a novel reasoning pre-training paradigm. Through pre-training language models with programs and their execution results, POET empowers language models to harvest the reasoning knowledge possessed by program executors via a data-driven approach. POET is conceptually simple and can be instantiated by different kinds of program executors. In this paper, we showcase two simple instances POET-Math and POET-Logic, in addition to a complex instance, POET-SQL. Experimental results on six benchmarks demonstrate that POET can significantly boost model performance in natural language reasoning, such as numerical reasoning, logical reasoning, and multi-hop reasoning. POET opens a new gate on reasoning-enhancement pre-training, and we hope our analysis would shed light on the future research of reasoning like program executors.

cs.CL↗

arXiv2022

Intrinsic ferromagnetic axion states and a single pair of Weyl fermions in the stable-state Mn\emph{X}$_{2}$\emph{B}$_{2}$\emph{T}$_{6}$-family materials

The intrinsic ferromagnetic (FM) axion insulators and Weyl semimetals (WSMs) with only single pair of Weyl points have drawn intensive attention but so far remain rare and elusive in real materials. Here, we propose a new class of Mn\emph{X}$_{2}$\emph{B}$_{2}$\emph{T}$_{6}$-B (\emph{X}=Ge, Sn, or Pb; \emph{B}=Sb or Bi; \emph{T}=Se or Te) family that is the stable structural form of this system. We find that the Mn\emph{X}$_{2}$\emph{B}$_{2}$\emph{T}$_{6}$-B family has not only the intrinsic FM axion insulators MnGe$_{2}$Bi$_{2}$Te$_{6}$-B, MnSn$_{2}$Bi$_{2}$Te$_{6}$-B, and MnPb$_{2}$Bi$_{2}$Te$_{6}$-B, but also the intrinsic WSM MnSn$_{2}$Sb$_{2}$Te$_{6}$-B with only a single pair of Weyl points. Thus, the Mn\emph{X}$_{2}$\emph{B}$_{2}$\emph{T}$_{6}$-B family can provide an ideal platform to explore the exotic topological magnetoelectric effect and the intrinsic properties related to Weyl points.

cond-mat.mtrl-sci↗

arXiv2022

ZeroFL: Efficient On-Device Training for Federated Learning with Local Sparsity

When the available hardware cannot meet the memory and compute requirements to efficiently train high performing machine learning models, a compromise in either the training quality or the model complexity is needed. In Federated Learning (FL), nodes are orders of magnitude more constrained than traditional server-grade hardware and are often battery powered, severely limiting the sophistication of models that can be trained under this paradigm. While most research has focused on designing better aggregation strategies to improve convergence rates and in alleviating the communication costs of FL, fewer efforts have been devoted to accelerating on-device training. Such stage, which repeats hundreds of times (i.e. every round) and can involve thousands of devices, accounts for the majority of the time required to train federated models and, the totality of the energy consumption at the client side. In this work, we present the first study on the unique aspects that arise when introducing sparsity at training time in FL workloads. We then propose ZeroFL, a framework that relies on highly sparse operations to accelerate on-device training. Models trained with ZeroFL and 95% sparsity achieve up to 2.3% higher accuracy compared to competitive baselines obtained from adapting a state-of-the-art sparse training framework to the FL setting.

cs.LG↗

arXiv2022

Federated Self-supervised Speech Representations: Are We There Yet?

The ubiquity of microphone-enabled devices has lead to large amounts of unlabelled audio data being produced at the edge. The integration of self-supervised learning (SSL) and federated learning (FL) into one coherent system can potentially offer data privacy guarantees while also advancing the quality and robustness of speech representations. In this paper, we provide a first-of-its-kind systematic study of the feasibility and complexities for training speech SSL models under FL scenarios from the perspective of algorithms, hardware, and systems limits. Despite the high potential of their combination, we find existing system constraints and algorithmic behaviour make SSL and FL systems nearly impossible to build today. Yet critically, our results indicate specific performance bottlenecks and research opportunities that would allow this situation to be reversed. While our analysis suggests that, given existing trends in hardware, hybrid SSL and FL speech systems will not be viable until 2027. We believe this study can act as a roadmap to accelerate work towards reaching this milestone much earlier.

