As the emergence and the thriving development of social networks, a huge number of short texts are accumulated and need to be processed. Inferring latent topics of collected short texts is an essential task for understanding its hidden structure and predicting new contents. A biterm topic model (BTM) was recently proposed for short texts to overcome the sparseness of document-level word co-occurrences by directly modeling the generation process of word pairs. Stochastic inference algorithms based on collapsed Gibbs sampling (CGS) and collapsed variational inference have been proposed for BTM. However, they either require large computational complexity, or rely on very crude estimation that does not preserve sufficient statistics. In this work, we develop a stochastic divergence minimization (SDM) inference algorithm for BTM to achieve better predictive likelihood in a scalable way. Experiments show that SDM-BTM trained by 30% data outperforms the best existing algorithm trained by full data.

This paper proposes a method of improving reception of digital satellite broadcasting in a moving vehicle. According to some studies, the antennas used for mobile reception will be smaller in the next generation and reception will be more difficult because of a fading multipath channel with delays in a low carrier-to-noise ratio. Commonly used approaches to reduce the inter symbol interference caused by a fading multipath channel with delays are pilot sequences and diversity reception. Digital satellite broadcasting, however, does not transmit pilot sequences for channel estimation and it is not possible to install multiple antennas in a vehicle. This paper does not propose any change to the broadcasting standards but discusses how to process currently available digital satellite signals to obtain better results. Our method does not rely on the pilot sequences or diversity reception, but consists of channel estimation and stochastic inference methods. For each task, two methods are proposed. The maximum likelihood estimation and higher order statistics matching methods are proposed for the estimation, and the marginal with the joint probability inference methods are proposed for the stochastic inference. The improvements were confirmed through experiments with numerical simulations and real data. The computational costs are also discussed for future implementation.