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

Numerical Analysis and Scientific Computing Commons™

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

Artificial Intelligence and Robotics

Institution
Keyword
Publication Year
Publication
Publication Type
File Type

Articles 3421 - 3450 of 4163

Full-Text Articles in Numerical Analysis and Scientific Computing

Texture-Based Deep Neural Network For Histopathology Cancer Whole Slide Image (Wsi) Classification, Nelson Zange Tsaku Aug 2019

Texture-Based Deep Neural Network For Histopathology Cancer Whole Slide Image (Wsi) Classification, Nelson Zange Tsaku

Master of Science in Computer Science Theses

Automatic histopathological Whole Slide Image (WSI) analysis for cancer classification has been highlighted along with the advancements in microscopic imaging techniques. However, manual examination and diagnosis with WSIs is time-consuming and tiresome. Recently, deep convolutional neural networks have succeeded in histopathological image analysis. In this paper, we propose a novel cancer texture-based deep neural network (CAT-Net) that learns scalable texture features from histopathological WSIs. The innovation of CAT-Net is twofold: (1) capturing invariant spatial patterns by dilated convolutional layers and (2) Reducing model complexity while improving performance. Moreover, CAT-Net can provide discriminative texture patterns formed on cancerous regions of histopathological …

Go to article

Fintech Empowerment: Data Science, Ai, And Machine Learning, Keng Siau, Michael Hilgers, Langtao Chen, Steve Liu, Fiona Fui-Hoon Nah, Richard Hall, Barry Flachsbart Aug 2019

Fintech Empowerment: Data Science, Ai, And Machine Learning, Keng Siau, Michael Hilgers, Langtao Chen, Steve Liu, Fiona Fui-Hoon Nah, Richard Hall, Barry Flachsbart

Research Collection School Of Computing and Information Systems

The article discusses how data science, artificial intelligence and machine learning are affecting the evolution of “fintech,” the technologies used to deliver financial services. After presenting fintech’s competitive advantages in combination with these other advanced technologies, the article posits that financial institutions that don’t move forward with the innovations will be eliminated from the marketplace.

Go to article

Adapting Bert For Target-Oriented Multimodal Sentiment Classification, Jianfei Yu, Jing Jiang Aug 2019

Adapting Bert For Target-Oriented Multimodal Sentiment Classification, Jianfei Yu, Jing Jiang

Research Collection School Of Computing and Information Systems

As an important task in Sentiment Analysis, Target-oriented Sentiment Classification (TSC) aims to identify sentiment polarities over each opinion target in a sentence. However, existing approaches to this task primarily rely on the textual content, but ignoring the other increasingly popular multimodal data sources (e.g., images), which can enhance the robustness of these text-based models. Motivated by this observation and inspired by the recently proposed BERT architecture, we study Target-oriented Multimodal Sentiment Classification (TMSC) and propose a multimodal BERT architecture. To model intra-modality dynamics, we first apply BERT to obtain target-sensitive textual representations. We then borrow the idea from self-attention …

Go to article

Synthetic, Yet Natural: Properties Of Wordnet Random Walk Corpora And The Impact Of Rare Words On Embedding Performance, Filip Klubicka, Alfredo Maldonado, Abhijit Mahalunkar, John D. Kelleher Jul 2019

Synthetic, Yet Natural: Properties Of Wordnet Random Walk Corpora And The Impact Of Rare Words On Embedding Performance, Filip Klubicka, Alfredo Maldonado, Abhijit Mahalunkar, John D. Kelleher

Conference papers

Creating word embeddings that reflect semantic relationships encoded in lexical knowledge resources is an open challenge. One approach is to use a random walk over a knowledge graph to generate a pseudo-corpus and use this corpus to train embeddings. However, the effect of the shape of the knowledge graph on the generated pseudo-corpora, and on the resulting word embeddings, has not been studied. To explore this, we use English WordNet, constrained to the taxonomic (tree-like) portion of the graph, as a case study. We investigate the properties of the generated pseudo-corpora, and their impact on the resulting embeddings. We find …

Go to article

Mathematics And Programming Exercises For Educational Robot Navigation, Ronald I. Greenberg Jul 2019

Mathematics And Programming Exercises For Educational Robot Navigation, Ronald I. Greenberg

Computer Science: Faculty Publications and Other Works

This paper points students towards ideas they can use towards developing a convenient library for robot navigation, with examples based on Botball primitives, and points educators towards mathematics and programming exercises they can suggest to students, especially advanced high school students.