cs.SD↗

arXiv2022

Federated Self-supervised Learning for Video Understanding

The ubiquity of camera-enabled mobile devices has lead to large amounts of unlabelled video data being produced at the edge. Although various self-supervised learning (SSL) methods have been proposed to harvest their latent spatio-temporal representations for task-specific training, practical challenges including privacy concerns and communication costs prevent SSL from being deployed at large scales. To mitigate these issues, we propose the use of Federated Learning (FL) to the task of video SSL. In this work, we evaluate the performance of current state-of-the-art (SOTA) video-SSL techniques and identify their shortcomings when integrated into the large-scale FL setting simulated with kinetics-400 dataset. We follow by proposing a novel federated SSL framework for video, dubbed FedVSSL, that integrates different aggregation strategies and partial weight updating. Extensive experiments demonstrate the effectiveness and significance of FedVSSL as it outperforms the centralized SOTA for the downstream retrieval task by 6.66% on UCF-101 and 5.13% on HMDB-51.

cs.CV↗

arXiv2022

Distributed Control for a Multi-Agent System to Pass through a Connected Quadrangle Virtual Tube

In order to guide the multi-agent system in a cluttered environment, a connected quadrangle virtual tube is designed for all agents to keep moving within it, whose basis is called the single trapezoid virtual tube. There is no obstacle inside the tube, namely the area inside the tube can be seen as a safety zone. Then, a distributed swarm controller is proposed for the single trapezoid virtual tube passing problem. This issue is resolved by a gradient vector field method with no local minima. Formal analyses and proofs are made to show that all agents are able to pass the single trapezoid virtual tube. Finally, a modified controller is put forward for convenience in practical use. For the connected quadrangle virtual tube, a modified switching logic is proposed to avoid the deadlock and prevent agents from moving outside the virtual tube. Finally, the effectiveness of the proposed method is validated by numerical simulations and real experiments.

cs.RO↗

arXiv2022

Robust Distributed Control within a Curve Virtual Tube for a Robotic Swarm under Self-Localization Drift and Precise Relative Navigation

To guide the movement of a robotic swarm in a corridor-like environment, a curve virtual tube with no obstacle inside is designed in our previous work. This paper generalizes the controller design to the condition that all robots have self-localization drifts and precise relative navigation, where the flocking algorithm is introduced to reduce the negative impact of the self-localization drift. It is shown that the cohesion behavior and the velocity alignment behavior are able to reduce the influence of the position measurement drift and the velocity measurement error, respectively. For the convenience in practical use, a modified vector field controller with five control terms is put forward. Finally, the effectiveness of the proposed method is validated by numerical simulations and real experiments.

cs.RO↗

arXiv2022

Occluded Video Instance Segmentation: A Benchmark

Can our video understanding systems perceive objects when a heavy occlusion exists in a scene? To answer this question, we collect a large-scale dataset called OVIS for occluded video instance segmentation, that is, to simultaneously detect, segment, and track instances in occluded scenes. OVIS consists of 296k high-quality instance masks from 25 semantic categories, where object occlusions usually occur. While our human vision systems can understand those occluded instances by contextual reasoning and association, our experiments suggest that current video understanding systems cannot. On the OVIS dataset, the highest AP achieved by state-of-the-art algorithms is only 16.3, which reveals that we are still at a nascent stage for understanding objects, instances, and videos in a real-world scenario. We also present a simple plug-and-play module that performs temporal feature calibration to complement missing object cues caused by occlusion. Built upon MaskTrack R-CNN and SipMask, we obtain a remarkable AP improvement on the OVIS dataset. The OVIS dataset and project code are available at http://songbai.site/ovis .

