Abstract
Domain adaptation deals with adapting classifiers trained on data from a source distribution, to work effectively on data from a target distribution. In this paper, we introduce the Nonlinear Embedding Transform (NET) for unsupervised domain adaptation. The NET reduces cross-domain disparity through nonlinear domain alignment. It also embeds the domain- aligned data such that similar data points are clustered together. This results in enhanced classification. To determine the parameters in the NET model (and in other unsupervised domain adaptation models), we introduce a validation procedure by sampling source data points that are similar in distribution to the target data. We test the NET and the validation procedure using popular image datasets and compare the classification results across competitive procedures for unsupervised domain adaptation.
| Original language | English (US) |
|---|---|
| Title of host publication | WS-17-01 |
| Subtitle of host publication | Artificial Intelligence and Operations Research for Social Good; WS-17-02: Artificial Intelligence, Ethics, and Society; WS-17-03: Artificial Intelligence for Connected and Automated Vehicles; WS-17-04: Artificial Intelligence for Cyber Security; WS-17-05: Artificial Intelligence for Smart Grids and Buildings; WS-17-06: Computer Poker and Imperfect Information Games; WS-17-07: Crowdsourcing, Deep Learning and Artificial Intelligence Agents; WS-17-08: Distributed Machine Learning; WS-17-09: Joint Workshop on Health Intelligence; WS-17-10: Human-Aware Artificial Intelligence; WS-17-11: Human-Machine Collaborative Learning; WS-17-12: Knowledge-Based Techniques for Problem Solving and Reasoning; WS-17-13: Plan, Activity, and Intent Recognition; WS-17-14: Symbolic Inference and Optimization; WS-17-15: What's Next for AI in Games? |
| Publisher | AI Access Foundation |
| Pages | 774-780 |
| Number of pages | 7 |
| Volume | WS-17-01 - WS-17-15 |
| ISBN (Electronic) | 9781577357865 |
| State | Published - Jan 1 2017 |
| Event | 31st AAAI Conference on Artificial Intelligence, AAAI 2017 - San Francisco, United States Duration: Feb 4 2017 → Feb 10 2017 |
Other
| Other | 31st AAAI Conference on Artificial Intelligence, AAAI 2017 |
|---|---|
| Country | United States |
| City | San Francisco |
| Period | 2/4/17 → 2/10/17 |
Fingerprint
ASJC Scopus subject areas
- Engineering(all)
Cite this
Model selection with nonlinear embedding for unsupervised domain adaptation. / Demakethepalli Venkateswara, Hemanth; Chakraborty, Shayok; McDaniel, Troy; Panchanathan, Sethuraman.
WS-17-01: Artificial Intelligence and Operations Research for Social Good; WS-17-02: Artificial Intelligence, Ethics, and Society; WS-17-03: Artificial Intelligence for Connected and Automated Vehicles; WS-17-04: Artificial Intelligence for Cyber Security; WS-17-05: Artificial Intelligence for Smart Grids and Buildings; WS-17-06: Computer Poker and Imperfect Information Games; WS-17-07: Crowdsourcing, Deep Learning and Artificial Intelligence Agents; WS-17-08: Distributed Machine Learning; WS-17-09: Joint Workshop on Health Intelligence; WS-17-10: Human-Aware Artificial Intelligence; WS-17-11: Human-Machine Collaborative Learning; WS-17-12: Knowledge-Based Techniques for Problem Solving and Reasoning; WS-17-13: Plan, Activity, and Intent Recognition; WS-17-14: Symbolic Inference and Optimization; WS-17-15: What's Next for AI in Games?. Vol. WS-17-01 - WS-17-15 AI Access Foundation, 2017. p. 774-780.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Model selection with nonlinear embedding for unsupervised domain adaptation
AU - Demakethepalli Venkateswara, Hemanth
AU - Chakraborty, Shayok
AU - McDaniel, Troy
AU - Panchanathan, Sethuraman
PY - 2017/1/1
Y1 - 2017/1/1
N2 - Domain adaptation deals with adapting classifiers trained on data from a source distribution, to work effectively on data from a target distribution. In this paper, we introduce the Nonlinear Embedding Transform (NET) for unsupervised domain adaptation. The NET reduces cross-domain disparity through nonlinear domain alignment. It also embeds the domain- aligned data such that similar data points are clustered together. This results in enhanced classification. To determine the parameters in the NET model (and in other unsupervised domain adaptation models), we introduce a validation procedure by sampling source data points that are similar in distribution to the target data. We test the NET and the validation procedure using popular image datasets and compare the classification results across competitive procedures for unsupervised domain adaptation.
AB - Domain adaptation deals with adapting classifiers trained on data from a source distribution, to work effectively on data from a target distribution. In this paper, we introduce the Nonlinear Embedding Transform (NET) for unsupervised domain adaptation. The NET reduces cross-domain disparity through nonlinear domain alignment. It also embeds the domain- aligned data such that similar data points are clustered together. This results in enhanced classification. To determine the parameters in the NET model (and in other unsupervised domain adaptation models), we introduce a validation procedure by sampling source data points that are similar in distribution to the target data. We test the NET and the validation procedure using popular image datasets and compare the classification results across competitive procedures for unsupervised domain adaptation.
UR - http://www.scopus.com/inward/record.url?scp=85046095379&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85046095379&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85046095379
VL - WS-17-01 - WS-17-15
SP - 774
EP - 780
BT - WS-17-01
PB - AI Access Foundation
ER -