eCommons

 

AUTOMATED MACHINE LEARNING AND TASK SPACE NAVIGATION

Other Titles

Author(s)

Abstract

Machine learning has become almost ubiquitous in our lives. Training these models have even entered mainstream with the advent of efficient deep learning frameworks and advancements in hardware innovations. There are, however, two noteworthy limitations that prevent practitioners outside the machine learning community from successfully training models that achieve the desirable performance. The first one is the heavy requirement of expert knowledge. There are many components around building a successful machine learning model that each one could impact its final performance. Selecting these components is not easy: it requires trials and errors and a lot of human knowledge! The second is the inefficiency of training deep learning models, especially in the problem picking a good pretraiend feature extractor for a given task. This work tackles both of these problems by incorporating AutoML techniques. As introduced and detailed in later chapters, these frameworks could effectively automate many processes and complexity within training machine learning models.

Journal / Series

Volume & Issue

Description

79 pages

Sponsorship

Date Issued

2021-05

Publisher

Keywords

Location

Effective Date

Expiration Date

Sector

Employer

Union

Union Local

NAICS

Number of Workers

Committee Chair

Hariharan, Bharath
Udell, Madeleine Richards

Committee Co-Chair

Committee Member

Degree Discipline

Computer Science

Degree Name

M.S., Computer Science

Degree Level

Master of Science

Related Version

Related DOI

Related To

Related Part

Based on Related Item

Has Other Format(s)

Part of Related Item

Related To

Related Publication(s)

Link(s) to Related Publication(s)

References

Link(s) to Reference(s)

Previously Published As

Government Document

ISBN

ISMN

ISSN

Other Identifiers

Rights

Rights URI

Types

dissertation or thesis

Accessibility Feature

Accessibility Hazard

Accessibility Summary

Link(s) to Catalog Record