Open Access. Powered by Scholars. Published by Universities.®

Artificial Intelligence and Robotics Commons™

Open Access. Powered by Scholars. Published by Universities.®

2022

Discipline
Institution
Keyword
Publication
Publication Type

Articles 211 - 240 of 965

Full-Text Articles in Artificial Intelligence and Robotics

Parallel Simulation System Of Equipment Precision Maintenance Based On Cloud-Edge-End Architecture, Yanqiang Di, Ting Li, Shaochong Feng, Qiongyao Liu, Jianhong Lü, Zhijia Chen, Yang Zhang, Pengfei Cao Sep 2022

Parallel Simulation System Of Equipment Precision Maintenance Based On Cloud-Edge-End Architecture, Yanqiang Di, Ting Li, Shaochong Feng, Qiongyao Liu, Jianhong Lü, Zhijia Chen, Yang Zhang, Pengfei Cao

Journal of System Simulation

Abstract: Aiming at the demands of equipment precision maintenance, based on the previous research results of equipment parallel simulation, the algorithm of equipment remaining useful life (RUL)prediction is optimized and a parallel simulation system for equipment precision maintenance with cloud-edge-end architecture is designed. At the equipment end, the system collects equipment status data and preprocesses it with edge devices. At the cloud end, based on simulation model, in parallel with the equipment entity, the system dynamically predicts the RUL of equipment. The prediction results are applied to the formulation and deduction of equipment maintenance plans to support the equipment …

Go to article

Unmanned Air Vehicles Launching Aircraft Combat System And Key Technologies For Penetrating Counterair, Minghao Li, Wenhao Bi, An Zhang, Wenxuan Sun Sep 2022

Unmanned Air Vehicles Launching Aircraft Combat System And Key Technologies For Penetrating Counterair, Minghao Li, Wenhao Bi, An Zhang, Wenxuan Sun

Journal of System Simulation

Abstract: Penetrating counterair is an important countermeasure to the anti-access/area denial environment. Under this operational requirement, the unmanned air vehicles launching aircraft (UAVLA)has received much attention due to the advantages of load quantity and variety, operational range and duration, and development time and cost. Based on the review of the concept development and supporting research related to the UAVLA, the top-level concepts of operations such as the component systems,operational process, operational events tracking, and information interaction of the UAVLA combat system (UAVLACS) are designed. The key technologies of the system are prospected from four aspects:intelligent cognition of battlefield situation under …

Go to article

Opnet Based Simulation Of Hybrid Tdma Protocol For Helicopters Datalink, Yanfang Fu, Nan Zhang, Jianing Wei, Shaochun Qu, Ying Lu, Chang Liu Sep 2022

Opnet Based Simulation Of Hybrid Tdma Protocol For Helicopters Datalink, Yanfang Fu, Nan Zhang, Jianing Wei, Shaochun Qu, Ying Lu, Chang Liu

Journal of System Simulation

Abstract: For the current time slot allocation problem of the data link, an improved hybrid time slot allocation protocol based on grey relational analysis is proposed and implemented. Through the aggregation of throughput, delay and load of current message buffer by grey relational analysis, the comprehensive evaluation index is obtained, and the time slot is allocated dynamically. The fixed time slot allocation is also adopted to ensure that at least one time slot is available for nodes in the network.Simulation results show that compared with the fixed TDMA(time division multiple access) protocol and the P-TDMA(priority-TDMA) protocol, the proposed protocol …

Go to article

Research On Opponent Modeling Framework For Multi-Agent Game Confrontation, Junren Luo, Wanpeng Zhang, Weilin Yuan, Zhenzhen Hu, Shaofei Chen, Jing Chen Sep 2022

Research On Opponent Modeling Framework For Multi-Agent Game Confrontation, Junren Luo, Wanpeng Zhang, Weilin Yuan, Zhenzhen Hu, Shaofei Chen, Jing Chen

Journal of System Simulation

Abstract: As the key technology of multi-agent game confrontation, opponent modeling is a typical cognitive modeling method of agent's behavior. Several typical models of multi-agent game confrontation,non-stationary problems, and meta-game theory are introduced; opponent modeling methods that concludes the frontier theory of opponent modeling are summarized, and the applications and challenges are analyzed. Based on the theory of meta-game, a general opponent modeling framework is constructed with three modules: opponent policy recognition and generation, opponent policy space reconstruction,and opponent exploitation. It is expected to provide theoretical and methodological reference for opponent modeling in multi-agent game confrontation.

