Artificial Intelligence and Robotics Commons^™

Language-Driven Region Pointer Advancement For Controllable Image Captioning, Annika Lindh, Robert J. Ross, John D. Kelleher

Conference papers

Controllable Image Captioning is a recent sub-field in the multi-modal task of Image Captioning wherein constraints are placed on which regions in an image should be described in the generated natural language caption. This puts a stronger focus on producing more detailed descriptions, and opens the door for more end-user control over results. A vital component of the Controllable Image Captioning architecture is the mechanism that decides the timing of attending to each region through the advancement of a region pointer. In this paper, we propose a novel method for predicting the timing of region pointer advancement by treating the …

Nearest Centroid: A Bridge Between Statistics And Machine Learning, Manoj Thulasidas

Research Collection School Of Computing and Information Systems

In order to guide our students of machine learning in their statistical thinking, we need conceptually simple and mathematically defensible algorithms. In this paper, we present the Nearest Centroid algorithm (NC) algorithm as a pedagogical tool, combining the key concepts behind two foundational algorithms: K-Means clustering and K Nearest Neighbors (k- NN). In NC, we use the centroid (as defined in the K-Means algorithm) of the observations belonging to each class in our training data set and its distance from a new observation (similar to k-NN) for class prediction. Using this obvious extension, we will illustrate how the concepts of …

Interventional Few-Shot Learning, Zhongqi Yue, Zhang Hanwang, Qianru Sun, Xian-Sheng Hua

Research Collection School Of Computing and Information Systems

We uncover an ever-overlooked deficiency in the prevailing Few-Shot Learning (FSL) methods: the pre-trained knowledge is indeed a confounder that limits the performance. This finding is rooted from our causal assumption: a Structural Causal Model (SCM) for the causalities among the pre-trained knowledge, sample features, and labels. Thanks to it, we propose a novel FSL paradigm: Interventional Few-Shot Learning (IFSL). Specifically, we develop three effective IFSL algorithmic implementations based on the backdoor adjustment, which is essentially a causal intervention towards the SCM of many-shot learning: the upper-bound of FSL in a causal view. It is worth noting that the contribution …

Causal Intervention For Weakly-Supervised Semantic Segmentation, Zhang Dong, Hanwang Zhang, Jinhui Tang, Xian-Sheng Hua, Qianru Sun

Research Collection School Of Computing and Information Systems

We present a causal inference framework to improve Weakly-Supervised Semantic Segmentation (WSSS). Specifically, we aim to generate better pixel-level pseudo-masks by using only image-level labels --- the most crucial step in WSSS. We attribute the cause of the ambiguous boundaries of pseudo-masks to the confounding context, e.g., the correct image-level classification of "horse'' and "person'' may be not only due to the recognition of each instance, but also their co-occurrence context, making the model inspection (e.g., CAM) hard to distinguish between the boundaries. Inspired by this, we propose a structural causal model to analyze the causalities among images, contexts, and …

Artificial Intelligence For Social Impact: Learning And Planning In The Data-To-Deployment Pipeline, Andrew Perrault, Fei Fang, Arunesh Sinha, Milind Tambe

Research Collection School Of Computing and Information Systems

With the maturing of artificial intelligence (AI) and multiagent systems research, we have a tremendous opportunity to direct these advances toward addressing complex societal problems. In pursuit of this goal of AI for social impact, we as AI researchers must go beyond improvements in computational methodology; it is important to step out in the field to demonstrate social impact. To this end, we focus on the problems of public safety and security, wildlife conservation, and public health in low-resource communities, and present research advances in multiagent systems to address one key cross-cutting challenge: how to effectively deploy our limited intervention …

Lightweight Deep Learning For Botnet Ddos Detection On Iot Access Networks, Eric A. Mccullough

Graduate Theses/Dissertations

With the proliferation of the Internet of Things (IoT), computer networks have rapidly expanded in size. While Internet of Things Devices (IoTDs) benefit many aspects of life, these devices also introduce security risks in the form of vulnerabilities which give hackers billions of promising new targets. For example, botnets have exploited the security flaws common with IoTDs to gain unauthorized control of hundreds of thousands of hosts, which they then utilize to carry out massively disruptive distributed denial of service (DDoS) attacks. Traditional DDoS defense mechanisms rely on detecting attacks at their target and deploying mitigation strategies toward the attacker …

