Artificial Intelligence and Robotics Commons^™

Adadeep: A Usage-Driven, Automated Deep Model Compression Framework For Enabling Ubiquitous Intelligent Mobiles, Sicong Liu, Junzhao Du, Kaiming Nan, Zimu Zhou, Hui Liu, Zhangyang Wang, Yingyan Lin

Research Collection School Of Computing and Information Systems

Recent breakthroughs in deep neural networks (DNNs) have fueled a tremendously growing demand for bringing DNN-powered intelligence into mobile platforms. While the potential of deploying DNNs on resource-constrained platforms has been demonstrated by DNN compression techniques, the current practice suffers from two limitations: 1) merely stand-alone compression schemes are investigated even though each compression technique only suit for certain types of DNN layers; and 2) mostly compression techniques are optimized for DNNs’ inference accuracy, without explicitly considering other application-driven system performance (e.g., latency and energy cost) and the varying resource availability across platforms (e.g., storage and processing capability). To this …

Ai And The Future Of Work: What We Know Today, Steven M. Miller, Thomas H. Davenport

Research Collection School Of Computing and Information Systems

To contribute to a better understanding of the contemporary realities of AI workplace deployments, the authors recently completed 29 case studies of people doing their everyday work with AI-enabled smart machines. Twenty-three of these examples were from North America, mostly in the US. Six were from Southeast Asia, mostly in Singapore. In this essay, we compare our findings on job and workplace impacts to those reported in the MIT Task Force on the Work of the Future report, as we consider that to be the most comprehensive recent study on this topic.

Solving The Vehicle Routing Problem With Simultaneous Pickup And Delivery And Occasional Drivers By Simulated Annealing, Vincent F. Yu, Grace Aloina, Panca Jodiawan, Aldy Gunawan, Tsung-Chi Huang

Research Collection School Of Computing and Information Systems

This research studies the vehicle routing problem with simultaneous pickup and delivery with an occasional driver (VRPSPDOD). VRPSPDOD is a new variant of the vehicle routing problems with simultaneous pickup and delivery (VRPSPD). Different from VRPSPD, in VRPSPDOD, occasional drivers are employed to work with regular vehicles to service customers’ pickup and delivery requests in order to minimize the total cost. We formulate a mixed integer linear programming model for VRPSPD and propose a heuristic algorithm based on simulated annealing (SA) to solve the problem. The results of comprehensive numerical experiments show that the proposed SA performs well in terms …

Building Action Sets In A Deep Reinforcement Learner, Yongzhao Wang, Arunesh Sinha, Sky C.H. Wang, Michael P. Wellman

Research Collection School Of Computing and Information Systems

In many policy-learning applications, the agent may execute a set of actions at each decision stage. Choosing among an exponential number of alternatives poses a computational challenge, and even representing actions naturally expressed as sets can be a tricky design problem. Building upon prior approaches that employ deep neural networks and iterative construction of action sets, we introduce a reward-shaping approach to apportion reward to each atomic action based on its marginal contribution within an action set, thereby providing useful feedback for learning to build these sets. We demonstrate our method in two environments where action spaces are combinatorial. Experiments …

Rmm: Reinforced Memory Management For Class-Incremental Learning, Yaoyao Liu, Qianru Sun, Qianru Sun

Research Collection School Of Computing and Information Systems

Class-Incremental Learning (CIL) [38] trains classifiers under a strict memory budget: in each incremental phase, learning is done for new data, most of which is abandoned to free space for the next phase. The preserved data are exemplars used for replaying. However, existing methods use a static and ad hoc strategy for memory allocation, which is often sub-optimal. In this work, we propose a dynamic memory management strategy that is optimized for the incremental phases and different object classes. We call our method reinforced memory management (RMM), leveraging reinforcement learning. RMM training is not naturally compatible with CIL as the …

Hierarchical Control Of Multi-Agent Reinforcement Learning Team In Real-Time Strategy (Rts) Games, Weigui Jair Zhou, Budhitama Subagdja, Ah-Hwee Tan, Darren Wee Sze Ong

Research Collection School Of Computing and Information Systems

Coordinated control of multi-agent teams is an important task in many real-time strategy (RTS) games. In most prior work, micromanagement is the commonly used strategy whereby individual agents operate independently and make their own combat decisions. On the other extreme, some employ a macromanagement strategy whereby all agents are controlled by a single decision model. In this paper, we propose a hierarchical command and control architecture, consisting of a single high-level and multiple low-level reinforcement learning agents operating in a dynamic environment. This hierarchical model enables the low-level unit agents to make individual decisions while taking commands from the high-level …

