Deep learning

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering

Final Year Thesis Oral Presentation

Title: "Deep learning"

By

Minsam KIM

Abstract

The major breakthrough of deep-learning was a variety of pre-training
methods for initialization of the network before fine-tuning of the
weights in a supervised manner. This paper discusses about application of
such deep learning techniques for univariate time-series analysis, with
focus on forecasting. The control setup uses deep networks that are not
pre-trained, in order to quantify how different pre-training methods
improve the network performances. Appropriate modifications to different
algorithms were made in order to make the experiments particularly
effective for time-series data.

Date:                   Wednesday, 6 May 2015

Time:                   3:10 - 3:50pm

Venue:                  Room 5503
                        Lifts 25/26

Committee Members:      Prof. James Kwok (Supervisor)
                        Dr. Brian Mak (Reader)