Development Of Computational To Ols To Target Microrna, Luo Song

Dissertations and Theses (Open Access)

MicroRNAs (a.k.a, miRNAs) play an important role in disease development. However, few of their structures have been determined and structure-based computational methods remain challenging in accurately predicting their interactions with small molecules. To address this issue, my thesis is to develop integrated approaches to screening for novel inhibitors by targeting specific structure motifs in miRNAs. The project starts with implementing a tool to find potential miRNA targets with desired motifs. I combined both sequence information of miRNAs and known RNA structure data from Protein Data Bank (PDB) to predict the miRNA structure and identify the motif to target, then I …

Attentional Parsing Networks, Marcus Karr

Master's Theses

Convolutional neural networks (CNNs) have dominated the computer vision field since the early 2010s, when deep learning largely replaced previous approaches like hand-crafted feature engineering and hierarchical image parsing. Meanwhile transformer architectures have attained preeminence in natural language processing, and have even begun to supplant CNNs as the state of the art for some computer vision tasks.

This study proposes a novel transformer-based architecture, the attentional parsing network, that reconciles the deep learning and hierarchical image parsing approaches to computer vision. We recast unsupervised image representation as a sequence-to-sequence translation problem where image patches are mapped to successive layers …

Heterogeneous Univariate Outlier Ensembles In Multidimensional Data, Guansong Pang, Longbing Cao

Research Collection School Of Computing and Information Systems

In outlier detection, recent major research has shifted from developing univariate methods to multivariate methods due to the rapid growth of multidimensional data. However, one typical issue of this paradigm shift is that many multidimensional data often mainly contains univariate outliers, in which many features are actually irrelevant. In such cases, multivariate methods are ineffective in identifying such outliers due to the potential biases and the curse of dimensionality brought by irrelevant features. Those univariate outliers might be well detected by applying univariate outlier detectors in individually relevant features. However, it is very challenging to choose a right univariate detector …

A Study Of Multi-Task And Region-Wise Deep Learning For Food Ingredient Recognition, Jingjing Chen, Bin Zhu, Chong-Wah Ngo, Tat-Seng Chua, Yu-Gang Jiang

Research Collection School Of Computing and Information Systems

Food recognition has captured numerous research attention for its importance for health-related applications. The existing approaches mostly focus on the categorization of food according to dish names, while ignoring the underlying ingredient composition. In reality, two dishes with the same name do not necessarily share the exact list of ingredients. Therefore, the dishes under the same food category are not mandatorily equal in nutrition content. Nevertheless, due to limited datasets available with ingredient labels, the problem of ingredient recognition is often overlooked. Furthermore, as the number of ingredients is expected to be much less than the number of food categories, …

Enhanced Traffic Incident Analysis With Advanced Machine Learning Algorithms, Zhenyu Wang

Computational Modeling & Simulation Engineering Theses & Dissertations

Traffic incident analysis is a crucial task in traffic management centers (TMCs) that typically manage many highways with limited staff and resources. An effective automatic incident analysis approach that can report abnormal events timely and accurately will benefit TMCs in optimizing the use of limited incident response and management resources. During the past decades, significant efforts have been made by researchers towards the development of data-driven approaches for incident analysis. Nevertheless, many developed approaches have shown limited success in the field. This is largely attributed to the long detection time (i.e., waiting for overwhelmed upstream detection stations; meanwhile, downstream stations …

Goamlp: Network Intrusion Detection With Multilayer Perceptron And Grasshopper Optimization Algorithm, Farshid Bagheri Saravi

Student Scholarship

In this paper, an intrusion detection system is introduced that uses data mining and machine learning concepts to detect network intrusion patterns. In the proposed method, an artificial neural network (ANN) is used as a learning technique in intrusion detection. The metaheuristic algorithm with the swarm-based approach is used to reduce intrusion detection errors. In the proposed method, the Grasshopper Optimization Algorithm (GOA) is used for better and more accurate learning of ANNs to reduce intrusion detection error rate. The role of the GOAMLP algorithm is to minimize the intrusion detection error in the neural network by selecting useful parameters …

Computational Cognition And Deep Learning, Andy Malinsky

The Compass

No abstract provided.

