Deep learning is a new area of machine learning research that allows deep neural networks composed of multiple processing layers to learn representations of data with multiple levels of abstraction. Deep learning has helped in achieving the objective of pushing machine learning closer to one of its original goals of artificial intelligence. It has become state-of-the-art machine learning technique in the fields of computer vision, speech recognition and text processing. Although it has enjoyed great success in the fields of computer science, its application in scientific fields has been very limited. This is mainly due to the scarcity of and complex nature of scientific datasets since they are collected from expensive and time-consuming scientific experiments and computations. This thesis explores how to design and build novel deep neural network architectures that can handle the challenges associated with such datasets and automatically learn the underlying science behind those scientific phenomena using deep learning, for the advancement of the overall process of scientific knowledge discovery.

Creator

• Jha, Dipendra Kumar

DOI

• https://doi.org/10.21985/n2-hh5t-yw69

Subject

• Computer engineering

Language

• en

Alternate Identifier

• etdadmin_upload_711601
• http://dissertations.umi.com/northwestern:14960

Keyword

• deep neural networks
• machine learning
• deep learning
• artificial intelligence

Date created

• 2019-01-01

Resource type

• Dissertation

Rights statement

• In Copyright