Go to article

Siamese Object Tracking Algorithm Combined With The Intersection Over Union Loss, Wei Zhou, Yuxiang Liu, Guangping Liao, Xin Ma Sep 2022

Siamese Object Tracking Algorithm Combined With The Intersection Over Union Loss, Wei Zhou, Yuxiang Liu, Guangping Liao, Xin Ma

Journal of System Simulation

Abstract: To improve the accuracy of the object bounding box regression prediction of the SiamRPN,solve the problem of low discrimination of positive samples in classification prediction and the lack of correlation between regression prediction and classification prediction, an improved object tracking algorithm of SiamRPN which combined with IoU(intersection over union) loss is proposed. A joint optimization module of IoU-smooth L1 is designed to optimize the IoU loss of the best positive sample and the smooth L1 loss of other positive samples jointly. According to the regression prediction results, the weighted classification prediction is performed on the positive samples with the …

Go to article

Relative-Residual-Based Dynamic Schedule For Decoding Of Ldpc Codes, Fatang Chen, Hebin Li, Zhihao Zhang, Zhiqiang Mei Sep 2022

Relative-Residual-Based Dynamic Schedule For Decoding Of Ldpc Codes, Fatang Chen, Hebin Li, Zhihao Zhang, Zhiqiang Mei

Journal of System Simulation

Abstract: In order to solve the problems of the oscillation phenomenon and greedy characteristics in the dynamic scheduling decoding algorithm for low-density parity-check (LDPC) codes, the relative-residual-based dynamic schedule (RRB-BP) algorithm is proposed based on variable-to-check residual belief propagation (VC-RBP) algorithm. The variable nodes are grouped, then the relative residual value of the message passed by the variable nodes to the check node is taken as a reference, and the node with the largest relative residual value is updated in priority to accelerate the decoding convergence speed. For variable nodes oscillating in the decoding process, the posterior LLR (log likelihood …

Go to article

Energy-Efficient Scheduling Of Multi-Objective Flexible Job Shop Considering Interval Processing Time, Hongliang Zhang, Renman Ding, Gongjie Xu Sep 2022

Energy-Efficient Scheduling Of Multi-Objective Flexible Job Shop Considering Interval Processing Time, Hongliang Zhang, Renman Ding, Gongjie Xu

Journal of System Simulation

Abstract: Based on the comprehensive consideration of economic indicators and environmental factors, the energy-efficient scheduling problem of multi-objective flexible job shop with uncertain processing time is studied. The interval number is used to describe uncertain processing time of the workpiece, and the optimization model for energy-efficient problem of interval flexible job shop scheduling is established to minimize the maximum interval completion time and total energy consumption. According to the domination relation of interval possibility degree, an effective interval multi-objective evolutionary algorithm is designed. The simulation experiments of the interval multi-objective evolutionary algorithm, SPEA-II and NSGA-II are carried out through 15 …

Go to article

Particle Swarm Algorithm For Solving Emergency Material Dispatch Considering Urgency, Li Zhang, Huizhen Zhang, Dong Liu, Yuxin Lu Sep 2022

Particle Swarm Algorithm For Solving Emergency Material Dispatch Considering Urgency, Li Zhang, Huizhen Zhang, Dong Liu, Yuxin Lu

Journal of System Simulation

Abstract: In the early stage of major public health events, medical supplies are rapidly consumed and severely insufficient. In order to distribute medical supplies in a reasonable and efficient manner, research on the distribution of emergency medical materials is carried out. The entropy method is introduced to determine the urgency of demand points, thus could give priority to the demand points with high urgency and make the distribution routing as short as possible on that basis to realize the construction of a split delivery and multi-objective emergency medical materials scheduling model based on different urgency of demand points. Meanwhile the …

