Sequence-to-sequence (seq2seq) voice conversion (VC) models are attractive owing to their ability convert prosody. Nonetheless, without sufficient data, seq2seq VC can suffer from unstable training and mispronunciation problems in the converted speech, thus far practical. To tackle these shortcomings, we propose transfer knowledge other speech processing tasks where large-scale corpora easily a...

Short-term forecasting of passenger flow is critical for transit management and crowd regulation. Spatial dependencies, temporal inter-station correlations driven by other latent factors, exogenous factors bring challenges to the short-term forecasts urban rail networks. An innovative deep learning approach, Multi-Graph Convolutional-Recurrent Neural Network (MGC-RNN) proposed forecast in syste...

Virtual assistants are the cutting edge of end user interaction, thanks to endless set of capabilities across multiple services. The natural language techniques thus need to be evolved to match the level of power and sophistication that users expect from virtual assistants. In this report we investigate an existing deep learning model for semantic parsing, and we apply it to the problem of conv...

Since the accurate prediction of porosity is one critical factors for estimating oil and gas reservoirs, a novel method based on Imaged Sequence Samples (ISS) to (Seq2Seq) model fused by Transcendental Learning (TL) proposed using well-logging data. Firstly, investigate correlation between logging features porosity, original are normalized selected computing their with obtain point samples. Sec...

Developing an effective task offloading strategy has been a focus of research to improve the processing speed IoT devices in recent years. Some reinforcement learning-based policies can dependence heuristic algorithms on models through continuous interactive exploration edge environment; however, when environment changes, such learning cannot adapt and need spend time retraining. This paper pro...

A sequence-to-sequence attention reading comprehension model was implemented to fulfill Question Answering task defined in Stanford Question Answering Dataset (SQuAD). The basic structure was bidirectional LSTM (BiLSTM) encodings with attention mechanism as well as BiLSTM decoding. Several adjustments such as dropout, learning rate decay, and gradients clipping were used. Finally, the model ach...

In this paper, we propose a generative model which learns the relationship between language and human action in order to generate a human action sequence given a sentence describing human behavior. The proposed generative model is a generative adversarial network (GAN), which is based on the sequence to sequence (SEQ2SEQ) model. Using the proposed generative network, we can synthesize various a...