Probabilistic user modelling methods for improving human-in-the-loop machine learning for prediction
Loading...
Journal Title
Journal ISSN
Volume Title
School of Science |
Doctoral thesis (article-based)
| Defence date: 2021-05-28
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Author
Date
2021
Major/Subject
Mcode
Degree programme
Language
en
Pages
54 + app. 72
Series
Aalto University publication series DOCTORAL DISSERTATIONS, 58/2021
Abstract
In many machine learning applications and in particular those with only few training data, human involvement in the form of data provider or expert of the task is crucial. However, human interaction with a machine learning model is constrained by (i) the interaction channels, i.e., how human knowledge can be applied in the model, and (ii) the interaction budget, i.e., how much the user is willing to interact with the model. This thesis presents new methods to improve these constraints in human-in-the-loop machine learning. The core idea of the thesis is to jointly model the available data with a model of the human user, i.e., the user model, in a unified probabilistic model and then perform sequential probabilistic inference on the joint model to design improved interaction. The thesis contributes on two types of prediction tasks. The first task is expert knowledge elicitation for high-dimensional prediction. Experts in a field usually have information beyond training data which can help to improve the prediction performance. User models, as priors and likelihood functions, are proposed to directly connect expert knowledge about the relevance of parameters to a model responsible for prediction. The user model can account for complex user behaviour such as users updating their knowledge during the interaction. Furthermore, sequential experimental design on the joint model is employed to query the most informative expert knowledge earlier to minimize the amount of interaction. The second task is personalized recommendation where the goal is to predict the most relevant item for a user with as few interactions as possible. The interactions are based on user relevance feedback on the recommendations. The thesis proposes user models that are able to receive and integrate feedback on multiple domains and sources by providing a joint probabilistic model connecting all feedback types. Sequential inference on the joint model, using Thompson sampling, was employed to find the targeted recommendation with minimum interaction. Simulated experiments and user studies in both tasks demonstrate improved prediction performance only after few interactions with the users. The research highlights the benefits of joint probabilistic modelling of the user and prediction model in interactive tasks.Description
Defence is held on 28.5.2021 12:00 – 16:00
Zoom link https://aalto.zoom.us/j/3182801227
Supervising professor
Kaski, Samuel, Prof., Aalto University, Department of Computer Science, FinlandThesis advisor
Peltola, Tomi, Dr., Silo.AI, FinlandKeywords
interactive machine learning, Bayesian inference, probabilistic user modelling
Other note
Parts
-
[Publication 1]: Pedram Daee, Joel Pyykkö, Dorota Głowacka, and Samuel Kaski. Interactive Intent Modeling from Multiple Feedback Domains. In Proceedings of the 21st International Conference on Intelligent User Interfaces, Sonoma, California, USA, pages 71–75, March 2016.
DOI: 10.1145/2856767.2856803 View at publisher
-
[Publication 2]: Pedram Daee, Tomi Peltola, Marta Soare, and Samuel Kaski. Knowledge elicitation via sequential probabilistic inference for high-dimensional prediction. Machine Learning, volume 106, issue 9-10, pages 1599–1620, 2017.
Full text in Acris/Aaltodoc: http://urn.fi/URN:NBN:fi:aalto-201708036366DOI: 10.1007/s10994-017-5651-7 View at publisher
-
[Publication 3]: Iiris Sundin, Tomi Peltola, Luana Micallef, Homayun Afrabandpey, Marta Soare, Muntasir Mamun Majumder, Pedram Daee, Chen He, Baris Serim, Aki Havulinna, Caroline Heckman, Giulio Jacucci, Pekka Marttinen, and Samuel Kaski. Improving genomics-based predictions for precision medicine through active elicitation of expert knowledge. Bioinformatics, volume 34, issue 13, pages i395–i403, 2018.
Full text in Acris/Aaltodoc: http://urn.fi/URN:NBN:fi:aalto-201808014342DOI: 10.1093/bioinformatics/bty257 View at publisher
-
[Publication 4]: Pedram Daee, Tomi Peltola, Aki Vehtari, and Samuel Kaski. User Modelling for Avoiding Overfitting in Interactive Knowledge Elicitation for Prediction. In Proceedings of the 23rd International Conference on Intelligent User Interfaces, Tokyo, Japan, pages 305–310, March 2018.
Full text in Acris/Aaltodoc: http://urn.fi/URN:NBN:fi:aalto-201812106382DOI: 10.1145/3172944.3172989 View at publisher
-
[Publication 5]: Giulio Jacucci, Oswald Barral, Pedram Daee, Markus Wenzel, Baris Serim, Tuukka Ruotsalo, Patrik Pluchino, Jonathan Freeman, Luciano Gamberini, Samuel Kaski, Benjamin Blankertz. Integrating Neurophysiological Relevance Feedback in Intent Modeling for Information Retrieval. Journal of the Association for Information Science and Technology, volume 70, issue 9, pages, 917–930, 2019.
Full text in Acris/Aaltodoc: http://urn.fi/URN:NBN:fi:aalto-201904022387DOI: 10.1002/asi.24161 View at publisher