Go to article

Design Of Optical Compound Eye Simulation Software For Small Aircraft Applications, Qiming Qi, Ruigang Fu, Ping Wang, Min Wang, Hongqi Fan Sep 2022

Design Of Optical Compound Eye Simulation Software For Small Aircraft Applications, Qiming Qi, Ruigang Fu, Ping Wang, Min Wang, Hongqi Fan

Journal of System Simulation

Abstract: Optical compound eye has the advantages of large field of view, multiple viewing angles and high resolution. With another advantage that it can conformal combine with small aircraft, optical compound eye has application value in reconnaissance and surveillance, target detection, image navigation and other aspects. An optical compound eye simulation software for small aircraft is designed for the current situation of long development period of optical compound eye design and high cost of flight test in practical applications. The software integrates compound eye imaging, aircraft simulation and data management, and each functional module is extensible. The simulation results show …

Go to article

Mesoscopic Modeling And Simulation Of Mixed Traffic Flow Of Buses And Vehicles, Yiting Zhu, Yun Yan, Zhaocheng He Sep 2022

Mesoscopic Modeling And Simulation Of Mixed Traffic Flow Of Buses And Vehicles, Yiting Zhu, Yun Yan, Zhaocheng He

Journal of System Simulation

Abstract: Aiming at the problem that the existing mesoscopic simulation models only convert buses into several standard vehicles and ignore the movement difference between buses and vehicles, a mesoscopic simulation model of mixed traffic flow is proposed. In the process of road driving, on the one aspect, we consider the feature that bus speed is usually lower than vehicle speed, and correspondingly establish the reduction function of bus speed; on the other aspect, we consider the influences of bus-station queue overflow on the adjacent lanes, and correspondingly construct the lane-based speed model of mixed flow.Moreover, we use the …

Go to article

Electrical Resistance Tomography And Flow Pattern Identification Method Based On Deep Residual Neural Network, Weiguo Tong, Shichao Zeng, Lifeng Zhang, Zhe Hou, Jiayue Guo Sep 2022

Electrical Resistance Tomography And Flow Pattern Identification Method Based On Deep Residual Neural Network, Weiguo Tong, Shichao Zeng, Lifeng Zhang, Zhe Hou, Jiayue Guo

Journal of System Simulation

Abstract: Aiming at the low accuracy of inverse problem imaging and flow pattern recognition in electrical resistance tomography (ERT), a two-phase flow electrical resistance tomography and flow pattern recognition method based on the deep residual neural network is proposed. The finite element method is used to model the ERT forward problem to construct the "boundary voltage-conductivity distribution-flow pattern category" dataset of various gas-liquid two-phase flow distributions. The residual neural network for ERT image reconstruction and flow pattern identification of gas-liquid two-phase flow is built and trained. The two outputs of the residual neural network are processed respectively to obtain …

Go to article

Study On Invulnerability Of Urban Agglomeration Passenger Traffic Network Considering Time Characteristics, Chengbing Li, Yunfei Li, Peng Wu Sep 2022

Study On Invulnerability Of Urban Agglomeration Passenger Traffic Network Considering Time Characteristics, Chengbing Li, Yunfei Li, Peng Wu

Journal of System Simulation

Abstract: The cascading failure invulnerability study of comprehensive passenger transport network in urban agglomeration is helpful to improve the safety and transportation efficiency of intercity travel. In order to be consistent with the actual situation, a comprehensive passenger transport network model for urban agglomerations is constructed based on multi-layer complex network theory and actual passenger flow. The passenger transport network cascading failure invulnerability model considering time characteristics is established with unit time step. The spatial and temporal evaluation indexes are put forward to analyze the network situation in each period. Taking Hu-Bao-E-Yu urban agglomeration as an example, the results show …

Go to article

Research On Ofdm Signal Detection Method Based On Intrawell Stochastic Resonance Of Bistable System, Gaohui Liu, Ying Liang Sep 2022

Research On Ofdm Signal Detection Method Based On Intrawell Stochastic Resonance Of Bistable System, Gaohui Liu, Ying Liang