cs.CV↗

arXiv2022

Distributed Control for a Robotic Swarm to Pass through a Curve Virtual Tube

Robotic swarm systems are now becoming increasingly attractive for many challenging applications. The main task for any robot is to reach the destination while keeping a safe separation from other robots and obstacles. In many scenarios, robots need to move within a narrow corridor, through a window or a doorframe. In order to guide all robots to move in a cluttered environment, a curve virtual tube with no obstacle inside is carefully designed in this paper. There is no obstacle inside the tube, namely the area inside the tube can be seen as a safety zone. Then, a distributed swarm controller is proposed with three elaborate control terms: a line approaching term, a robot avoidance term and a tube keeping term. Formal analysis and proofs are made to show that the curve virtual tube passing problem can be solved in a finite time. For the convenience in practical use, a modified controller with an approximate control performance is put forward. Finally, the effectiveness of the proposed method is validated by numerical simulations and real experiments. To show the advantages of the proposed method, the comparison between our method and the control barrier function method is also presented in terms of calculation speed.

cs.RO↗

arXiv2022

NFormer: Robust Person Re-identification with Neighbor Transformer

Person re-identification aims to retrieve persons in highly varying settings across different cameras and scenarios, in which robust and discriminative representation learning is crucial. Most research considers learning representations from single images, ignoring any potential interactions between them. However, due to the high intra-identity variations, ignoring such interactions typically leads to outlier features. To tackle this issue, we propose a Neighbor Transformer Network, or NFormer, which explicitly models interactions across all input images, thus suppressing outlier features and leading to more robust representations overall. As modelling interactions between enormous amount of images is a massive task with lots of distractors, NFormer introduces two novel modules, the Landmark Agent Attention, and the Reciprocal Neighbor Softmax. Specifically, the Landmark Agent Attention efficiently models the relation map between images by a low-rank factorization with a few landmarks in feature space. Moreover, the Reciprocal Neighbor Softmax achieves sparse attention to relevant -- rather than all -- neighbors only, which alleviates interference of irrelevant representations and further relieves the computational burden. In experiments on four large-scale datasets, NFormer achieves a new state-of-the-art. The code is released at \url{https://github.com/haochenheheda/NFormer}.

cs.CV↗

arXiv2022

Invariant graphs of rational maps

Let $f$ be a postcritically finite rational map. We prove that, as $n$ large enough, there exists an $f^n$-invariant (finite connected) graph on $\widehat{\mathbb{C}}$ such that it contains the postcritical set of $f$.

math.DS↗

arXiv2022

UniSAr: A Unified Structure-Aware Autoregressive Language Model for Text-to-SQL

Existing text-to-SQL semantic parsers are typically designed for particular settings such as handling queries that span multiple tables, domains or turns which makes them ineffective when applied to different settings. We present UniSAr (Unified Structure-Aware Autoregressive Language Model), which benefits from directly using an off-the-shelf language model architecture and demonstrates consistently high performance under different settings. Specifically, UniSAr extends existing autoregressive language models to incorporate three non-invasive extensions to make them structure-aware: (1) adding structure mark to encode database schema, conversation context, and their relationships; (2) constrained decoding to decode well structured SQL for a given database schema; and (3) SQL completion to complete potential missing JOIN relationships in SQL based on database schema. On seven well-known text-to-SQL datasets covering multi-domain, multi-table and multi-turn, UniSAr demonstrates highly comparable or better performance to the most advanced specifically-designed text-to-SQL models. Importantly, our UniSAr is non-invasive, such that other core model advances in text-to-SQL can also adopt our extensions to further enhance performance.

cs.CL↗

arXiv2022

HiTab: A Hierarchical Table Dataset for Question Answering and Natural Language Generation

