Deep Learning Algorithms and Their Applications in Speech Signal Processing and Recognition
- 2016-08-15 (Mon.), 10:30 AM
- Recreation Hall, 2F, Institute of Statistical Science
- Dr. Yu Tsao
- Research Center for Information Technology Innovation, Academia Sinica
Abstract
Pattern recognition and machine learning have become indispensable tools for managing big data from diverse sources such as images, text, speech, and music. Recently, deep learning algorithms have been applied to a number of different areas including speech recognition, object recognition, and language understanding. Deep learning allows computers to represent high-level concepts from very low-level data. This hierarchical deep structure mimics human cognitive functions such as those involved in vision perception. The advantage of deep learning lies in its ability to perform unsupervised feature learning, which will prove invaluable for integrating vast amounts of unlabeled data. Deep learning has led to remarkable performance improvements in the field of speech recognition, allowing systems to automatically learn the most important features in speech for determining phonemes and likely utterances. In this talk, we will introduce our recent research accomplishments on deep learning algorithms and their applications in speech signal processing and pattern recognition tasks.