Journal of System Simulation

Abstract: In order to solve the problem of weak OFDM (orthogonal frequency division multiplexing)signal detection at the receiver in OFDM transmission system, the intrawell stochastic resonance of bistable system is combined with the OFDM signal enhancement and demodulation process. Analytical expression is derived for the time required to change from zero state to potential well state for the intrawell stochastic resonance system under the excitation of multicarrier signals, and the energy loss of multicarrier signals in one symbol caused by the transient response is analyzed. The steady-state output equation of system is derived, and the problem of superimposing the …

Go to article

Architecture Design And Prototype Verification Of Railway Vehicle Dynamics Cloud Platform, Junjie Sheng, Zhao Tang, Shaodi Dong, Shuyang Wu, Hao Liang Sep 2022

Architecture Design And Prototype Verification Of Railway Vehicle Dynamics Cloud Platform, Junjie Sheng, Zhao Tang, Shaodi Dong, Shuyang Wu, Hao Liang

Journal of System Simulation

Abstract: Since almost all the software in the railway vehicle field is controlled by foreign capital, it is difficult to catch up with the development of independent vehicle system software based on single machine deployment mode in a short time. In view of this, a set of autonomous and controllable vehicle system dynamics software architecture based on cloud platform is proposed. Based on the railway vehicle system dynamics and cloud services, a cloud platform with automatic process modeling, cloud computing,post-processing analysis is built. A simulation model of a trailer caris applied in the platform, and compared with the SIMPACK …

Go to article

A High Resolution Reconstruction Method Of Temperature Distribution In Acoustic Tomography, Lifeng Zhang, Yu Miao Sep 2022

A High Resolution Reconstruction Method Of Temperature Distribution In Acoustic Tomography, Lifeng Zhang, Yu Miao

Journal of System Simulation

Abstract: Accurate measurement temperature distribution is important for industrial production. In order to solve the number of mesh divisions will impact reconstruction accuracy in acoustic tomography, the TR-RBF (Tikhonov regularization-radial basis function) reconstruction algorithm is rebuilt to reconstruct the temperature field with high resolution. The Tikhonov regularization is used to reconstruct the ultrasound time of flight (TOF) to obtain a temperature distribution on coarse grids, and use local weighted regression method to smooth processing; use RBF neural networks to predict the temperature distribution on fine grids. Through numerical simulation with and without noise, compared with ART,SVD and Tikhonov, the proposed …

Go to article

Research On Cooperative Adaptive Cruise Control Strategy Based On Improved Mpc, Qiming Wang, Jiangyue Jiang, Zhichao Lü, Hanzu Zhang Sep 2022

Research On Cooperative Adaptive Cruise Control Strategy Based On Improved Mpc, Qiming Wang, Jiangyue Jiang, Zhichao Lü, Hanzu Zhang

Journal of System Simulation

Abstract: To solve the problems of environmental interference, sensor noise and poor tracking stability of time-varying speed, an improved MPC (model predictive control) algorithm based on KF (kalman filtering) is proposed. The longitudinal kinematics model of CACC(cooperative adaptive cruise control)between vehicles is established and the discrete state space equation is created. KF is used to reduce the noise of state variables, and at the same time, the prediction model is designed for robustness. The CACC control objectives are analyzed under different working conditions and the objective optimization functions are created. Verify by building Simulink and CarSim co-simulation model, the simulation …

Go to article

Simulation Of The Market Exclusive Competition Between Platforms, Wen Zheng, Zhe Zhang, Jingyi Zhu Sep 2022

Simulation Of The Market Exclusive Competition Between Platforms, Wen Zheng, Zhe Zhang, Jingyi Zhu

Journal of System Simulation

Abstract: As for the problem of an exclusive competition between the platforms, 2 competing PlatformsAgent, 100 ConsumersAgent and 300 SellersAgent are introduced and encapsulated into a closed market environment in BarriersModelSwarm. A two-sided market system is constructed with the cross-network externality. Through BarriersObserverSwarm, the Agents attribute information and behavior strategy are cross-called, and the virtual connection class Orderand ArrayList class in the Virtual Connection Classes are generated to run cyclically. The unilateral dependence degree of Consumers/SellersAgent is triggered, which restores the exclusive of the two-sided market competition in comparison with the platform transaction scale, market concentration, and platform cumulative capital. …

