This paper empirically studies whether BERT can really learn to conduct natural language inference (NLI) without utilizing hidden dataset bias; and how efficiently it can learn if it could.This is done via creating a simple entailment judgment case which involves only binary predicates in plain English.The results show that the learning process of BERT is very slow.However, the efficiency of learning can be greatly improved (data reduction by a factor of 1,500) if task-related features are added.This suggests that domain knowledge greatly helps when conducting NLI with neural networks.

Chao-Chun Liang, Yu-Shiang Wong, Yi-Chung Lin, Keh-Yih Su. Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers). 2018.

A goal-oriented meaning-based statistical framework is presented in this paper to solve the math word problem that requires multiple arithmetic operations with understanding, reasoning and explanation. It first analyzes and transforms sentences into their meaning-based logical forms, which represent the associated context of each quantity with role-tags (e.g., nsubj, verb, etc.). Logic forms with role-tags provide a flexible and simple way to specify the physical meaning of a quantity. Afterwards, the main-goal of the problem is decomposed recursively into its associated sub-goals. For each given sub-goal, the associated operator and operands are selected with statistical models. Lastly, it performs inference on logic expressions to get the answer and explains how the answer is obtained in a human comprehensible way. This process thus resembles the human cognitive understanding of the problem and produces a more meaningful problem solving interpretation.

In this paper, we propose a probabilistic approach to verifying the meaningful phrases, named concepts, in a user utterance. In this approach, a concept is verified not only according to its confidence measures assessed by various confidence measurement methods hut also according to neighboring concepts and their confidence levels. Two corpora are used to evaluate the proposed approach. One is collected from landline phone calls, and the other is collected from wireless phone calls. Experimental results show that the proposed model significantly outperforms the models using only confidence measures. The relative reduction in verification error is 23.7% on the corpus of landline phone calls, and 34.2% on the corpus of wireless phone calls. These substantial improvements show the proposed probabilistic verification approach is effective in integrating contextual confidence information and different confidence measures to verify the concepts of user utterances.

For telephone-based spoken dialogue systems, the responses to users should be specific and short. Therefore, it is highly demanded to segment a topical text into specific event segments which can be use to answer users’ queries. However, the lexical cohesion approach, which has been widely used to segment text into topics, is not suitable for segmenting text into smaller units, like events. In this paper, we present a two-stage approach to partition text into event segments. In the first stage, a trigram chunk tagger is used to label the segmentation tags. In the second stage, the unreliable segmentation tags are detected and then verified by a probabilistic verification model. Compared with the chunk tagger, the verification model can explore more contextual information and is less sensitive to the sparseness of training data. Experimental results show that the proposed two-stage approach significantly outperforms the chunk tagger approach. The improvements on precision and recall rates are 27% to 83% in different testing tasks.

Speech synthesis based on the Multiband Resynthesis OverLap-Add (MBROLA) algorithm produces high quality speech without requiring too much effort to design the diphone database, and using a low computational power.The main drawback of this algorithm is the slightly metallic sound or buzziness that can be perceived on voiced segments.We are working on a speech synthesizer based on the MBROLA algorithm, and trying to improve its speech quality by means of on an enhanced phase control strategy.

A joint learning algorithm is proposed in this paper to enhance the integrated speech and language model operating in the character-synchronous mode. A character-synchronous approach is first proposed to integrate speech and language information, including morphology and parts-of-speech, for ranking the candidates right after each Chinese character is uttered. Since the search space is cut down very efficiently by applying high-level knowledge in early time, the character-synchronous score function enables our system to operate in real time. To further enhance system performance, a joint learning algorithm is then derived to adjust all parameters of the speech and language processing modules simultaneously, according to their contributions to the overall discrimination power, to minimize the error rate. The proposed approaches are compared with a baseline system, which directly couples an HMM-based speech recognizer with a bigram language model. The performance of 75.71% character accuracy rate is obtained for the baseline system when it is tested on the task of recognizing 1000 Chinese spoken sentences with a very large vocabulary (90495 words) in the speaker-independent isolated-character mode. The character accuracy rate is improved to 88.26% with the character-synchronous approach, and 94.16% after the joint learning algorithm is applied.

A joint learning algorithm which enabled the parameters in an integrated speech and language model to be trained jointly, was proposed in this paper. The integrated model enhanced the spoken language system with high level knowledge, and operated in a character-synchronous mode. This integration model was tested on the task of recognizing isolated Chinese characters in the speaker independent mode with very large vocabulary of 90,495 words, and the performance of 88.26% character accuracy rate was obtained. In contrast, only 75.71% accuracy rate was achieved with the baseline system, which directly coupled the speech recognizer with a character bi-gram language module. Afterwards, the parameters of both speech and language modules were jointly adjusted according to their contribution in discrimination. The dynamic range variations among the parameters in different modules were also well tuned during the learning processes. After applying this procedure to the character-synchronous integration model, a very promising result of 94.16% character accuracy (75.96% error reduction rate) was obtained.>

To enhance the performance of spoken language processing, a unified framework is proposed to integrate speech and language information together. This framework uses probabilistic formulation to characterize different language analyses generated from a language processing module. As probabilistic formulations are used in both speech and language processing modules, information from both modules can be easily integrated. To further improve the performance, a discrimination and robustness oriented learning procedure is proposed to adjust the parameters of probabilistic formulations. Significant improvement has been observed in the task of reading aloud Chinese computer manuals, which operates in a speaker dependent, isolated word mode.>