Go to article

Modeling Intra-Relation In Math Word Problems With Different Functional Multi-Head Attentions, Jierui Li, Lei Wang, Jipeng Zhang, Yan Wang, Bing Tian Dai, Dongxiang Zhang Jul 2019

Modeling Intra-Relation In Math Word Problems With Different Functional Multi-Head Attentions, Jierui Li, Lei Wang, Jipeng Zhang, Yan Wang, Bing Tian Dai, Dongxiang Zhang

Research Collection School Of Computing and Information Systems

Several deep learning models have been proposed for solving math word problems (MWPs) automatically. Although these models have the ability to capture features without manual efforts, their approaches to capturing features are not specifically designed for MWPs. To utilize the merits of deep learning models with simultaneous consideration of MWPs’ specific features, we propose a group attention mechanism to extract global features, quantity-related features, quantity-pair features and question-related features in MWPs respectively. The experimental results show that the proposed approach performs significantly better than previous state-of-the-art methods, and boost performance from 66.9% to 69.5% on Math23K with training-test split, from …

Go to article

Meta-Transfer Learning For Few-Shot Learning, Qianru Sun, Yaoyao Liu, Tat-Seng Chua, Bernt Schiele Jun 2019

Meta-Transfer Learning For Few-Shot Learning, Qianru Sun, Yaoyao Liu, Tat-Seng Chua, Bernt Schiele

Research Collection School Of Computing and Information Systems

Meta-learning has been proposed as a framework to address the challenging few-shot learning setting. The key idea is to leverage a large number of similar few-shot tasks in order to learn how to adapt a base-learner to a new task for which only a few labeled samples are available. As deep neural networks (DNNs) tend to overfit using a few samples only, meta-learning typically uses shallow neural networks (SNNs), thus limiting its effectiveness. In this paper we propose a novel few-shot learning method called meta-transfer learning (MTL) which learns to adapt a deep NN for few shot learning tasks. Specifically, …

Go to article

Depressiongnn: Depression Prediction Using Graph Neural Network On Smartphone And Wearable Sensors, Param Bidja May 2019

Depressiongnn: Depression Prediction Using Graph Neural Network On Smartphone And Wearable Sensors, Param Bidja

Honors Scholar Theses

Depression prediction is a complicated classification problem because depression diagnosis involves many different social, physical, and mental signals. Traditional classification algorithms can only reach an accuracy of no more than 70% given the complexities of depression. However, a novel approach using Graph Neural Networks (GNN) can be used to reach over 80% accuracy, if a graph can represent the depression data set to capture differentiating features. Building such a graph requires 1) the definition of node features, which must be highly correlated with depression, and 2) the definition for edge metrics, which must also be highly correlated with depression. In …

Go to article

New Clock-Driven Algorithm Based On Separation Of Synaptic Conductance Computation, Zhijie Wang, Peng Xia, Han Fang, Xiaochun Gu Apr 2019

New Clock-Driven Algorithm Based On Separation Of Synaptic Conductance Computation, Zhijie Wang, Peng Xia, Han Fang, Xiaochun Gu

Journal of System Simulation

Abstract: In order to reduce the computing time when simulating the biologic neural network, an efficient clock-driven algorithm based on the separation of synaptic conductance computation is presented. It is found that the calculation of the synaptic state variables can be separated into two independent parts: one called conductance coefficient related with the pre-synaptic neuron, and the other called synaptic current. By introducing the data structure of the virtual synapse cluster to storing sequences of synaptic conductance coefficient, the former part can be calculated independently according to the spiking states of pre-synaptic neuron at each time step. When calculating the …

Go to article

Grasping Simulation Of Linkage Underactuated Mechanical Finger, Xiali Li, Tianyi Lan, Licheng Wu Apr 2019

Grasping Simulation Of Linkage Underactuated Mechanical Finger, Xiali Li, Tianyi Lan, Licheng Wu

Journal of System Simulation

Abstract: For verifying the design rationality and the property of a linkage underactuated finger, Solidworks is utilized to simulate the grasp operation of the finger with some different objects at the different positions. The simulations can be used to analyze the grasping range, the underactuated characteristic, the uniformity of grasp motion and the mechanical property. The finger mechanism contains springs that brings adaptive grasp capability of different objects. The results show that the finger can make suitable capability for grasping the objects whose size can vary from 0.106 to 0.851 times of the finger length. And the finger can give …

