Hybrid algorithm for predicting the temperature-variation-induced wavelength drift of DBR semiconductor lasers.

In this paper, a hybrid algorithm to predict the wavelength drift induced by ambient temperature variation in distributed Bragg reflector semiconductor lasers is proposed. This algorithm combines the global search capability of a genetic algorithm (GA) and the supermapping ability of an extreme learning machine (ELM), which not only avoids the randomness of ELM but also improves its generalization performance. In addition, a tenfold cross-validation method is employed to determine the optimal activation function and the number of hidden layer nodes for ELM to construct the most suitable model. After applying multiple sets of test data, the results demonstrate that GA-ELM can quickly and accurately predict the wavelength drift, with an average rms error of 4.09×10-4nm and average mean absolute percentage error of 0.21 %. This model is expected to combine the temperature and current tuning models for a wavelength in follow-up research to achieve rapid tuning and high stability of a wavelength without additional devices.

H-VECTORS: Improving the robustness in utterance-level speaker embeddings using a hierarchical attention model.

In this paper, a hierarchical attention network is proposed to generate robust utterance-level embeddings (H-vectors) for speaker identification and verification. Since different parts of an utterance may have different contributions to speaker identities, the use of hierarchical structure aims to learn speaker related information locally and globally. In the proposed approach, frame-level encoder and attention are applied on segments of an input utterance and generate individual segment vectors. Then, segment level attention is applied on the segment vectors to construct an utterance representation. To evaluate the quality of the learned utterance-level speaker embeddings on speaker identification and verification, the proposed approach is tested on several benchmark datasets, such as the NIST SRE2008 Part1, the Switchboard Cellular (Part1), the CallHome American English Speech ,the Voxceleb1 and Voxceleb2 datasets. In comparison with some strong baselines, the obtained results show that the use of H-vectors can achieve better identification and verification performances in various acoustic conditions.