Abstract

Human activity recognition has been an active research area in recent years. The difficulty of this problem lies in the complex dynamical motion patterns embedded through the sequential frames. The Long Short-Term Memory (LSTM) recurrent neural network is capable of processing complex sequential information since it utilizes special gating schemes for learning representations from long input sequences. It has the potential to model various time-series data, where the current hidden state has to be considered in the context of the past hidden states. Unfortunately, the conventional LSTMs do not consider the impact of spatio-temporal dynamics corresponding to the given salient motion patterns, when they gate the information that ought to be memorized through time. To address this problem, we propose a differential gating scheme for the LSTM neural network, which emphasizes the change in information gain caused by the salient motions between the successive video frames. This change in information gain is quantified by Derivative of States (DoS), and thus the proposed LSTM model is termed differential Recurrent Neural Network (dRNN). Based on the energy profiling of DoS, we further propose to employ the State Energy Profile (SEP) to search for salient dRNN states and construct more informative representations. To better understand the scene and human appearance information, the dRNN model is extended by connecting Convolutional Neural Networks (CNN) and stacked dRNNs into an end-to-end model. Lastly, the dissertation continues to discuss and compare the combined and the individual orders of DoS used within the dRNN. We propose to control the LSTM gates via individual order of DoS and stack multiple levels of LSTM cells in increasing orders of state derivatives. To this end, we have introduced a new family of LSTMs, expanding the applications of LSTMs and advancing the performances of the state-of-the-art methods.

Notes

If this is your thesis or dissertation, and want to learn how to access it or for more information about readership statistics, contact us at STARS@ucf.edu

Graduation Date

2020

Semester

Spring

Advisor

Hua, Kien

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

Computer Science

Degree Program

Computer Science

Format

application/pdf

Identifier

CFE0008069; DP0023208

URL

https://purls.library.ucf.edu/go/DP0023208

Language

English

Release Date

May 2020

Length of Campus-only Access

None

Access Status

Doctoral Dissertation (Open Access)

Share

COinS