Haoli Bai, Wei Zhang, Lu Hou, Lifeng Shang, Jin Jin, Xin Jiang, Qun Liu, Michael Lyu, Irwin King. Proceedings of the 59th Annual Meeting Association for Computational Linguistics and 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Traditional discriminative correlation filter (DCF) has received great popularity due to its high computational efficiency. However, the lightweight framework of DCF cannot promise robust performance when tracker faces appearance variations within background. These unpredictable always distract filter. Most existing DCF-based trackers either utilize deep convolutional features or incorporate additional constraints elevate tracking robustness. Despite some improvements, both them hamper speed...

Autonomous robots must navigate reliably in unknown environments even under compromised exteroceptive perception, or perception failures.Such failures often occur when harsh lead to degraded sensing, the algorithm misinterprets scene due limited generalization.In this paper, we model as invisible obstacles and pits, train a reinforcement learning (RL) based local navigation policy guide our legged robot.Unlike previous works relying on heuristics anomaly detection update navigational...

Laser-induced fluorescence spectra have been recorded for uranyl chloride isolated in a solid Ar matrix. Pulsed excitation was examined using XeCl excimer laser (308 nm) and dye operating the 19500-27500 cm-1 range. Several absorption emission band systems were observed. The characterized by nearly harmonic vibrational progression with frequency of 840 starting at 20323 cm-1. electronic dominated five progressions frequencies approximately 710 Comparisons theoretical calculations indicate...

Current unmanned aerial vehicle (UAV) visual tracking algorithms are primarily limited with respect to: (i) the kind of size variation they can deal with, (ii) implementation speed which hardly meets real-time requirement. In this work, a UAV algorithm powerful estimation ability is proposed. Specifically, overall task allocated to two 2D filters: translation filter for location prediction in space domain, scale and aspect ratio optimization domain. Besides, an efficient two-stage...

A multi-scale quantum harmonic oscillator algorithm (MQHOA) is a population-based proposed recently. It utilizes the wave function to locate global optimum of numerical optimization problem. As MQHOA employs elitism replace worst particle in each iteration cycle, it reduces one particles run, which will cripple diversity population and slow down convergence speed. Therefore, be easily trapped into local optima. In this paper, we suggest new with truncated mean stabilization (TS-MQHOA) policy...

Phase measurement profilometry (PMP) is primarily employed to analyze the morphology of a functional surface with precision. Historically, one most complex and persistent challenges in PMP has been reducing errors stemming from inconsistent indicators at edges surface. In response this challenge, we propose an optimized error compensation methodology specifically designed handle edge artefacts. This introduces Hilbert transform object albedo as tools detect artefact region that need be...

With the development of intelligent manufacturing, production and assembly accuracy components in factories is increasing line with growing demand. However, traditional manual quality inspection inefficient, inaccurate, costly. To this end, digital optical imaging techniques are used to achieve inspection. during reconstruction process, high reflectivity object materials affects speed results. overcome these problems, study investigated three-dimensional (3D) based on laser scanning. It...

Visual tracking has yielded promising applications with unmanned aerial vehicle (UAV). In literature, the advanced discriminative correlation filter (DCF) type trackers generally distinguish foreground from background a learned regressor which regresses implicit circulated samples into fixed target label. However, predefined and unchanged regression results in low robustness adaptivity to uncertain scenarios. this work, we exploit local maximum points of response map generated detection...

Inspired by the swarm intelligence in selforganizing behavior of real Particle Swarm Optimization various algorithms were proposed recently for many research fields data mining such as clustering Compared with previous approaches K-means main advantage based is that no additional information needed initial partitioning or number clusters In this paper, we discuss analysis way a combination advantages particle optimization clustering, since has good global searching quickly.Firstly, center...

Deep neural networks have proven to be effective in solving computer vision and natural language processing problems. To fully leverage its power, manually designed network templates, i.e., Residual Networks, are introduced deal with various tasks. These hand-crafted rely on a large number of parameters, which both data-dependent laborious. On the other hand, architectures suitable for specific tasks also grown exponentially their size topology, prohibits brute force search. address these...

Multi-modal optimization is a troublesome problem faced by algorithms. The multiscale quantum harmonic oscillator algorithm (MQHOA) utilizes group statistics strategy to evaluate the state of population and neglects individual state. It will lead particles be trapped in local optima when addressing multi-modal problems. This paper proposes modified MQHOA introducing strict metastability constraints (MQHOA-SMC). new adopts joint constraint mechanism make particle states mutual with each...

Visual tracking has yielded promising applications with unmanned aerial vehicle (UAV). In literature, the advanced discriminative correlation filter (DCF) type trackers generally distinguish foreground from background a learned regressor which regresses implicit circulated samples into fixed target label. However, predefined and unchanged regression results in low robustness adaptivity to uncertain scenarios. this work, we exploit local maximum points of response map generated detection...

The emotional support dialogue system is an emerging and challenging task in natural language processing to alleviate people’s distress. Each utterance the has features such as emotion, intent, commonsense knowledge. Previous research indicated subpar performance strategy prediction accuracy response generation quality due overlooking certain underlying factors. To address these issues, we propose Advanced Multi-Task Learning Feature-Fusion for Emotional Support Conversation (AdMISC), which...

Abstract Surface defect detection is crucial in industrial production, and due to the conveyor speed, real-time requires 30 60 Frames Per Second, which exceeds capability of most existing methods. This demand for high FPS has driven need lightweight models. Despite significant advancements deep learning-based that have enabled single-stage models such as YOLO series achieve relatively fast detection, methods still face challenges detecting multi-scale defects tiny on complex surfaces while...