Deep Learning Predicts Ebv Status In Gastric Cancer Based On Spatial Patterns Of Lymphocyte Infiltration, Baoyi Zhang, Kevin Yao, Min Xu, Jia Wu, Chao Cheng

Computer Vision Faculty Publications

EBV infection occurs in around 10% of gastric cancer cases and represents a distinct subtype, characterized by a unique mutation profile, hypermethylation, and overexpression of PD-L1. Moreover, EBV positive gastric cancer tends to have higher immune infiltration and a better prognosis. EBV infection status in gastric cancer is most commonly determined using PCR and in situ hybridization, but such a method requires good nucleic acid preservation. Detection of EBV status with histopathology images may complement PCR and in situ hybridization as a first step of EBV infection assessment. Here, we developed a deep learning-based algorithm to directly predict EBV infection …

Evaluation Of Deep Neural Network Prospr For Accurate Protein Distance Predictions On Casp14 Targets, Jacob A. Stern, Bryce Eric Hedelius, Olivia Fisher, Wendy M. Billings, Dennis Della Corte

Faculty Publications

The field of protein structure prediction has recently been revolutionized through the introduction of deep learning. The current state-of-the-art tool AlphaFold2 can predict highly accurate structures; however, it has a prohibitively long inference time for applications that require the folding of hundreds of sequences. The prediction of protein structure annotations, such as amino acid distances, can be achieved at a higher speed with existing tools, such as the ProSPr network. Here, we report on important updates to the ProSPr network, its performance in the recent Critical Assessment of Techniques for Protein Structure Prediction (CASP14) competition, and an evaluation of its …

Multi-Modal Transformers Excel At Class-Agnostic Object Detection, Muhammad Maaz, Hanoona Bangalath Rasheed, Salman Hameed Khan, Fahad Shahbaz Khan, Rao Muhammad Anwer, Ming-Hsuan Yang

Computer Vision Faculty Publications

What constitutes an object? This has been a longstanding question in computer vision. Towards this goal, numerous learning-free and learning-based approaches have been developed to score objectness. However, they generally do not scale well across new domains and for unseen objects. In this paper, we advocate that existing methods lack a top-down supervision signal governed by human-understandable semantics. To bridge this gap, we explore recent Multi-modal Vision Transformers (MViT) that have been trained with aligned image-text pairs. Our extensive experiments across various domains and novel objects show the state-of-the-art performance of MViTs to localize generic objects in images. Based on …

Situate: An Agent-Based System For Situation Recognition, Max Henry Quinn

Dissertations and Theses

Computer vision and machine learning systems have improved significantly in recent years, largely based on the development of deep learning systems, leading to impressive performance on object detection tasks. Understanding the content of images is considerably more difficult. Even simple situations, such as "a handshake", "walking the dog", "a game of ping-pong", or "people waiting for a bus", present significant challenges. Each consists of common objects, but are not reliably detectable as a single entity nor through the simple co-occurrence of their parts.

In this dissertation, toward the goal of developing machine learning systems that demonstrate properties associated with understanding, …

Analysis And Strategy Of Ai Ethical Problems, Zhaoxiang Zhang, Jiyu Zhang, Tieniu Tan

Bulletin of Chinese Academy of Sciences (Chinese Version)

Artificial intelligence (AI) is the core of the fourth industrial revolution, and it has brought challenges to ethics and social governance. On the basis of explaining the current ethical risks of artificial intelligence, the study furtherly analyzes the current consensus on ethics, governance principles, and governance approaches of artificial intelligence. Moreover, the study also proposes to take "co-construction, co-governance and sharing" as the guiding theory to gradually build a multi-dimensional ethical governance system, including education reform, ethical norms, technical supports, legal regulations, and international cooperation.

