Using Hidden Markov Model to Predict the Potential Intent of User's Gaze Behavior

Study between visual gaze behavior and implied intent, it provides a new idea for exploring the human-computer interaction mode of non-verbal communication. Applying hidden Markov model to implicit intention inference of visual gaze, find predictive models for both efficiency and accuracy, reduction...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2021 International Conference on Machine Learning and Intelligent Systems Engineering (MLISE) S. 38 - 41
Hauptverfasser: Guogang, Gaozheng, Wushiqian, Yumin
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.07.2021
Schlagworte:
component
Hidden Markov Model
Hidden Markov models
Human computer interaction
Markov processes
Predicted intent
Predictive models
Training
Visual gaze
Visualization
Viterbi algorithm
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Study between visual gaze behavior and implied intent, it provides a new idea for exploring the human-computer interaction mode of non-verbal communication. Applying hidden Markov model to implicit intention inference of visual gaze, find predictive models for both efficiency and accuracy, reduction of lag in training models using historical data. The subjects' gaze data were collected, based on the hidden Markov model, two different gaze patterns are constructed, the model parameters are trained by Baum- Welch algorithm, next, the viterbi algorithm is used to solve the maximum probability hidden state sequence, then, the implicit intent prediction of the gaze behavior. On this basis, the architecture of the model and its effectiveness are verified, the relationship between human intention and gaze behavior is further discussed.
DOI:10.1109/MLISE54096.2021.00015