No abstract

No abstract

Recently deep neural networks (DNNs) have been used to learn speaker features. However, the quality of the learned features is not sufficiently good, so a complex back-end model, either neural or probabilistic, has to be used to address the residual uncertainty when applied to speaker verification, just as with raw features. This paper presents a convolutional timedelay deep neural network structure (CT-DNN) for speaker feature learning. Our experimental results on the Fisher database demonstrated that this CT-DNN can produce highquality speaker features: even with a single feature (0.3 seconds including the context), the EER can be as low as 7.68%. This effectively confirmed that the speaker trait is largely a deterministic short-time property rather than a long-time distributional pattern, and therefore can be extracted from just dozens of frames.

Emerging evidence support that temporal dynamics is pivotal for signaling molecules in orchestrating smart responses to diverse stimuli. p53 is such a signaling molecule that employs temporal dynamics for the selective activation of downstream target genes and ultimately for cell fate decision. Yet how this fine-tuned p53 machinery is quantitatively decoded remains largely unclear. Here we report a quantitative mechanism defining how p53 dynamics orchestrates with binding affinity to target genes for cell fate decision. Treating cells with a genotoxic drug doxorubicin at various doses and durations, we found that a mild and prolonged challenge triggered sequential p53 pulses and ultimately resulted in a terminal pulse enacting apoptosis in a comparable rate with that induced by an acute and high-dose treatment. To transactivate proapoptotic genes and thereafter executing apoptosis, p53 must exceed a certain threshold and accumulate for sufficient time at levels above it. Effective cumulative levels above the threshold, defined as E∫p53, but not the total accumulation levels of p53, precisely discriminate survival and apoptotic cells. p53 accumulation below this threshold, even with prolonging time to reach a total level comparable to that from the accumulation over the threshold, could not transactivate proapoptotic genes to which the binding affinity of p53 is lower than that of proarrest genes, and this property is independent of dynamic features. Our findings indicate that the dynamic feature per se does not directly control cell fate, but rather it orchestrates with the binding affinity to target genes to confer an appropriate time window for cell fate choice. Our study provides a quantitative mechanism unifying p53 dynamics and binding affinity to target genes, providing novel insights to understand how p53 can respond quantitatively to chemotherapeutic drugs, and guiding the design of metronomic regimens for chemotherapeutic drugs.

Li1.3Al0.3Ti1.7(PO4)3 powder was synthesized with CO(NH2)2 as a molten flux and Li3PO4 SEI layer was used to improve cycle stability of the battery.