Yaping Zhang

Cong Ma, Yaping Zhang, Mei Tu et al.

Text image machine translation (TIMT) aims to translate texts embedded in images from one source language to another target language. Existing methods, both two-stage cascade and one-stage end-to-end architectures, suffer from different issues. The cascade models can benefit from the large-scale optical character recognition (OCR) and MT datasets but the two-stage architecture is redundant. The end-to-end models are efficient but suffer from training data deficiency. To this end, in our paper, we propose an end-to-end TIMT model fully making use of the knowledge from existing OCR and MT datasets to pursue both an effective and efficient framework. More specifically, we build a novel modal adapter effectively bridging the OCR encoder and MT decoder. End-to-end TIMT loss and cross-modal contrastive loss are utilized jointly to align the feature distribution of the OCR and MT tasks. Extensive experiments show that the proposed method outperforms the existing two-stage cascade models and one-stage end-to-end models with a lighter and faster architecture. Furthermore, the ablation studies verify the generalization of our method, where the proposed modal adapter is effective to bridge various OCR and MT models.

Kaisheng Liao, Yaodong Zhao, Jie Gu et al.

A novel and efficient end-to-end learning model for automatic modulation classification is proposed for wireless spectrum monitoring applications, which automatically learns from the time domain in-phase and quadrature data without requiring the design of hand-crafted expert features. With the intuition of convolutional layers with pooling serving as the role of front-end feature distillation and dimensionality reduction, sequential convolutional recurrent neural networks are developed to take complementary advantage of parallel computing capability of convolutional neural networks and temporal sensitivity of recurrent neural networks. Experimental results demonstrate that the proposed architecture delivers overall superior performance in signal to noise ratio range above -10~dB, and achieves significantly improved classification accuracy from 80\% to 92.1\% at high signal to noise ratio range, while drastically reduces the average training and prediction time by approximately 74% and 67%, respectively. Response patterns learned by the proposed architecture are visualized to better understand the physics of the model. Furthermore, a comparative study is performed to investigate the impacts of various sequential convolutional recurrent neural network structure settings on classification performance. A representative sequential convolutional recurrent neural network architecture with the two-layer convolutional neural network and subsequent two-layer long short-term memory neural network is developed to suggest the option for fast automatic modulation classification.

Bin Liu, Shuai Nie, Yaping Zhang et al.

LSTM-based speaker verification usually uses a fixed-length local segment randomly truncated from an utterance to learn the utterance-level speaker embedding, while using the average embedding of all segments of a test utterance to verify the speaker, which results in a critical mismatch between testing and training. This mismatch degrades the performance of speaker verification, especially when the durations of training and testing utterances are very different. To alleviate this issue, we propose the deep segment attentive embedding method to learn the unified speaker embeddings for utterances of variable duration. Each utterance is segmented by a sliding window and LSTM is used to extract the embedding of each segment. Instead of only using one local segment, we use the whole utterance to learn the utterance-level embedding by applying an attentive pooling to the embeddings of all segments. Moreover, the similarity loss of segment-level embeddings is introduced to guide the segment attention to focus on the segments with more speaker discriminations, and jointly optimized with the similarity loss of utterance-level embeddings. Systematic experiments on Tongdun and VoxCeleb show that the proposed method significantly improves robustness of duration variant and achieves the relative Equal Error Rate reduction of 50% and 11.54% , respectively.