Restormer: Efficient Transformer For High-Resolution Image Restoration, Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang

Computer Vision Faculty Publications

Since convolutional neural networks (CNNs) perform well at learning generalizable image priors from large-scale data, these models have been extensively applied to image restoration and related tasks. Recently, another class of neural architectures, Transformers, have shown significant performance gains on natural language and high-level vision tasks. While the Transformer model mitigates the shortcomings of CNNs (i.e., limited receptive field and in-adaptability to input content), its computational complexity grows quadratically with the spatial resolution, therefore making it infeasible to apply to most image restoration tasks involving high-resolution images. In this work, we propose an efficient Transformer model by making several key …

Using Deep Learning To Predict The Path Of A Shuttlecock In Badminton, Sachleen Singh

Student Theses and Dissertations

With this thesis we intend to predict the movement of the shuttlecock given ten frames of a video for the each of the next ten frames. We also present a new regression model for the prediction of the frames called PIFR [Present Imputation and Future Regression] which consists of two models prepared for the same. Both the models try to predict the “future” position of the shuttlecock, the players and their rackets. There are two parts of the prediction network: An object detection stage based YOLO and the regression model. [40]

Fuzzy Information Granulation And Improved Rvm For Rolling Bearing Life Prediction, Xiaoman Hu, Wang Yan, Zhicheng Ji

Journal of System Simulation

Abstract: Aiming at the low accuracy in life prediction and unpredictable problems of degenerative performance trends and fluctuation ranges, etc. Of the bearing life prediction, an improved complete ensemble empirical mode decomposition with adaptive noise analysis and fuzzy information granulating method of improved relevance vector machine is proposed. Focusing on bearing data containing a lot of noise, through the improved complete ensemble empirical mode decomposition with adaptive noise analysis in combination with wavelet packet denoising, the principal component analysis is carride out by exitracing a variety of characeteristics of the signal, the effective information is extracted by granulating the fuzzy …

Simulation Of Pedestrian In Multifunctional Passageway Of Metro Station Area Based On Social Force Model, Wang Xi, Zhang Rui, Fei Shuo, Minghang Yang

Journal of System Simulation

Abstract: Transitional passageway connecting subway stations and commercial facilities is generally designed as the multifunctional passageway, in which the traffic function is the main and the service function is the auxiliary. The impact of the service facilities on both sides of the passage on pedestrian traffic is difficult to be quantitatively analyzed and simulated and modeled. Through measurement, it is found that the viscous effect of service facilities on pedestrian traffic is mainly slowing down the speed or changing the trajectory direction. Through analyzing the horizontal influence range of service facilities, dividing the passage into different areas, and introducing the …

Midcourse Guidance Method Based On Fading Memory Filter For Intercepting Near-Space Gliding Target, Yonghua Fan, Yilun Huangfu, Xiaowen Guo, Chenlu Li, Guofei Li

Journal of System Simulation

Abstract: Aiming at the interception of hypersonic gliding target in near space, a sliding mode guidance law based on fading memory filtering algorithm is proposed. The accurate motion parameters of the hypersonic target are obtained based on the current statistical model fading memory EKF(Extended Kalman Filter) algorithm. Based on the filtering information and sliding mode control theory, a sliding mode guidance law is designed to adjust the interception trajectory online according to the target maneuver law. The simulation results show that the proposed fading memory filtering algorithm can effectively track the gliding target with high filtering accuracy. For a …

Kernel Block Diagonal Representation Subspace Clustering And Its Convergence Analysis, Maoshan Liu, Zhicheng Ji, Wang Yan, Jianfeng Wang

Journal of System Simulation

Abstract: Focus on the problems that the linear block diagonal representation subspace clustering cannot effectively handle non-linear visual data, and the regular regularizers cannot directly pursue the k-block diagonal matrix, a kernel block diagonal representation subspace clustering is proposed. In the proposed algorithm, the original input space is mapped into the kernel Hilbert space which is linearly separable, and the spectral clustering is performed in the feature space. The convergence analysis is given, and the strong convex of variables and the boundedness of function is utilized to verify the monotonically decreasing of objective function and the boundedness and convergence of …

Research On Flexible Job-Shop Dynamic Scheduling Based On Game Theory, Yichen You, Wang Yan, Zhicheng Ji

Journal of System Simulation

Abstract: To quickly and effectively respond to the machine fault disturbance events in Flexible Job-shop Scheduling Problem (FJSP), a flexible job-shop dynamic scheduling based on game theory is established. A pre-scheduling scheme is generated under Non-Dominated Sort Genetic Algorithm-Ⅱ (NSGA-Ⅱ) algorithm which introduces self-adapted crossover operators to improve the population diversity. For FJSP dynamic scheduling with machine fault, a multi-stage complete information game model is built to better balance the stability and robustness indicators and respond quickly to the machine fault, in which the stability and robustness indicators are mapped to the game players, and a hybrid Nash Equilibrium which …