Go to article

Modeling And Simulation Of False Report Filtering Scheme Based On Position In Wireless Sensor Networks, Zhixiong Liu, Limiao Li Apr 2019

Modeling And Simulation Of False Report Filtering Scheme Based On Position In Wireless Sensor Networks, Zhixiong Liu, Limiao Li

Journal of System Simulation

Abstract: In wireless sensor networks, the adversary can inject false reports from compromised nodes. Previous security designs cannot detect faked reports that are forged coordinately by a group of compromised nodes. Furthermore, in sparse areas, some events failed to be reported to sink. This paper proposes a position based filtering scheme (PFS). It derives the optimal coverage degree ω, and the nodes are deployed accordingly. After deployment, each node distributes its position to some other nodes. When a report is generated for an observed event, it must carry t distinct MACs (Message Authentication Codes) along with positions of all detecting …

Go to article

Multi-Chromosome Genetic Algorithm For Multiple Traveling Salesman Problem, Duofu Ye, Liu Gang, He Bing Apr 2019

Multi-Chromosome Genetic Algorithm For Multiple Traveling Salesman Problem, Duofu Ye, Liu Gang, He Bing

Journal of System Simulation

Abstract: A multi-traveling salesman model with time window is established, and two objective functions for the number of traveling salesmen and the sum of travel time are designed. A multi - chromosome coding method is designed to develop complex mutation operator tree, which overcomes the problem of large searching space of traditional genetic algorithms. The performances of algorithms are compared by simulation, and the simulation results show that the genetic algorithm with complex multi-chromosome mutation tree can balance the two objective functions of the number of TSP and total travel time well, improve the algorithm of travel speed, and reduce …

Go to article

Virtual Scene Simulation Of Pilot Response In Taking-Off And Landing Processes Of Carrier-Based Aircraft, Ke Peng, Chenglin Xu, Songyang Liu, Minggao Li, Zhao Xin Apr 2019

Virtual Scene Simulation Of Pilot Response In Taking-Off And Landing Processes Of Carrier-Based Aircraft, Ke Peng, Chenglin Xu, Songyang Liu, Minggao Li, Zhao Xin

Journal of System Simulation

Abstract: A research and implementation method based on biomechanics simulation and virtual scene simulation was put forward to investigate the dynamics responses of pilots during the taking off and landing processes of carrier-based aircraft. The human biomechanics calculation of pilots and 3D virtual scene simulation of flight scene were combined to design and develop the integrated virtual simulation platform. This platform has the functions of data management and analysis, human biological dynamics calculation, synchronous analysis of key results, 3D virtual scene simulation of taking-off and landing processes and human response. The verification was carried out using the experiments from …

Go to article

Modeling And Timing Simulation Of Turbojet Engine Starting Process, Wang Lei, Liying Yang, Hongda Zhang, Yuqing He Apr 2019

Modeling And Timing Simulation Of Turbojet Engine Starting Process, Wang Lei, Liying Yang, Hongda Zhang, Yuqing He

Journal of System Simulation

Abstract: The turbojet engine is complicated in its starting process and requires collaboration and cooperation involving all the execution structures, thus higher demand is put forward for the control system. How to model the engine starting process, and then accurately describe and analyze the impact of actuator states on engine performance directly affect the engine control system performance. According to the working principle of the JetCat-P400 German turbojet starting process, the rotor speed mathematical model of the turbojet engine starting process is constructed by the component method. The parameters of the model are identified by combining the experimental data of …

Go to article

Wind Speed Noise Reduction In Wind Farm Based On Variational Mode Decomposition, Xinghua Xu, Tinglong Pan, Dinghui Wu Apr 2019

Wind Speed Noise Reduction In Wind Farm Based On Variational Mode Decomposition, Xinghua Xu, Tinglong Pan, Dinghui Wu

Journal of System Simulation

Abstract: Wind power forecast is based on the existing data, and the wind speed data in wind power are mixed with different types of noise. In order to improve the precision of prediction, noise reduction is needed. However, the traditional empirical mode decomposition noise reduction method has the phenomenon of mode mixing. To improve the effect of the noise reduction, a kind of noise reduction method based on variational mode decomposition is proposed. The variational mode decomposition is a new method which has good noise immunity and no mode mixing. To thoroughly research the application of variational mode decomposition …

Go to article

Predictive Control For Permanent Magnet Synchronous Motor Based On Imc Observer, Zhiling Ren, Zhongbao Zhang, Limin Hou, Guangquan Zhang, Lin Dong, Zhao Xing Apr 2019

