Learn More
We have produced a draft sequence of the rice genome for the most widely cultivated subspecies in China, Oryza sativa L. ssp. indica, by whole-genome shotgun sequencing. The genome was 466 megabases in size, with an estimated 46,022 to 55,615 genes. Functional coverage in the assembled sequences was 92.0%. About 42.2% of the genome was in exact(More)
We apply the recently proposed Context-Dependent Deep-Neural-Network HMMs, or CD-DNN-HMMs, to speech-to-text transcription. For single-pass speaker-independent recognition on the RT03S Fisher portion of phone-call transcription benchmark (Switchboard), the word-error rate is reduced from 27.4%, obtained by discriminatively trained Gaussian-mixture HMMs, to(More)
Emotional expression and understanding are normal instincts of human beings, but auto-matical emotion recognition from speech without referring any language or linguistic information remains an unclosed problem. The limited size of existing emotional data samples , and the relative higher dimensionality have outstripped many dimensionality reduction and(More)
Age Specific Human-Computer Interaction (ASHCI) has vast potential applications in daily life. However, automatic age estimation technique is still underdeveloped. One of the main reasons is that the aging effects on human faces present several unique characteristics which make age estimation a challenging task that requires non-standard classification(More)
Eighty-four post-1990 empirical studies of international tourism demand modeling and forecasting using econometric approaches are reviewed. New developments are identified and it is shown that applications of advanced econometric methods improve the understanding of international tourism demand. An examination of the 22 studies which compare forecasting(More)
as much as one third—from 27.4%, obtained by discrimina-tively trained Gaussian-mixture HMMs with HLDA features, to 18.5%—using 300+ hours of training data (Switchboard), 9000+ tied triphone states, and up to 9 hidden network layers. In this paper, we evaluate the effectiveness of feature transforms developed for GMM-HMMs—HLDA, VTLN, and fMLLR—applied to(More)
We propose a novel regularized adaptation technique for context dependent deep neural network hidden Markov models (CD-DNN-HMMs). The CD-DNN-HMM has a large output layer and many large hidden layers, each with thousands of neurons. The huge number of parameters in the CD-DNN-HMM makes adaptation a challenging task, esp. when the adaptation set is small. The(More)
We show empirically that in SGD training of deep neural networks , one can, at no or nearly no loss of accuracy, quantize the gradients aggressively—to but one bit per value—if the quan-tization error is carried forward across minibatches (error feedback). This size reduction makes it feasible to parallelize SGD through data-parallelism with fast processors(More)
Mutations in XLF/Cernunnos (XLF) cause lymphocytopenia in humans, and various studies suggest an XLF role in classical nonhomologous end joining (C-NHEJ). We now find that XLF-deficient mouse embryonic fibroblasts are ionizing radiation (IR) sensitive and severely impaired for ability to support V(D)J recombination. Yet mature lymphocyte numbers in(More)