The 22nd International Conference on Algorithmic Learning Theory (ALT
2011, http://www-alg.ist.hokudai.ac.
be held at Aalto University in Espoo, Finland, October 5-7, 2011. The
conference is on the theoretical foundations of machine learning. The
conference will be co-located with the 14th International Conference
on Discovery Science (DS 2011, http://ds2011.org/).
Topics of Interest:
* Comparison of the strength of learning models and the design and
evaluation of novel algorithms for learning problems in
established learning-theoretic settings such as
o statistical learning theory,
o on-line learning,
o inductive inference,
o query models,
o unsupervised, semi-supervised and active learning.
* Analysis of the theoretical properties of existing algorithms:
o families of algorithms could include
+ boosting,
+ kernel-based methods, SVM,
+ Bayesian networks,
+ methods for reinforcement learning or learning in
repeated games,
+ graph- and/or manifold-based methods,
+ methods for latent-variable estimation and/or clustering,
+ MDL,
+ decision tree methods,
+ information-based methods,
o analyses could include generalization, convergence or
computational efficiency.
* Definition and analysis of new learning models. Models might
o identify and formalize classes of learning problems
inadequately addressed by existing theory or
o capture salient properties of important concrete applications.
2011, http://www-alg.ist.hokudai.ac.
be held at Aalto University in Espoo, Finland, October 5-7, 2011. The
conference is on the theoretical foundations of machine learning. The
conference will be co-located with the 14th International Conference
on Discovery Science (DS 2011, http://ds2011.org/).
Topics of Interest:
* Comparison of the strength of learning models and the design and
evaluation of novel algorithms for learning problems in
established learning-theoretic settings such as
o statistical learning theory,
o on-line learning,
o inductive inference,
o query models,
o unsupervised, semi-supervised and active learning.
* Analysis of the theoretical properties of existing algorithms:
o families of algorithms could include
+ boosting,
+ kernel-based methods, SVM,
+ Bayesian networks,
+ methods for reinforcement learning or learning in
repeated games,
+ graph- and/or manifold-based methods,
+ methods for latent-variable estimation and/or clustering,
+ MDL,
+ decision tree methods,
+ information-based methods,
o analyses could include generalization, convergence or
computational efficiency.
* Definition and analysis of new learning models. Models might
o identify and formalize classes of learning problems
inadequately addressed by existing theory or
o capture salient properties of important concrete applications.