Predictive Control For Permanent Magnet Synchronous Motor Based On Imc Observer, Zhiling Ren, Zhongbao Zhang, Limin Hou, Guangquan Zhang, Lin Dong, Zhao Xing

Journal of System Simulation

Abstract: Predictive control strategy is applied to permanent magnet synchronous motor control system, it can achieve fast dynamic response and high-precision tracking control, but depending on the mathematical model of the motor, the load disturbance will affect the control performance of system. A new control method combining an observer based on internal model control (IMC) and predictive control was presented. The inner-loop current controller and the outer-loop speed controller based separately on model predictive control algorithm and deadbeat current predictive control algorithm were designed to form dual-loop predictive control system. An IMC observer is designed to estimate the load disturbance, …

Go to article

Distribution Routing Optimization Of Fresh Agricultural Products Based On Road Conditions, Wang Heng, Yaxing Xu, Zhenfeng Wang, Tianpeng Zhou, Dechun Tian Apr 2019

Distribution Routing Optimization Of Fresh Agricultural Products Based On Road Conditions, Wang Heng, Yaxing Xu, Zhenfeng Wang, Tianpeng Zhou, Dechun Tian

Journal of System Simulation

Abstract: Reasonable arrangement of distribution route of fresh agricultural product can effectively guarantee the freshness of products, improve the distribution efficiency, and reduce the distribution cost. In practical distribution, the road condition is an important factor affecting the arrangement of the distribution route. According to the different road conditions, the speed characteristic models are set up. Meanwhile, considering the perishable and vulnerable characteristics of fresh agricultural products, the function of time window penalty cost and the function of customer satisfaction are established. Based on the comprehensive consideration of factors, such as road condition, time window and fresh consumption, a multi-objective …

Go to article

Simulation Of Domestic Hot Water Using Behavior Of College Students Based On Group Characteristics, Zhou Jin, Xiaoping Yin, Xu Feng Apr 2019

Simulation Of Domestic Hot Water Using Behavior Of College Students Based On Group Characteristics, Zhou Jin, Xiaoping Yin, Xu Feng

Journal of System Simulation

Abstract: The use of domestic hot water is influenced by user’s gender and living habits, so the composition and behavior characteristics of user group have great influence on the design and operation control of domestic hot water system. The classification of hot water user group of college students was made by means of statistical analysis of original data, and the parameters related to the probability distributions of hot water use behavior were obtained. The simulation of domestic hot water using by college students was conducted and compared with the data generated from original recorded data, and the results show good …

Go to article

Evaluation Of Green Smart Cities In China Based On Entropy Weight - Cloud Model, Chen Li, Haixia Zhang Apr 2019

Evaluation Of Green Smart Cities In China Based On Entropy Weight - Cloud Model, Chen Li, Haixia Zhang

Journal of System Simulation

Abstract: Based on the research on green smart city at home and abroad; and aiming at the shortcomings and deficiencies of traditional evaluation methods, this paper proposes an evaluation method of combining entropy and cloud model based on the cloud model which can realize the conversion of qualitative concept and quantitative value. This method synthetically considers the subjective and objective factors; carries on the correlation analysis to the index; determines the set of evaluation indicators; uses the X-conditional cloud generator in cloud model to obtain the different levels of membership matrix corresponding to each evaluation object; and carries on the …

Go to article

Image Feature Extraction And Online Grading Method For Weight And Shape Of Strawberry, Zhang Qing, Xiangjun Zou, Guichao Lin, Yanhui Sun Apr 2019

Image Feature Extraction And Online Grading Method For Weight And Shape Of Strawberry, Zhang Qing, Xiangjun Zou, Guichao Lin, Yanhui Sun

Journal of System Simulation

Abstract: To deal with the classification problems of strawberry in production, a machine vision based strawberry weight and shape grading method was proposed. The strawberry image was segmented by thresholding to extract the fruit. The area and perimeter parameters of the fruit were then calculated and used to build the strawberry weight grading model through regression analysis. Elliptic Fourier descriptor was used to extract the shape features of the fruit, and these shape features were applied to train a support vector machine (SVM) which represented the strawberry shape grading model. 200 samples of strawberries were selected to test both …

Go to article

Dp-Q(Λ): Real-Time Path Planning For Multi-Agent In Large-Scale Web3d Scene, Fengting Yan, Jinyuan Jia Apr 2019