Energy-Based Neural Modelling For Large-Scale Multiple Domain Dialogue State Tracking, Anh Duong Trinh, Robert J. Ross, John D. Kelleher

Conference papers

Scaling up dialogue state tracking to multiple domains is challenging due to the growth in the number of variables being tracked. Furthermore, dialog state tracking models do not yet explicitly make use of relationships between dialogue variables, such as slots across domains. We propose using energy-based structure prediction methods for large-scale dialogue state tracking task in two multiple domain dialogue datasets. Our results indicate that: (i) modelling variable dependencies yields better results; and (ii) the structured prediction output aligns with the dialogue slot-value constraint principles. This leads to promising directions to improve state-of-the-art models by incorporating variable dependencies into their …

Autonomous Vehicles And The Ethical Tension Between Occupant And Non-Occupant Safety, Jason Borenstein, Joseph Herkert, Keith Miller

The Journal of Sociotechnical Critique

Given that the creation and deployment of autonomous vehicles is likely to continue, it is important to explore the ethical responsibilities of designers, manufacturers, operators, and regulators of the technology. We specifically focus on the ethical responsibilities surrounding autonomous vehicles that these stakeholders have to protect the safety of non-occupants, meaning individuals who are around the vehicles while they are operating. The term “non-occupants” includes, but is not limited to, pedestrians and cyclists. We are particularly interested in how to assign moral responsibility for the safety of non-occupants when autonomous vehicles are deployed in a complex, land-based transportation system.

Machine Learning Augmentation Micro-Sensors For Smart Device Applications, Mohammad H. Hasan

Department of Mechanical and Materials Engineering: Dissertations, Theses, and Student Research

Novel smart technologies such as wearable devices and unconventional robotics have been enabled by advancements in semiconductor technologies, which have miniaturized the sizes of transistors and sensors. These technologies promise great improvements to public health. However, current computational paradigms are ill-suited for use in novel smart technologies as they fail to meet their strict power and size requirements. In this dissertation, we present two bio-inspired colocalized sensing-and-computing schemes performed at the sensor level: continuous-time recurrent neural networks (CTRNNs) and reservoir computers (RCs). These schemes arise from the nonlinear dynamics of micro-electro-mechanical systems (MEMS), which facilitates computing, and the inherent ability …

New Methods For Deep Learning Based Real-Valued Inter-Residue Distance Prediction, Jacob Barger

Theses

Background: Much of the recent success in protein structure prediction has been a result of accurate protein contact prediction--a binary classification problem. Dozens of methods, built from various types of machine learning and deep learning algorithms, have been published over the last two decades for predicting contacts. Recently, many groups, including Google DeepMind, have demonstrated that reformulating the problem as a multi-class classification problem is a more promising direction to pursue. As an alternative approach, we recently proposed real-valued distance predictions, formulating the problem as a regression problem. The nuances of protein 3D structures make this formulation appropriate, allowing predictions …

Under Impact Of Non-Motor Vehicle Violation An Analysis On Vehicle Operation Efficiency, Sun Di, Zhou Jin, Sijia Liu, Xiaoming Zhang, Xueying Gao

Journal of System Simulation

Abstract: In order to study the impact of non-motor vehicle violations on motor vehicle traffic efficiency at signalized intersections, the non-motor vehicle traffic behaviors are analyzed. An actual intersection is selected as study object, the main violation behaviors of on selected intersection are analyzed by using the linear regression model. For traffic light violation behavior of non-motor vehicles, the violation rate is modeled by using logistic model, and the results are analyzed. At the signal controlled intersection, non-motor vehicle violations affect the normal motor vehicles, and time is delayed. The delay time of motor vehicle under non-motor vehicle …

Time-Varying Parameter System Modeling Method Based On Zonotope-Ellipsoid Double Filtering, Ziyun Wang, Peiyu Wang, Yacong Zhan

