Unconstrained real world environments are often full of clutter. Therefore, robustness to occlusion is vital for any artificial recognition system. Recently, the Deep Boltzmann Machine (DBM) has been shown to be good at generative modeling and recognition of visual objects. In this work we develop an extension to the DBM framework to make the system more accurate when recognizing handwritten di...

Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi Author Kyunghyun Cho Name of the doctoral dissertation Foundations and Advances in Deep Learning Publisher Unit Department of Information and Computer Science Series Aalto University publication series DOCTORAL DISSERTATIONS 21/2014 Field of research Machine Learning Manuscript submitted 2 September 2013 Date of the defence 21 March ...

Parallel and distributed algorithms have become a necessity in modern machine learning tasks. In this work, we focus on parallel asynchronous gradient descent [1, 2, 3] and propose a zealous variant that minimizes the idle time of processors to achieve a substantial speedup. We then experimentally study this algorithm in the context of training a restricted Boltzmann machine on a large collabor...

Abstract Neural-network quantum state (NQS) has attracted significant interests as a powerful wave-function ansatz to model phenomena. In particular, variant of NQS based on the restricted Boltzmann machine (RBM) been adapted ground spin lattices and electronic structures small molecules in devices. Despite these progresses, challenges remain with RBM-NQS-based simulations. this work, we presen...

The papers in this series are intended for internal use and are distributed by the author. Copies may be ordered from the library of Department of Computer Science. Abstract The primary motivation of this work has been the study of phase transitions in the simulated annealing schedules of Boltzmann machines. At rst, we introduce and focus on the harmonium Boltzmann machine, which can be used to...

A Boltzmann game is an n-player repeated game, in which Boltzmann machines are employed by players to choose their optimal strategy for each round of the game. Players only have knowledge about the machine they have selected and their own strategy set. Information about other the players and the game’s pay-off function are concealed from all players. Players therefore select their strategies in...

Neal, R.M., Connectionist learning of belief networks, Artificial Intelligence 56 (1992) 71-113. Connectionist learning procedures are presented for "sigmoid" and "noisy-OR" varieties of probabilistic belief networks. These networks have previously been seen primarily as a means of representing knowledge derived from experts. Here it is shown that the "Gibbs sampling" simulation procedure for s...

The wind speed forecasting in Hong Kong is more difficult than in other places in the same latitude for two reasons: the great affect from the urbanization of Hong Kong in the long term, and the very high wind speeds brought by the tropical cyclones. Therefore, prediction model with higher learning ability is in need for the wind speed forecast in Hong Kong. In this paper, we try to employ the ...