Dp-Q(Λ): Real-Time Path Planning For Multi-Agent In Large-Scale Web3d Scene, Fengting Yan, Jinyuan Jia

Journal of System Simulation

Abstract: The path planning of multi-agent in an unknown large-scale scene needs an efficient and stable algorithm, and needs to solve multi-agent collision avoidance problem, and then completes a real-time path planning in Web3D. To solve above problems, the DP-Q(λ) algorithm is proposed; and the direction constraints, high reward or punishment weight training methods are used to adjust the values of reward or punishment by using a probability p (0-1 random number). The value from reward or punishment determines its next step path planning strategy. If the next position is free, the agent could walk to it. The above strategy …

Go to article

Design And Realization Of Ship Virtual Fire Training System Based On Hmd, Zhang Bo, Sun Jun, Shang Lei, Hongji Xu, Yang Cheng Apr 2019

Design And Realization Of Ship Virtual Fire Training System Based On Hmd, Zhang Bo, Sun Jun, Shang Lei, Hongji Xu, Yang Cheng

Journal of System Simulation

Abstract: Aiming at the shortcomings of ship fire training in real environment and the advantages and application of virtual reality technology in education and training, the virtual fire training system of ship based on head mount display (HMD) is developed. The system is developed using the Unity 3D Virtual Reality Development Engine as a platform, using client / server mode, supporting multi-person cooperative operation, and running the terminal for HTC Vive virtual reality hardware. The system combines the fire diffuser mechanism, the water shooting hydrodynamic equation and the particle system; completes the model establishment and visualization process, and improves …

Go to article

Systematic Error Analysis Based On Hardware-In-The-Loop Simulation Environment, Zhang Duo, Manguo Liu, Zhang Xiao, Pingan Liu, Yixuan He Apr 2019

Systematic Error Analysis Based On Hardware-In-The-Loop Simulation Environment, Zhang Duo, Manguo Liu, Zhang Xiao, Pingan Liu, Yixuan He

Journal of System Simulation

Abstract: Aiming at the complex and diverse components in the semi-physical simulation environment of a certain type of anti-tank missile, the errors present different types of problems, and a systematic error analysis method based on the environment is proposed. By analyzing the composition characteristics and the error characteristics of the simulation computer, flight motion simulator, target simulator and simulation dedicated interface; the delay error of the simulation computer, the mechanical installation error of the flight motion simulator, the delay error of the target simulator and the transmission error of the simulation dedicated interface are acquired. Furthermore, by comparing the local …

Go to article

Design And Simulation Of Heterogeneous Magnetron Memristive Chua Circuit, Xiaohong Zhang, Lijuan Wan, Keliu Long Apr 2019

Design And Simulation Of Heterogeneous Magnetron Memristive Chua Circuit, Xiaohong Zhang, Lijuan Wan, Keliu Long

Journal of System Simulation

Abstract: A nonlinear circuit containing two heterogeneous magnetron memristors which based on the three-order classic Chua circuit is designed, its dimensionless fifth-order mathematical model is proposed. Due to the difference in the design of the characteristic equation of the two magnetron memristors, their positions form a symmetrical structure with respect to the capacitor. The existence of chaotic properties is proved by analyzing the stability of the system equilibrium point and Lyapunov exponents. The model of five-order heterogeneous memristive Chua system and two heterogeneous magnetron memristors are implemented by Multisim. Numerical calculation and experimental results show that heterogeneous magnetron memristive Chua …

Go to article

Active Disturbance Rejection Sliding Mode Control For Time-Delay Systems, Suzhen Wang, Guofa Sun, Shengrong Liu Apr 2019

Active Disturbance Rejection Sliding Mode Control For Time-Delay Systems, Suzhen Wang, Guofa Sun, Shengrong Liu

Journal of System Simulation

Abstract: Aiming at the time-delay systems with uncertainty, this paper designs a third-order nonlinear feedback extended state observer (ESO) by using the coordinate transformation. The ESO can estimate both the system states and the total disturbance signal, and the nonlinear feedback has fast convergence. By the first-order approximation of the time-delay, the output feedback control problem of the time-delay system is transformed into controlling the second-order nonlinear uncertain system with unknown function. On the basis of the tracking differentiator (TD) and the designed ESO, the second-order sliding mode controller (SMC) is designed to realize the fast and accurate tracking …

Go to article

Process Simulation And Optimization On Giant Block Erection Of Longitudinal Staggered Arrangement, Wenmin Zhu, Minghua Zhu, Xiumin Fan Apr 2019