Single-Frame Image Motion Parallax Key Point Estimation Combined With Self-Supervised Learning, Zhihao Huo, Weidong Jin, Tang Peng

Journal of System Simulation

Abstract: The motion parallax key point FOE (Focus of Expansion) is an important parameter of railway catenary video inspection. The current method of calculating FOE requires multi-frame image matching estimation, which has high time complexity. Aiming at the single-frame image FOE estimation, a single-frame image FOE estimation algorithm fused with self-supervised learning is proposed. A full convolutional network F-VGG(Fully-Visual Geometry Group) is built as the FOE predictor, and the training label of the sample data is automatically generated through the fusion agent task, which realizes the end-to-end single-frame image FOE estimation. The experimental results show that the method has an …

Research On Intelligent Gait Recognition Method Based On Plantar Pressure Perception, Xueqin Liu, Liu Ning, Su Zhong, Jingxiao Wang, Chaojie Yuan

Journal of System Simulation

Abstract: In view of the complexity and low accuracy of gait recognition in the past, an intelligent gait recognition method based on plantar pressure perception is proposed. The pressure data of the gait of plantar periodic motion is collected and the obtained gait data is classified by the vector machines,the intelligent gait recognition of plantar pressure perception is realized, and the accuracy of gait feature analysis is improved. Through experiment verification, the overall classification accuracy of the classifier is more than 90%, which verifies the rationality of the feature extraction. By evaluating the real state and the results of …

Short-Term Wind Power Prediction Method Based On Random Forest, Liu Xing, Wang Yan, Zhicheng Ji

Journal of System Simulation

Abstract: In order to effectively predict the power and value fluctuation range of the short-term wind, a wind power prediction method based on clustering and kernel principal component analysis combined with random forest algorithm is proposed. The clustering analysis data processing method is used to preprocess the meteorological wind power generation data to improve the data quality, and the kernel principal component analysis method is used to reduce the dimensionality of the eight groups of characteristic data to remove the correlation of the wind power data, the random forest algorithm is used to forecast the wind power, to obtain …

Access Control Mechanism Of Uav Cluster Based On Blockchain Smart Contract, Ting Duan, Weiping Wang, Yifan Zhu, Wang Tao, Meigen Huang

Journal of System Simulation

Abstract: A Unmanned Aerial Vehicle (UAV) cluster access control mechanism based on Ethereum blockchain smart contract is proposed to solve the problems of the strategic stability and low security of UAV cluster access control mechanism. The role-based access control mechanism model is improved, and the formal definition of the access control model for UAV cluster is given. The access control architecture of UAV cluster based on blockchain technology is proposed, and the corresponding basic framework and execution process is proposed, which can effectively reduce the cost of UAV cluster operation management resources, solve the problem of incomplete state …

Modeling And Simulation Of Radiation Measurement System Based On Monte Carlo Method, Jinghai Cheng, Hongzhi Wang, Luoyuan Xu, Xia Tian

Journal of System Simulation

Abstract: A method is applied to build a virtual simulation radiometric measurement system. The mathematical and physical models of gamma ray interaction with matter, radiation sources, measurement electronics system and protective materials are constructed by using Monte Carlo method. Through numerical calculation and scene simulation of the radiation measurement system, virtual simulation acquisition and energy spectrum processing of radiation measurement data are realized. It, the system, can simulate single channel measurement and computer multi-channel measurement experiments. It can realize energy measurement, activity measurement and energy spectrum measurement of mixed, unknown or custom radiation sources in different size crystals. It can …

System Performance Evaluation Method Based On Multi-Source Prior Data, Haozhe Liu, Li Wei, Ma Ping, Yang Ming

Journal of System Simulation

Abstract: When using the Bayes method to evaluate the performance of the system with multi-source prior data, the multi-source prior data is fused, the posterior distribution is calculated by synthesizing the fused prior distribution and test data. The parameters of posterior distribution are estimated to obtain the performance evaluation results. A weighted fusion method of multi-source prior data based on Kullback-Leibler divergence is proposed, which can effectively integrate the multi-source prior data. The commonly used Markov Chain Monte Carlo method is used to estimate the parameters of Bayes posterior distribution. The influence of different proposal distributions on the sampling results …