Go to article

Turbofan Engine Fault Prediction Based On Evidential Reasoning And Belief Rule Base, Hailong Zhu, Ruxia Jia, Liang Zhang, Wei He Sep 2022

Turbofan Engine Fault Prediction Based On Evidential Reasoning And Belief Rule Base, Hailong Zhu, Ruxia Jia, Liang Zhang, Wei He

Journal of System Simulation

Abstract: Aiming at the fault prediction problem of a turbofan engine, a fault prediction model based on evidential reasoning (ER) and belief rule base (BRB) is proposed. In order to describe the health state of turbofan engine, ER algorithm is adopted to fuse the state information. Combined with prior knowledge, a hybrid driven simulation prediction of BRB model is established. Projection covariance matrix adaptive evolution strategy (P-CMA-ES) is used to optimize the model parameters. The validity of the model is verified by experiments. Experimental results show that the proposed method not only accurately predicts the probability of failure …

Go to article

Modulation Recognition Algorithm Based On Truncated Migration And Parallel Resnet, Yecai Guo, Qingwei Wang Sep 2022

Modulation Recognition Algorithm Based On Truncated Migration And Parallel Resnet, Yecai Guo, Qingwei Wang

Journal of System Simulation

Abstract: A truncated migration data preprocessing algorithm is proposed for the problem of limited time series characteristics of the signal extracted by convolutional neural network. The distance unit at one end of the sampling matrix is truncated, migrated to the other end to form a new matrix, allowing the convolutional neural network to extract more sampling points and compare more symbolic information.An improved parallel ResNet is proposed, which focuses on features in both horizontal and vertical directions simultaneously by two parallel branches. The results show that the algorithm has an accuracy rate of about 10% higher than that of ordinary …

Go to article

A Machine Learning Framework For Automatic Speech Recognition In Air Traffic Control Using Word Level Binary Classification And Transcription, Fowad Shahid Sohail Sep 2022

A Machine Learning Framework For Automatic Speech Recognition In Air Traffic Control Using Word Level Binary Classification And Transcription, Fowad Shahid Sohail

Theses and Dissertations

Advances in Artificial Intelligence and Machine learning have enabled a variety of new technologies. One such technology is Automatic Speech Recognition (ASR), where a machine is given audio and transcribes the words that were spoken. ASR can be applied in a variety of domains to improve general usability and safety. One such domain is Air Traffic Control (ATC). ASR in ATC promises to improve safety in a mission critical environment. ASR models have historically required a large amount of clean training data. ATC environments are noisy and acquiring labeled data is a difficult, expertise dependent task. This thesis attempts to …

Go to article

An Enterprise Risk Management Framework To Design Pro-Ethical Ai Solutions, Quintin P. Mcgrath Sep 2022

An Enterprise Risk Management Framework To Design Pro-Ethical Ai Solutions, Quintin P. Mcgrath

USF Tampa Graduate Theses and Dissertations

The effective use of Artificial Intelligence (AI) has immediate business benefits for an organization and its stakeholders through efficiency and quality gains, and the potential to explore and implement new business models. However, there are risks of unintended ethical consequences. Enterprise Risk Management (ERM) focuses on managing risk while maximizing business value from exploiting opportunities. Using applied ethics as a basis and the perspective that ethics includes both enabling human flourishing and not violating accepted norms, I argue that greater business value is achieved when an organization simultaneously targets the maximization of benefits and the minimization of harms for the …

Go to article

Learning Hierarchical Metrical Structure Beyond Measures, Junyan Jiang, Daniel Chin, Yixiao Zhang, Gus Xia Sep 2022

Learning Hierarchical Metrical Structure Beyond Measures, Junyan Jiang, Daniel Chin, Yixiao Zhang, Gus Xia