Process Simulation And Optimization On Giant Block Erection Of Longitudinal Staggered Arrangement, Wenmin Zhu, Minghua Zhu, Xiumin Fan

Journal of System Simulation

Abstract: The process of giant block erection of longitudinal staggered arrangement is complicated. If the design of process and tooling are not reasonable, the adjustment work in the construction site, which affects the shipbuilding cycle seriously, will be abundant. To solve these problems, the method of process simulation and optimization on giant block erection were researched. Simulation models were established for carrying out the process simulation, and the simulation process was detected in real time by establishing collision detection model and virtual human model. Based on the measured data of the ship hull, the automatic calculation method of raising height …

Go to article

Robust Adaptive Rate Control Algorithm And Its Simulation For P2p-Tv Streaming Media Systems, Fengjie Yin, Yang Hui, Zhang Ying Apr 2019

Robust Adaptive Rate Control Algorithm And Its Simulation For P2p-Tv Streaming Media Systems, Fengjie Yin, Yang Hui, Zhang Ying

Journal of System Simulation

Abstract: A control algorithm proposed to control the upstream rate based on sliding mode control in P2P-TV streaming media node. Through the rational design of sliding mode controller, the algorithm can adaptively control the numbers of signaling threads between neighbor pairs connected simultaneously and then control the upload rate. Further, it can achieve the fair use of resources. Due to the sliding mode control has strong robust with modeling uncertainties, time varying parameter fluctuations and external disturbances, it can avoid the system congestion under network condition change. Simulation results demonstrate that it can lead to the convergence of the …

Go to article

Modeling And Simulation Of Health Degradation Trend For Wind Turbine Bearing, Xinghui Dong, Xiaoshuang Ma, Youxing Cheng, Wang Shuai Apr 2019

Modeling And Simulation Of Health Degradation Trend For Wind Turbine Bearing, Xinghui Dong, Xiaoshuang Ma, Youxing Cheng, Wang Shuai

Journal of System Simulation

Abstract: By taking a wind turbine bearing as research object, the model of bearing temperature health’s degradation trend is established through using least squares surface fitting and the monitored parameters from Supervisory Control And Data Acquisition (SCADA). Bearings’ degradation trend with unsteady characteristics is decomposed by modified Ensemble Empirical Mode Decomposition(EEMD) to obtain several relatively steady components. Components are predicted respectively by time series neural network and the predicted results of all the components are added to obtain final prediction result. Comprehensive simulations and comparisons show that the proposed method can predict the health degradation trend of wind turbine bearings …

Go to article

Soil Data Construction Method For Cross-Country Trafficability Analysis, Kunwei Li, You Xiong, Zhang Xin, Fen Tang Apr 2019

Soil Data Construction Method For Cross-Country Trafficability Analysis, Kunwei Li, You Xiong, Zhang Xin, Fen Tang

Journal of System Simulation

Abstract: Soil is one of the most important factors influencing the off-road maneuver of the army. The combination of soil and weather factors makes cross-country trafficability analysis extremely complicated. The soil data from unified soil classification system (USCS) are the basis for the analysis of soil trafficability. In this paper, the random forest method is used to predict the type of soil by using various attribute information. The method extracts sample data from existing USCS soil data to construct multiple random forest models, then analyses the accuracy of random forests and the importance of characteristic variables, and finally uses the …

Go to article

Confusion Prediction From Eye-Tracking Data: Experiments With Machine Learning, Joni Salminen, Mridul Nagpal, Haewoon Kwak, Jisun An, Soon-Gyo Jung, Bernard J. Jansen Mar 2019

Confusion Prediction From Eye-Tracking Data: Experiments With Machine Learning, Joni Salminen, Mridul Nagpal, Haewoon Kwak, Jisun An, Soon-Gyo Jung, Bernard J. Jansen

Research Collection School Of Computing and Information Systems

Predicting user confusion can help improve information presentation on websites, mobile apps, and virtual reality interfaces. One promising information source for such prediction is eye-tracking data about gaze movements on the screen. Coupled with think-aloud records, we explore if user's confusion is correlated with primarily fixation-level features. We find that random forest achieves an accuracy of more than 70% when prediction user confusion using only fixation features. In addition, adding user-level features (age and gender) improves the accuracy to more than 90%. We also find that balancing the classes before training improves performance. We test two balancing algorithms, Synthetic Minority …

Go to article

First Prev Next Last