The Experts below are selected from a list of 5160 Experts worldwide ranked by ideXlab platform

Jenqder Chen - One of the best experts on this subject based on the ideXlab platform.

Changhua Lien - One of the best experts on this subject based on the ideXlab platform.

Longyeu Chung - One of the best experts on this subject based on the ideXlab platform.

Andre F T Martins - One of the best experts on this subject based on the ideXlab platform.

  • unbabel s submission to the wmt2019 ape shared task bert based encoder decoder for automatic post editing
    arXiv: Computation and Language, 2019
    Co-Authors: Antonio V Lopes, Amin M Farajian, Goncalo M Correia, Jonay Trenous, Andre F T Martins
    Abstract:

    This paper describes Unbabel's submission to the WMT2019 APE Shared Task for the English-German language pair. Following the recent rise of large, powerful, pre-trained models, we adapt the BERT pretrained model to perform Automatic Post-Editing in an encoder-decoder framework. Analogously to dual-encoder architectures we develop a BERT-based encoder-decoder (BED) model in which a single pretrained BERT encoder receives both the source src and machine translation tgt strings. Furthermore, we explore a Conservativeness factor to constrain the APE system to perform fewer edits. As the official results show, when trained on a weighted combination of in-domain and artificial training data, our BED system with the Conservativeness penalty improves significantly the translations of a strong Neural Machine Translation system by $-0.78$ and $+1.23$ in terms of TER and BLEU, respectively. Finally, our submission achieves a new state-of-the-art, ex-aequo, in English-German APE of NMT.

  • unbabel s submission to the wmt2019 ape shared task bert based encoder decoder for automatic post editing
    Proceedings of the Fourth Conference on Machine Translation (Volume 3: Shared Task Papers Day 2), 2019
    Co-Authors: Antonio V Lopes, Amin M Farajian, Goncalo M Correia, Jonay Trenous, Andre F T Martins
    Abstract:

    This paper describes Unbabel’s submission to the WMT2019 APE Shared Task for the English-German language pair. Following the recent rise of large, powerful, pre-trained models, we adapt the BERT pretrained model to perform Automatic Post-Editing in an encoder-decoder framework. Analogously to dual-encoder architectures we develop a BERT-based encoder-decoder (BED) model in which a single pretrained BERT encoder receives both the source src and machine translation mt strings. Furthermore, we explore a Conservativeness factor to constrain the APE system to perform fewer edits. As the official results show, when trained on a weighted combination of in-domain and artificial training data, our BED system with the Conservativeness penalty improves significantly the translations of a strong NMT system by -0.78 and +1.23 in terms of TER and BLEU, respectively. Finally, our submission achieves a new state-of-the-art, ex-aequo, in English-German APE of NMT.

Boming Zhang - One of the best experts on this subject based on the ideXlab platform.

  • robust restoration method for active distribution networks
    IEEE Transactions on Power Systems, 2016
    Co-Authors: Xin Chen, Boming Zhang
    Abstract:

    Distributed generations (DGs) introduce significant uncertainties to restoration of active distribution networks, in addition to roughly estimated load demands. An adjustable robust restoration optimization model with a two-stage objective is proposed in this paper, involving the uncertain DG outputs and load demands. The first stage generates optimal strategies for recovery of outage power and the second stage seeks the worst-case fluctuation scenarios. The model is formulated as a mixed-integer linear programming problem and solved using the column-and-constraint generation method. The feasibility and reliability of the strategies obtained via this robust optimization model can be guaranteed for all cases in the predefined uncertainty sets with good performance. A technique known as the uncertainty budget is used to adjust the Conservativeness of this model, providing a tradeoff between Conservativeness and robustness. Numerical tests are carried out on the modified PG&E 69-bus system and a modified 246-bus system to compare the robust optimization model against a deterministic restoration model, which verifies the superiority of this proposed model.

  • robust look ahead power dispatch with adjustable Conservativeness accommodating significant wind power integration
    IEEE Transactions on Sustainable Energy, 2015
    Co-Authors: Boming Zhang, Bin Wang
    Abstract:

    Robust look-ahead power dispatch is an effective and secure approach to account for uncertainties in wind power. Based on the robust look-ahead power dispatch model proposed in our previous work, in this paper, we focus on coping with adjustable uncertainty sets to reduce the Conservativeness of robust dispatch (RD). Robust look-ahead power dispatch with adjustable uncertainty sets is formulated using robust optimization (RO) and is transformed into biconvex programming. Several strategies exploiting different levels of uncertainty information are developed to adjust the Conservativeness by varying the desired confidence level. Monte Carlo simulations are carried out to compare the performance of the proposed approaches with another popular strategy. Test results show that the proposed method is effective in reducing the Conservativeness of RD and ensuring system security with controllable risk. Experiments on large-scale benchmark systems show the scalability of the proposed method.