Normalization Of Simulation System Credibility Index Based On Vague Set, Yuhang Ren, Li Wei, Ma Ping, Yang Ming

Journal of System Simulation

Abstract: Focus on various types of simulation system credibility indexes and the difficulty to convert the index results to credibility, a normalization method of credibility indexes based on Vague sets is proposed, which includes qualitative and quantitative conversion methods; Aiming at the problem of credibility Vague value index synthesis, the weighted arithmetic mean operator and the weighted geometric mean operator based on Vague set are given, and the applications are explained; According to the similarity principle of Vague sets, a method of transforming the credibility Vague value to the credibility single value is proposed, which improves the …

Trajectory Tracking Control Of Planetary Entry Phase Based On Neural Network And Fractional Sliding Mode, Cunli Fan, Dai Juan, Haitao Liu, Su Zhong, Zhu Cui, Wenting Xu

Journal of System Simulation

Abstract: A fractional order sliding mode control method based on Radial Basis Function (RBF) neural network is proposed to solve the landing accuracy being affected by the interference during the landing process of planetary probe. Based on sliding mode control, a trajectory tracking control method for the entry phase of the probe is designed. Fractional calculus is introduced to alleviate the chattering caused by sliding mode control. RBF neural network is used to estimate and compensate the atmospheric density uncertainty. The method is applied to Mars landing scene simulation. The simulation results show that the proposed control method can accurately …

An Electromechanical-Electromagnetic Transient Stability Simulation System For Ac/Dc Hybrid Power System, Weijie Dong, Huang Min, Guoqing He, Bao Wei, Yilong Wang, Liu Quan

Journal of System Simulation

Abstract: In order to improve the hybrid simulation speed of AC/DC power grid, it is necessary to improve the simulation method of sub grid parallel. An electromechanical transient stability simulation system is presented for AC/DC hybrid power grid. The AC/DC sub network module is used to divide the large AC/DC power grid into small AC/DC sub networks, so that each sub network can be simulated in parallel. The numerical calculation method is improved for the efficiency and accuracy of simulation calculation. Taking IEEE 10-39 bus as an example, the effectiveness of the method is verified in PSCAD (Power Systems Computer …

Supplier Selection Based On Supplier Portrait And Markov Monte Carlo Method, Bingli Sun, Song Xiao, Guanghong Gong

Journal of System Simulation

Abstract: The supplier selection problem is a complex multi-objective decision-making problem and the key is how to establish the supplier's portrait. For the supplier selection of aerospace equipment, enterprise qualification management, business risks, and product quality are comprehensively considered. Based on Bayesian theory, the multi-parameter joint distribution derivation of portrait sample data is realized. Combined with the mathematical model derived, a Markov Monte Carlo simulation method is proposed. And combined with Gibbs sampler, the supplier ranking and selection are achieved when data is difficult to obtain or missing, which provides a new idea for supplier selection in the aerospace …

Simulation Of Zero-Speed Correction Algorithm For Underground Space Individual Positioning, Yijing Wang, Su Zhong, Li Qing, Li Lei

Journal of System Simulation

Abstract: In view of the complex and dangerous collapse environment of the tunnel, the related safety hazards of the positioning system of the tunnel rescuer are intensively analyzed, and the simulation of the inertial device worn on the chest, waist, calf, and foot surface shows that the correction on the foot surface is the best. Focus on the error accumulation of inertial devices, according to the fact that the speed is near zero when the sole of the foot fully touches the ground during walking, the algorithm of zero-speed correction for acceleration and angular velocity is compared, and a combination …

A Natural Computing Method Based On Spatial Division Search Strategy, Xiaoqing Sun, Cheng Hao, Luyao Zhang, Weidong Ji, Wang Xu

Journal of System Simulation

Abstract: A natural computing method based on spatial division search strategy is proposed. The strategy can map the high-dimensional space to the three-dimensional Cartesian coordinate system by grouping the dimensional space into a group of three dimensions. The individual after spatial segmentation is numbered into subindividual, to increases the particle number while reducing the dimension, thus the individual is distributed over wider search space to effectively increases the diversity of the population. The algorithm iterates to a certain extent and can synthesize the individual into the original individual through the numbered index. By calculating the fitness value, some poor …