Machine Learning Faculty Publications

Music contains hierarchical structures beyond beats and measures. While hierarchical structure annotations are helpful for music information retrieval and computer musicology, such annotations are scarce in current digital music databases. In this paper, we explore a data-driven approach to automatically extract hierarchical metrical structures from scores. We propose a new model with a Temporal Convolutional Network-Conditional Random Field (TCN-CRF) architecture. Given a symbolic music score, our model takes in an arbitrary number of voices in a beat-quantized form, and predicts a 4-level hierarchical metrical structure from downbeat-level to section-level. We also annotate a dataset using RWC-POP MIDI files to facilitate …

Go to article

Led Down The Rabbit Hole: Exploring The Potential Of Global Attention For Biomedical Multi-Document Summarisation, Yulia Otmakhova, Hung Thinh Truong, Timothy Baldwin, Trevor Cohn, Karin Verspoor, Jey Han Lau Sep 2022

Led Down The Rabbit Hole: Exploring The Potential Of Global Attention For Biomedical Multi-Document Summarisation, Yulia Otmakhova, Hung Thinh Truong, Timothy Baldwin, Trevor Cohn, Karin Verspoor, Jey Han Lau

Natural Language Processing Faculty Publications

In this paper we report on our submission to the Multidocument Summarisation for Literature Review (MSLR) shared task. Specifically, we adapt PRIMERA (Xiao et al., 2022) to the biomedical domain by placing global attention on important biomedical entities in several ways. We analyse the outputs of the 23 resulting models, and report patterns in the results related to the presence of additional global attention, number of training steps, and the input configuration. © 2022, CC BY-SA.

Go to article

Unsupervised Lexical Substitution With Decontextualised Embeddings, Takashi Wada, Timothy Baldwin, Yuji Matsumoto, Jey Han Lau Sep 2022

Unsupervised Lexical Substitution With Decontextualised Embeddings, Takashi Wada, Timothy Baldwin, Yuji Matsumoto, Jey Han Lau

Natural Language Processing Faculty Publications

We propose a new unsupervised method for lexical substitution using pre-trained language models. Compared to previous approaches that use the generative capability of language models to predict substitutes, our method retrieves substitutes based on the similarity of contextualised and decontextualised word embeddings, i.e. the average contextual representation of a word in multiple contexts. We conduct experiments in English and Italian, and show that our method substantially outperforms strong baselines and establishes a new state-of-the-art without any explicit supervision or fine-tuning. We further show that our method performs particularly well at predicting low-frequency substitutes, and also generates a diverse list of …

Go to article

Artificial Intelligence-Driven Design Of Fuel Mixtures, Nursulu Kuzhagaliyeva, Samuel Horváth, John Williams, Andre Nicolle, S. Mani Sarathy Sep 2022

Artificial Intelligence-Driven Design Of Fuel Mixtures, Nursulu Kuzhagaliyeva, Samuel Horváth, John Williams, Andre Nicolle, S. Mani Sarathy

Machine Learning Faculty Publications

High-performance fuel design is imperative to achieve cleaner burning and high-efficiency engine systems. We introduce a data-driven artificial intelligence (AI) framework to design liquid fuels exhibiting tailor-made properties for combustion engine applications to improve efficiency and lower carbon emissions. The fuel design approach is a constrained optimization task integrating two parts: (i) a deep learning (DL) model to predict the properties of pure components and mixtures and (ii) search algorithms to efficiently navigate in the chemical space. Our approach presents the mixture-hidden vector as a linear combination of each single component’s vectors in each blend and incorporates it into the …

Go to article

Beat Transformer: Demixed Beat And Downbeat Tracking With Dilated Self-Attention, Jingwei Zhao, Gus Xia, Ye Wang Sep 2022

Beat Transformer: Demixed Beat And Downbeat Tracking With Dilated Self-Attention, Jingwei Zhao, Gus Xia, Ye Wang

Machine Learning Faculty Publications

We propose Beat Transformer, a novel Transformer encoder architecture for joint beat and downbeat tracking. Different from previous models that track beats solely based on the spectrogram of an audio mixture, our model deals with demixed spectrograms with multiple instrument channels. This is inspired by the fact that humans perceive metrical structures from richer musical contexts, such as chord progression and instrumentation. To this end, we develop a Transformer model with both time-wise attention and instrument-wise attention to capture deep-buried metrical cues. Moreover, our model adopts a novel dilated self-attention mechanism, which achieves powerful hierarchical modelling with only linear complexity. …