Journal of System Simulation

Abstract: The traditional system modeling method using zonotopes as the feasible parameter sets islikely to increase the computational complexity of the algorithm due to the increasing dimensions of the zonotope shape matrix. This paper proposes a time-varying parameter modeling systems method based on zonotope-ellipsoid double filtering technique. Considering the time-varying parameters, a zonotope with the minimum volume is obtained during the iterations of the intersection with the constraint strip. After transforming the shape matrix of the zonotope, the dimensionality reduction is performed, instead of directly finding the row sum of the extended shape matrix, to reduce the algorithm conservativeness originated …

Research On Configurable Simulation Integration Technology For Equipment Software, Yuanyuan Wang, Yuxin Duan, Guangzhao Song

Journal of System Simulation

Abstract: The problems of the traditional distributed digital simulation system are analyzed. The method of constructing the digital simulation system based on the equipment software is studied. The configurable simulation integration middleware is designed. The overall structure of the configurable middleware is designed and the design ideas of each module are briefly summarized. An example of a digital simulation system with networked transformation of equipment shows that the middleware can be effectively and flexibly configured and can provide support for constructing digital simulation system with software, which can be promoted and used in other simulation systems.

Research On Multi-Objective Optimization Method Based On Model, Jianjun Liu, Guangya Si, Yanzheng Wang, Dachuan He

Journal of System Simulation

Abstract: There is a model-based algorithm for the optimization of multiple objective functions by means of black-box evaluation is proposed. The algorithm iteratively generates candidate solutions from a mixture distribution over the solution space and updates the mixture distribution based on the sampled solutions’ domination count, such that the future search is biased towards the set of Pareto optimal solutions. The proposed algorithm seeks to find a mixture distribution on the solution space so that each component of the mixture distribution is a degenerate distribution centered at a Pareto optimal solution and each estimated Pareto optimal solution is uniformly spread …

Research And Application Of A Lightweight Real-Time Human Posture Detection Model, Hongkun Zhu, Jiawei Yin, Wenyu Feng, Hua Liang, Minrui Fei, Kun Zhang

Journal of System Simulation

Abstract: The traditional OpenPose model has good accuracy but slow speed in human posture detection. In order to accelerate the detection speed and reduce the model on condition of the detection precision, based on the traditional OpenPose model, the residual network with second-order term fusion is used to extract the low-level features, the weights of the trained model are pruned by the L1 norm weight, and an improved OpenPose model is proposed. Experiments show that when the detection accuracy is approximately equal to original model, the model size reduces to about 8%, the parameters reduces by nearly 83%, and the …

Construction And Test Method Of A Semi-Physical Simulation System For Laser Driving Guidance Weapon, Zhang Xiang, Mengyan Liu, Zhang Peng, Kewei Zhu, Xiaodong Yan

Journal of System Simulation

Abstract: To achieve the previous semi-physical simulation of the laser beam-guided weapon, the two-dimensional translation system must be used. But the technical index of the two-dimensional translation system is not high enough to meet the requirements of the relative motion simulation speed and acceleration of the projectile. To solve the problem, a new method and test method for the semi-physical simulation system of laser driving guidance weapon is proposed. The “two-axis turret + load bracket” simulation system construction method is adopted to replace the linear motion of the two-dimensional translation system by the rotational angular motion of the …

Two Gd Atoms Adsorbed On Zigzag Graphene Nanoribbon:A First-Principles Study, Weifeng Xie, Zuo Xu

Journal of System Simulation

Abstract: A giant Rashba-type spin splitting is highly critical for the application of spintronics, but one-dimensional magnetic systems are rarely involved. In order to explore the characteristics and strength of Rashba effect in one-dimensional magnetic systems, two Gd atoms adsorbed on Zigzag graphene nanoribbon system is proposed. The characteristics and strength of the Rashba effect in different magnetization directions and the magnetic anisotropy of the system are analyzed through the first-principles calculations. The results show that the antiferromagnetic ground state system has strong Rashba strength and out-of-plane magnetic anisotropy. In addition, the Rashba effect in magnetic system needs to …

Generalized Zero-Inflated Binomial Distribution Model Aimed At Air Quality Data Analysis, Benyue Su, Pengpeng Xu, Sheng Min

