Technology
Efficient Algorithms for Non-convex Isotonic Regression through Submodular Optimization
Neural Information Processing Systems http://nips.cc/
Exact Rate-Distortion in Autoencoders via Echo Noise
Rob Brekelmans, Daniel Moyer, Aram Galstyan, Greg Ver Steeg
Made: Maskedautoencoder fordistributionestimation. InInternational Conferenceon Machine Learning, pages 881-889, 2015.
Regret Bounds for Learning State Representations in Reinforcement Learning
Ronald Ortner, Matteo Pirotta, Alessandro Lazaric, Ronan Fruit, Odalric-Ambrym Maillard
Neural Information Processing Systems http://nips.cc/
9b8b50fb590c590ffbf1295ce92258dc-AuthorFeedback.pdf
For example, when solving RL problems such as Atari7 games, we may test different representation methods. Fortheaveragereward30 setting, it is still an open question whether S-bounds areachievable. Ourapproach canbeadapted totheepisodic31 case when the regret bounds would benefit from the improved bounds available in this setting. The A-dependence is optimal as for UCRL2, while the optimal dependence onS is still an open question (also46 for the MDP case). The optimal dependence on|ฮฆ| in our setting is also open.
Quadrature-based features for kernel approximation
Marina Munkhoeva, Yermek Kapushev, Evgeny Burnaev, Ivan Oseledets
Neural Information Processing Systems http://nips.cc/
fe248e22b241ae5a9adf11493c8c12bc-Supplemental-Conference.pdf
In practice, however, the runtime is much smaller becauseofpruning. With this change in place, the solver can search for incomplete trees. KamPost differs from CART by using a different splitting criterion and by its post-relabelling of theleafnodes. The results also confirm thefindings from Figure 1that thevariance inthediscrimination value is often high, specifically for the small datasets. This means that for those instances it is difficult to generalize and overfitting interms ofdiscrimination isstill happening.
