In this paper we present a spoken query detection method based on posteriorgrams generated from Deep Boltzmann Machines (DBMs). The proposed method can be deployed in both semi-supervised and unsupervised training scenarios. The DBM-based posteriorgrams were evaluated on a series of keyword spotting tasks using the TIMIT speech corpus. In unsupervised training conditions, the DBM-approach improved upon our previous best unsupervised keyword detection performance using Gaussian mixture model-based posteriorgrams by over 10%. When limited amounts of labeled data were incorporated into training, the DBM-approach required less than one third of the annotated data in order to achieve a comparable performance of a system that used all of the annotated data for training.

The increasing consumption of shark products, along with the shark’s fishing vulnerabilities, has led to the decrease in certain shark populations. In this study we used a DNA barcoding method to identify the species of shark landings at fishing ports, shark fin products in retail stores, and shark fins detained by Taiwan customs. In total we identified 23, 24, and 14 species from 231 fishing landings, 316 fin products, and 113 detained shark fins, respectively. All the three sample sources were dominated by Prionace glauca, which accounted for more than 30% of the collected samples. Over 60% of the species identified in the fin products also appeared in the port landings, suggesting the domestic-dominance of shark fin products in Taiwan. However, international trade also contributes a certain proportion of the fin product markets, as four species identified from the shark fin products are not found in Taiwan’s waters, and some domestic-available species were also found in the customs-detained sample. In addition to the species identification, we also found geographical differentiation in the cox1 gene of the common thresher sharks (Alopias vulpinus), the pelagic thresher shark (A. pelagicus), the smooth hammerhead shark (Sphyrna zygaena), and the scalloped hammerhead shark (S. lewini). This result might allow fishing authorities to more effectively trace the origins as well as enforce the management and conservation of these sharks.

No abstract

This paper presents a stochastic estimation approach to adaptive interpolation of color filter array. It models an image as a 2-D locally stationary Gaussian process and achieves robustness against aliasing by employing an edge-sensitive weighting policy based on the stochastic characteristics of uniformly oriented edge indicators. Experimental results show that the algorithm can effectively eliminate the occurrence of perceptible artifact. Performance comparison in terms of peak signal-to-noise ratio and mean square error is provided to demonstrate the superiority of the proposed algorithm.Index Terms-Color filter array (CFA), color interpolation, demosaicing, edge detection, edge-sensitive weighting.

Recent improvements in speech recognition technology, along with increased computing power and bigger datasets, have considerably improved the state of the art in the field, making it possible for commercial apps such as Google Voice Search to serve users in their everyday mobile search needs. Deploying such systems in various countries has shown us the extent to which multilingualism is present in some cultures, and the need for better solutions to handle it in our speech recognition systems.In this paper, we describe a few early data sharing and model combination experiments we did to improve the recognition of English queries made to Mandarin Voice Search, in Taiwan. We obtained a 12% relative sentence accuracy improvement over a baseline system already including some support for English queries.

In this paper we propose discriminative training of hierarchical acoustic models for large vocabulary continuous speech recognition tasks. After presenting our hierarchical modeling framework, we describe how the models can be generated with either Minimum Classification Error or large-margin training. Experiments on a large vocabulary lecture transcription task show that the hierarchical model can yield more than 1.0% absolute word error rate reduction over non-hierarchical models for both kinds of discriminative training.

Electronic image stabilizer and video codec are two important components of a digital video camera. Both require motion information of the captured image sequence to perform their respective tasks. Since motion estimation is a computationally intensive operation, we propose three schemes for integrating the electronic image stabilizer with the video codec. The technical issues involved in the integration are discussed, and the simulation results are shown to illustrate the effectiveness of the integration schemes. I.

No abstract