Recent advancements in open-domain dialogue systems have been propelled by the emergence of high-quality large language models (LLMs) and various effective training methodologies. Nevertheless, the presence of toxicity within these models presents a significant challenge that can potentially diminish the user experience. In this study, we introduce an innovative training algorithm, an improvement upon direct preference optimization (DPO), called adversarial DPO (ADPO). The ADPO algorithm is designed to train models to assign higher probability distributions to preferred responses and lower distributions to unsafe responses, which are self-generated using the toxic control token. We demonstrate that ADPO enhances the model’s resilience against harmful conversations while minimizing performance degradation. Furthermore, we illustrate that ADPO offers a more stable training procedure compared to the traditional DPO. To the best of our knowledge, this is the first adaptation of the DPO algorithm that directly incorporates harmful data into the generative model, thereby reducing the need to artificially create safe dialogue data.

Recent dialogue systems typically operate through turn-based spoken interactions between users and agents. These systems heavily depend on accurate Automatic Speech Recognition (ASR), as transcription errors can significantly degrade performance in downstream dialogue tasks. To alleviate this challenge, robust ASR is required, and one effective method is to utilize the dialogue context from user and agent interactions for transcribing the subsequent user utterance. This method incorporates the transcription of the user’s speech and the agent’s response as model input, using the accumulated context generated by each turn. However, this context is susceptible to ASR errors because the ASR model generates it auto-regressively. Such noisy context can further degrade the benefits of context input, resulting in suboptimal ASR performance. In this paper, we introduce context noise representation learning to enhance robustness against noisy context, ultimately improving dialogue speech recognition accuracy. To maximize the advantage of context awareness, our approach involves decoder pre-training with text-based dialogue data and noise representation learning for a context encoder. Evaluated on DSTC11 (MultiWoZ 2.1 audio dialogues), it achieves a 24% relative reduction in Word Error Rate (WER) compared to wav2vec2.0 baselines and a 13% reduction compared to Whisper-large-v2. Notably, in noisy environments where user speech is barely audible, our method proves its effectiveness by utilizing contextual information for accurate transcription. Tested on audio data with strong noise level (Signal Noise Ratio of 0dB), our approach shows up to a 31% relative WER reduction compared to the wav2vec2.0 baseline, providing a reassuring solution for real-world noisy scenarios.

We propose a Korean multimodal dialogue system targeting emotion-based empathetic dialogues because most research in this field has been conducted in a few languages such as English and Japanese and in certain circumstances. Our dialogue system consists of an emotion detector, an empathetic response generator, a monitoring interface, a voice activity detector, a speech recognizer, a speech synthesizer, a gesture classification, and several controllers to provide both multimodality and empathy during a conversation between a human and a machine. For comparisons across visual influence on users, our dialogue system contains two versions of the user interface, a cat face-based user interface and an avatar-based user interface. We evaluated our dialogue system by investigating the dialogues in text and the average mean opinion scores under three different visual conditions, no visual, the cat face-based, and the avatar-based expressions. The experimental results stand for the importance of adequate visual expressions according to user utterances.

Research on open-domain dialogue systems that allow free topics is challenging in the field of natural language processing (NLP). The performance of the dialogue system has been improved recently by the method utilizing dialogue-related knowledge; however, non-English dialogue systems suffer from reproducing the performance of English dialogue systems because securing knowledge in the same language with the dialogue system is relatively difficult. Through experiments with a Korean dialogue system, this paper proves that the performance of a non-English dialogue system can be improved by utilizing English knowledge, highlighting the system uses cross-lingual knowledge. For the experiments, we 1) constructed a Korean version of the Wizard of Wikipedia dataset, 2) built Korean-English T5 (KE-T5), a language model pre-trained with Korean and English corpus, and 3) developed a knowledge-grounded Korean dialogue model based on KE-T5. We observed the performance improvement in the open-domain Korean dialogue model even only English knowledge was given. The experimental results showed that the knowledge inherent in cross-lingual language models can be helpful for generating responses in open dialogue systems.

Backchannel (BC), a short reaction signal of a listener to a speaker’s utterances, helps to improve the quality of the conversation. Several studies have been conducted to predict BC in conversation; however, the utilization of advanced natural language processing techniques using lexical information presented in the utterances of a speaker has been less considered. To address this limitation, we present a BC prediction model called BPM_MT (Backchannel prediction model with multitask learning), which utilizes KoBERT, a pre-trained language model. The BPM_MT simultaneously carries out two tasks at learning: 1) BC category prediction using acoustic and lexical features, and 2) sentiment score prediction based on sentiment cues. BPM_MT exhibited 14.24% performance improvement compared to the existing baseline in the four BC categories: continuer, understanding, empathic response, and No BC. In particular, for empathic response category, a performance improvement of 17.14% was achieved.