[TMP-094] Online Deep-AUTOML

Contact person: Joao Gama (joao.gama@inesctec.pt)

Internal Partners:

1. INESC TEC, Joao Gama, Bruno Veloso, and S´onia Teixeira
2. Consiglio Nazionale delle Ricerche (CNR), Giuseppe Manco and Luciano Caroprese
3. University of Leiden, Holger Hoos and Matthias K¨onig

External Partners:

1. Portucalense University, Bruno Veloso
2. University of British Columbia, Holger H. Hoos

Online AutoML in environments where the working conditions change over time.The main goal consists of studying online optimization methods for hyper-parameter tuning. In dynamic environments, the “optimal” hyper-parameters might change over time.Online AutoML consists of an exploration phase followed by an exploitation phase.The exploration phase is looking to find the set of hyper-parameters for the current working condition. The exploitation phase continuously monitors the learning process to detect degradation in the performance of the system which triggers a new exploitation phase.We consider complex problems described by pipelines where each step in the pipeline has its own hyper-parameters. We consider problems with many hyper-parameters where some of them might be irrelevant. Among the relevant parameters, the complexity of the model architecture (with particular reference to deep networks) is of particular relevance and is the objective of our study.

Results Summary

A Bayesian generative model is presented for recommending interesting items and trustworthy users to the targeted users in social rating networks with asymmetric and directed trust relationships. The proposed model is the first unified approach to the combination of the two recommendation tasks. Within the devised model, each user is associated with two latent-factor vectors, i.e., her susceptibility and expertise. Items are also associated with corresponding latent-factor vector representations. The probabilistic factorization of the rating data and trust relationships is exploited to infer user susceptibility and expertise. Statistical social-network modeling is instead used to constrain the trust relationships from a user to another to be governed by their respective susceptibility and expertise. The inherently ambiguous meaning of unobserved trust relationships between users is suitably disambiguated. An intensive comparative experimentation on real-world social rating networks with trust relationships demonstrates the superior predictive performance of the presented model in terms of RMSE and AUC.

Tangible Outcomes

1. hyper-Parameter Optimization for Latent Spaces in Dynamic Recommender Systems – Bruno Veloso, Luciano Caroprese, Matthias Konig, Sonia Teixeira, Giuseppe Manco, Holger H. Hoos, and Joao Gama in Machine Learning and Knowledge Discovery in Databases. Research Track – European Conference, ECML PKDD 2021, Bilbao, Spain, September 13-17, 2021
   https://gmanco.github.io/publication/veloso-2021/ 
2. Generator for preference data – Bruno Veloso, Luciano Caroprese, Matthias Konig, Sonia Teixeira, Giuseppe Manco, Holger H. Hoos, and Joao Gama
   https://github.com/BrunoMVeloso/ECMLPKDD2021 
3. Self Hyper-parameter tunning – Bruno Veloso, Luciano Caroprese, Matthias Konig, Sonia Teixeira, Giuseppe Manco, Holger H. Hoos, and Joao Gama
   https://github.com/BrunoMVeloso/ECMLPKDD2021