Go to article

Domain Adversarial Training On Conditional Variational Auto-Encoder For Controllable Music Generation, Jingwei Zhao, Gus Xia, Ye Wang Sep 2022

Domain Adversarial Training On Conditional Variational Auto-Encoder For Controllable Music Generation, Jingwei Zhao, Gus Xia, Ye Wang

Machine Learning Faculty Publications

The variational auto-encoder has become a leading framework for symbolic music generation, and a popular research direction is to study how to effectively control the generation process. A straightforward way is to control a model using different conditions during inference. However, in music practice, conditions are usually sequential (rather than simple categorical labels), involving rich information that overlaps with the learned representation. Consequently, the decoder gets confused about whether to “listen to” the latent representation or the condition, and sometimes just ignores the condition. To solve this problem, we leverage domain adversarial training to disentangle the representation from condition cues …

Go to article

Cmr3d: Contextualized Multi-Stage Refinement For 3d Object Detection, Dhanalaxmi Gaddam, Jean Lahoud, Fahad Shahbaz Khan, Rao Anwer, Hisham Cholakkal Sep 2022

Cmr3d: Contextualized Multi-Stage Refinement For 3d Object Detection, Dhanalaxmi Gaddam, Jean Lahoud, Fahad Shahbaz Khan, Rao Anwer, Hisham Cholakkal

Computer Vision Faculty Publications

Existing deep learning-based 3D object detectors typically rely on the appearance of individual objects and do not explicitly pay attention to the rich contextual information of the scene. In this work, we propose Contextualized Multi-Stage Refinement for 3D Object Detection (CMR3D) framework, which takes a 3D scene as input and strives to explicitly integrate useful contextual information of the scene at multiple levels to predict a set of object bounding-boxes along with their corresponding semantic labels. To this end, we propose to utilize a context enhancement network that captures the contextual information at different levels of granularity followed by a …

Go to article

Decentralized Personalized Federated Learning: Lower Bounds And Optimal Algorithm For All Personalization Modes, Abdurakhmon Sadiev, Ekaterina Borodich, Aleksandr Beznosikov, Darina Dvinskikh, Saveliy Chezhegov, Rachael Tappenden, Martin Takac, Alexander Gasnikov Sep 2022

Decentralized Personalized Federated Learning: Lower Bounds And Optimal Algorithm For All Personalization Modes, Abdurakhmon Sadiev, Ekaterina Borodich, Aleksandr Beznosikov, Darina Dvinskikh, Saveliy Chezhegov, Rachael Tappenden, Martin Takac, Alexander Gasnikov

Machine Learning Faculty Publications

This paper considers the problem of decentralized, personalized federated learning. For centralized personalized federated learning, a penalty that measures the deviation from the local model and its average, is often added to the objective function. However, in a decentralized setting this penalty is expensive in terms of communication costs, so here, a different penalty — one that is built to respect the structure of the underlying computational network — is used instead. We present lower bounds on the communication and local computation costs for this problem formulation and we also present provably optimal methods for decentralized personalized federated learning. Numerical …

Go to article

Self-Supervised Learning For Invariant Representations From Multi-Spectral And Sar Images, Pallavi Jain, Bianca Schoen Phelan, Robert J. Ross Sep 2022

Self-Supervised Learning For Invariant Representations From Multi-Spectral And Sar Images, Pallavi Jain, Bianca Schoen Phelan, Robert J. Ross

Articles

Self-Supervised learning (SSL) has become the new state of the art in several domain classification and segmentation tasks. One popular category of SSL are distillation networks such as Bootstrap Your Own Latent (BYOL). This work proposes RS-BYOL, which builds on BYOL in the remote sensing (RS) domain where data are non-trivially different from natural RGB images. Since multi-spectral (MS) and synthetic aperture radar (SAR) sensors provide varied spectral and spatial resolution information, we utilise them as an implicit augmentation to learn invariant feature embeddings. In order to learn RS based invariant features with SSL, we trained RS-BYOL in two ways, …

Go to article

First Prev Next Last