no code implementations • 23 Sep 2023 • Sam Davidson, Salvatore Romeo, Raphael Shu, James Gung, Arshit Gupta, Saab Mansour, Yi Zhang
One of the major impediments to the development of new task-oriented dialogue (TOD) systems is the need for human evaluation at multiple stages and iterations of the development process.
no code implementations • 1 Aug 2023 • Qingyang Wu, James Gung, Raphael Shu, Yi Zhang
Dialogue act annotations are important to improve response generation quality in task-oriented dialogue systems.
1 code implementation • 24 May 2023 • Mujeen Sung, James Gung, Elman Mansimov, Nikolaos Pappas, Raphael Shu, Salvatore Romeo, Yi Zhang, Vittorio Castelli
Intent classification (IC) plays an important role in task-oriented dialogue systems.
2 code implementations • 25 Apr 2023 • James Gung, Raphael Shu, Emily Moeng, Wesley Rose, Salvatore Romeo, Yassine Benajiba, Arshit Gupta, Saab Mansour, Yi Zhang
With increasing demand for and adoption of virtual assistants, recent work has investigated ways to accelerate bot schema design through the automatic induction of intents or the induction of slots and dialogue states.
no code implementations • 16 Feb 2023 • Shamik Roy, Raphael Shu, Nikolaos Pappas, Elman Mansimov, Yi Zhang, Saab Mansour, Dan Roth
Conventional text style transfer approaches focus on sentence-level style transfer without considering contextual information, and the style is described with attributes (e. g., formality).
no code implementations • 20 Dec 2022 • Raphael Shu, Elman Mansimov, Tamer Alkhouli, Nikolaos Pappas, Salvatore Romeo, Arshit Gupta, Saab Mansour, Yi Zhang, Dan Roth
The conversational model interacts with the environment by generating and executing programs triggering a set of pre-defined APIs.
1 code implementation • 27 May 2022 • Woojung Kim, Keondo Park, Kihyuk Sohn, Raphael Shu, Hyung-Sin Kim
Compared to a FSSL approach based on weight sharing, the prototype-based inter-client knowledge sharing significantly reduces both communication and computation costs, enabling more frequent knowledge sharing between more clients for better accuracy.
1 code implementation • 15 Apr 2021 • Raphael Shu, Kang Min Yoo, Jung-Woo Ha
Results show that the reward optimization with BLEURT is able to increase the metric scores by a large margin, in contrast to limited gain when training with smoothed BLEU.
no code implementations • INLG (ACL) 2021 • Hong Chen, Raphael Shu, Hiroya Takamura, Hideki Nakayama
In this paper, we focus on planning a sequence of events assisted by event graphs, and use the events to guide the generator.
1 code implementation • EMNLP 2020 • Jason Lee, Raphael Shu, Kyunghyun Cho
Given a continuous latent variable model for machine translation (Shu et al., 2020), we train an inference network to approximate the gradient of the marginal log probability of the target sentence, using only the latent variable as input.
1 code implementation • 20 Aug 2019 • Raphael Shu, Jason Lee, Hideki Nakayama, Kyunghyun Cho
By decoding multiple initial latent variables in parallel and rescore using a teacher model, the proposed model further brings the gap down to 1. 0 BLEU point on WMT'14 En-De task with 6. 8x speedup.
no code implementations • ACL 2019 • Raphael Shu, Hideki Nakayama, Kyunghyun Cho
In this work, we attempt to obtain diverse translations by using sentence codes to condition the sentence generation.
1 code implementation • NAACL 2019 • Jiali Yao, Raphael Shu, Xinjian Li, Katsutoshi Ohtsuki, Hideki Nakayama
Input method editor (IME) converts sequential alphabet key inputs to words in a target language.
no code implementations • ICLR 2019 • Jiali Yao, Raphael Shu, Xinjian Li, Katsutoshi Ohtsuki, Hideki Nakayama
The input method is an essential service on every mobile and desktop devices that provides text suggestions.
no code implementations • 27 Sep 2018 • Raphael Shu, Hideki Nakayama
Planning is important for humans when producing complex languages, which is a missing part in current language generation models.
no code implementations • 14 Aug 2018 • Raphael Shu, Hideki Nakayama
Structural planning is important for producing long sentences, which is a missing part in current language generation models.
no code implementations • ACL 2018 • Raphael Shu, Hideki Nakayama
However, as the algorithm produces hypotheses in a monotonic left-to-right order, a hypothesis can not be revisited once it is discarded.
3 code implementations • ICLR 2018 • Raphael Shu, Hideki Nakayama
For each word, the composition of basis vectors is determined by a hash code.
Ranked #10 on Machine Translation on IWSLT2015 German-English
1 code implementation • 6 Jul 2017 • Raphael Shu, Hideki Nakayama
Neural machine translation models rely on the beam search algorithm for decoding.
no code implementations • 11 Apr 2017 • Raphael Shu, Hideki Nakayama
For extended periods of time, sequence generation models rely on beam search algorithm to generate output sequence.
no code implementations • WS 2017 • Raphael Shu, Hideki Nakayama
Recently, the attention mechanism plays a key role to achieve high performance for Neural Machine Translation models.
no code implementations • COLING 2016 • Natsuda Laokulrat, Sang Phan, Noriki Nishida, Raphael Shu, Yo Ehara, Naoaki Okazaki, Yusuke Miyao, Hideki Nakayama
Automatic video description generation has recently been getting attention after rapid advancement in image caption generation.
no code implementations • WS 2016 • Raphael Shu, Akiva Miura
To enhance Neural Machine Translation models, several obvious ways such as enlarging the hidden size of recurrent layers and stacking multiple layers of RNN can be considered.