Journal of System Simulation

Abstract: For the problem of the quality monitoring and counting of excessive gas emissions in chemical industry parks, a generalized zero-inflated binomial distribution model is constructed. Statistics show that the times of number of excessive gas emissions has a typical zero-inflated feature. The traditional zero-inflated Poisson model and negative binomial regression model and so on will underestimate the probability of zero inflation. A generalized zero-inflated binomial distribution model is constructed by extending the traditional binomial regression model to a more general form. This model satisfies the characteristic that the expectation is less than the variance, and better solves the problems …

Hyper-Heuristic De Algorithm For Solving Zero-Wait Fermentation Process Schedulinge, Shen Peng, Wang Yan, Zhicheng Ji, Jianhua Zhang

Journal of System Simulation

Abstract: A class of zero-wait fermentation process scheduling issues with batch process characteristics are researched. In order to solve the problem of easy deterioration in the process, a super heuristic difference algorithm is proposed, and the maximum makespan is minimized as the optimization goal. The algorithm is divided into two layers. The upper layer is an improved adaptive differential evolution algorithm to select and sort the heuristic operations in lower layer. The lower layer is combined and sorted into a new algorithm to operate on the problem domain, adding simulated annealing algorithm to avoid falling into local optimization. The method …

Modeling And Relevance Analysis Of Urban Epidemic Transmission And Work Resumption Intensity, Gan Mi, Yunyi Tian, Wenchang Zhang, Xihan Zhao

Journal of System Simulation

Abstract: On the basis of multi-source big data, the model for analyzing the population migration changes and manpower gaps caused by the COVID-19 epidemic in 34 typical cities across the country is constructed , and the work resumption intensity of other cities is predicited by using the migration base constructed. The SEIR model is used to estimate the basic reproduction number in each city since the simulate results show that it can emulate the transmission trend of this epidemic accurately, and the retrospective matrix analysis of the work resumption intensity is combined with the manpower gap to summarize the anti-epidemic …

Research On Dynamic Flexible Job Shop Scheduling Problem Based On Dynamic Interaction Layer, Zhang Xiang, Wang Yan, Zhicheng Ji

Journal of System Simulation

Abstract: In order to quickly response to the unforeseen circumstances in flexible job shop, a dynamic flexible job shop scheduling model is constructed, which takes the overall production time and the completion time of emergency orders as the optimization objectives. For the model, a dynamic interaction layer (DIL) model, which has a better performance on DFJSP, is proposed to replace the scroll window. Particle swarm genetic hybrid algorithm (PSGA) is designed to combine the particle swarm optimization algorithm with the genetic algorithm to enhance the ability of local search. Aiming at the unexpected urgent orders in flexible job shop, …

Echo Simulation And Verification Of High Resolution Range Profile, Xiaolin Li, Cheng Yu, Haifei Zang, Shuge Wang, Liu Li, Qingqing Yuan

Journal of System Simulation

Abstract: In order to realize the echo simulation and verification of high-resolution range profile in laboratory environment, the multi-scattering point model and wideband LFM echo signal model are given, and the time-domain convolution and high-precision delay realization methods in echo simulation process are described. According to the different signal bandwidth forms of the tested equipments, a wideband echo simulator is used to realize the corresponding high-resolution range image echo. By comparing with the digital simulation results of the target characteristic modeling software, the fidelity of the high-resolution range echo simulation is verified. Simulation verification is carried out with an aircraft …

Multi-Agent Simulation Model For Covid-19 Virus Prevention And Control, Lihu Pan, Shipeng Qin, Xiaowen Li, Feiping Lu, Fenyu Yang

Journal of System Simulation

Abstract: The prevention and control of the novel coronavirus (COVID-19) is the priority work to maintain the public health security of the world nowadays. The COVID-19 prevention and control model using multi-agent modeling and simulation technology is proposed. The model can simulate the different dynamic development trend of the epidemic under different prevention and control measures. Taking Taiyuan as an example, according to the researched COVID-19 transmission rules, the prevention and control simulation of COVID-19 has been achieved under the designing rule of the interactive infection process and status transition process between various resident agents. Multi-scenario simulation experiments are realized …