Tutorials
Monday, September 7 (half day tutorials)
Learning from Multi-label Data (morning)
Grigorios Tsoumakas (Aristotle University of Thessaloniki) , Min Ling Zhang (Hohai University) and Zhi-Hua Zhou (Nanjing University)
Keywords: problem transformation and algorithm adaptation, exploiting label structure, learning with large number of labels, multi-instance multi-label learning, the Mulan open-source software, real-world applications.
Language and Document Analysis: Motivating Latent Variable Models (morning)
Wray Buntine (NICTA, Helsinki Institute of IT)
Keywords: natural language processing and linguistics, document analysis, information access, latent variable models, topic models.
Methods for Large Network Analysis (afternoon)
Vladimir Batagelj (University of Ljubljana)
Keywords: large networks, connectivity, network condensation, properties of vertices/lines, islands, two-mode network, pattern search (motifs), triadic spectrum, clustering in large networks., visualization of large networks.
Friday, September 11 (half day tutorials)
Evaluation in Machine Learning (morning)
Pádraig Cunningham (University College Dublin)
Keywords: loss functions, accuracy, precision recall, F-Score, ROC and AUC analysis, aggregation using win-tie-loss analysis, cross validation, leave-one-out validation, Statistical tests for ML evaluation, comparing across domains, evaluation blunders.
Transfer Learning for Reinforcement Learning Domains (afternoon)
Alessandro Lazaric (INRIA Lille) and Matthew Taylor (University of Southern California)
Keywords: transfer learning in AI and ML, theoretical foundation of transfer in RL, task relatedness, transferred knowledge, fixed state-action variables, multitask learning, inter-task mapping, fully autonomous transfer.
Graphical Models (morning)
Tiberio Caetano (NICTA)
Keywords: belief propagation and junction trees, maximum-likelihood and maximum-a-posteriori estimation, conditional random fields, Markov random fields, Bayesian Networks, nonparametric modeling.