Tables are often created with hierarchies, but existing works on table reasoning mainly focus on flat tables and neglect hierarchical tables. Hierarchical tables challenge existing methods by hierarchical indexing, as well as implicit relationships of calculation and semantics. This work presents HiTab, a free and open dataset to study question answering (QA) and natural language generation (NLG) over hierarchical tables. HiTab is a cross-domain dataset constructed from a wealth of statistical reports (analyses) and Wikipedia pages, and has unique characteristics: (1) nearly all tables are hierarchical, and (2) both target sentences for NLG and questions for QA are revised from original, meaningful, and diverse descriptive sentences authored by analysts and professions of reports. (3) to reveal complex numerical reasoning in statistical analyses, we provide fine-grained annotations of entity and quantity alignment. HiTab provides 10,686 QA pairs and descriptive sentences with well-annotated quantity and entity alignment on 3,597 tables with broad coverage of table hierarchies and numerical reasoning types. Targeting hierarchical structure, we devise a novel hierarchy-aware logical form for symbolic reasoning over tables, which shows high effectiveness. Targeting complex numerical reasoning, we propose partially supervised training given annotations of entity and quantity alignment, which helps models to largely reduce spurious predictions in the QA task. In the NLG task, we find that entity and quantity alignment also helps NLG models to generate better results in a conditional generation setting. Experiment results of state-of-the-art baselines suggest that this dataset presents a strong challenge and a valuable benchmark for future research.

cs.CL↗

arXiv2022

Stellar Triples as a Source for Ba Stars

Barium stars have been studied extensively over the past few decades, yet our current understanding of how these intriguing objects formed leaves much to be desired. Many trends observed in systems containing barium stars cannot be satisfactorily explained by classical binary evolution models, naturally raising the question of whether triples and other higher-order multiples can give rise to such exotic objects. In this paper, we study the possibility that a Roche Lobe overflow from a tertiary in a hierarchical triple system can potentially lead to surface barium enrichment within the inner binary, while at the same time causing the inner binary to merge, thereby producing a barium star. This possibility has the potential to form a large proportion of Barium stars, as Roche Lobe overflow from a tertiary is typically much more stable for close orbits than that from a binary companion. Various formation channels and mechanisms by which this can be achieved are considered, and constraints on relative formation rates are placed on each scenario. We conclude that a significant, if not dominant, proportion of barium stars are formed from hierarchical triple systems, and that further studies are required in this area before a complete understanding of Barium star populations can be achieved.

astro-ph.SR↗

arXiv2022

Flower: A Friendly Federated Learning Research Framework

Federated Learning (FL) has emerged as a promising technique for edge devices to collaboratively learn a shared prediction model, while keeping their training data on the device, thereby decoupling the ability to do machine learning from the need to store the data in the cloud. However, FL is difficult to implement realistically, both in terms of scale and systems heterogeneity. Although there are a number of research frameworks available to simulate FL algorithms, they do not support the study of scalable FL workloads on heterogeneous edge devices. In this paper, we present Flower -- a comprehensive FL framework that distinguishes itself from existing platforms by offering new facilities to execute large-scale FL experiments and consider richly heterogeneous FL device scenarios. Our experiments show Flower can perform FL experiments up to 15M in client size using only a pair of high-end GPUs. Researchers can then seamlessly migrate experiments to real devices to examine other parts of the design space. We believe Flower provides the community with a critical new tool for FL study and development.

cs.LG↗

arXiv2022

Decoupled IoU Regression for Object Detection

Non-maximum suppression (NMS) is widely used in object detection pipelines for removing duplicated bounding boxes. The inconsistency between the confidence for NMS and the real localization confidence seriously affects detection performance. Prior works propose to predict Intersection-over-Union (IoU) between bounding boxes and corresponding ground-truths to improve NMS, while accurately predicting IoU is still a challenging problem. We argue that the complex definition of IoU and feature misalignment make it difficult to predict IoU accurately. In this paper, we propose a novel Decoupled IoU Regression (DIR) model to handle these problems. The proposed DIR decouples the traditional localization confidence metric IoU into two new metrics, Purity and Integrity. Purity reflects the proportion of the object area in the detected bounding box, and Integrity refers to the completeness of the detected object area. Separately predicting Purity and Integrity can divide the complex mapping between the bounding box and its IoU into two clearer mappings and model them independently. In addition, a simple but effective feature realignment approach is also introduced to make the IoU regressor work in a hindsight manner, which can make the target mapping more stable. The proposed DIR can be conveniently integrated with existing two-stage detectors and significantly improve their performance. Through a simple implementation of DIR with HTC, we obtain 51.3% AP on MS COCO benchmark, which outperforms previous methods and achieves state-of-the-art.

cs.CV↗