Yike Wu, Yi Huang, Nan Hu, Yuncheng Hua, Guilin Qi, Jiaoyan Chen, Jeff Z. Pan. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

The modern digital world and associated innovative and state-of-the-art applications that characterize its presence, render the current digital age a captivating era for many worldwide. These innovations include dialogue systems, such as Apple’s Siri, Google Now, and Microsoft’s Cortana, that stay on the personal devices of users and assist them in their daily activities. These systems track the intentions of users by analyzing their speech, context by looking at their previous turns, and several other external details, and respond or act in the form of speech output. For these systems to work efficiently, a dialogue state tracking (DST) module is required to infer the current state of the dialogue in a conversation by processing previous states up to the current state. However, developing a DST module that tracks and exploit dialogue states effectively and accurately is challenging. The notable challenges that warrant immediate attention include scalability, handling the unseen slot-value pairs during training, and retraining the model with changes in the domain ontology. In this article, we present a new end-to-end framework by combining BERT, Stacked Bidirectional LSTM (BiLSTM), and a multiple attention mechanism to formalize DST as a classification problem and address the aforementioned issues. The BERT-based module encodes the user’s and system’s utterances. The Stacked BiLSTM extracts the contextual features and multiple attention mechanisms to calculate the attention between its hidden states and the utterance embeddings. We experimentally evaluated our method against the current approaches over a variety of datasets. The results indicate a significant overall improvement. The proposed model is scalable in terms of sharing the parameters and it considers the unseen instances during training.

Controllable Text Generation (CTG) aims to modify the output of a Language Model (LM) to meet specific constraints. For example, in a customer service conversation, responses from the agent should ideally be soothing and address the user’s dissatisfaction or complaints. This imposes significant demands on controlling language model output. However, demerits exist among traditional methods. Promoting and fine-tuning language models exhibit the “hallucination” phenomenon and cannot guarantee complete adherence to constraints. Conditional language models (CLM), which map control codes into LM representations or latent space, require training the modified language models from scratch and a high amount of customized dataset is demanded. Decoding-time methods employ Bayesian Rules to modify the output of the LM or model constraints as a combination of energy functions and update the output along the low-energy direction. Both methods are confronted with the efficiency sampling problem. Moreover, there are no methods that consider the relation between constraints weights and the contexts, as is essential in actual applications such as customer service scenarios. To alleviate the problems mentioned above, we propose Controllable Text Generation with Generative Adversarial Networks (CTGGAN), which utilizes a language model with logits bias as the Generator to produce constrained text and employs the Discriminator with learnable constraint weight combinations to score and update the generation. We evaluate the method in the text completion task and Chinese customer service dialogues scenario, and our method shows superior performance in metrics such as PPL and Dist-3. In addition, CTGGAN also exhibits efficient decoding compared to other methods.

To solve collaborative tasks, Multi-Agent Systems (MAS) have been widely used because of their collaboration capabilities and flexibility. They have been applied in various fields, such as robotic collaboration and intelligent transportation systems. In multi-agent collaborative problems, coordinating the collaboration strategies between agents is the most challenging aspect. Traditional Multi-Agent Reinforcement Learning (MARL) methods attempt to transform the team objective into individual objectives for each agent through value decomposition. However, the value allocation mechanisms in existing methods struggle to accurately measure the contribution differences between agents. In this paper, we propose an MARL method based on structured coordination by leveraging the local interaction structure among agents. Depending on the application scenarios, we present explicit and implicit implementation approaches. The explicit method constructs an explicit collaboration graph based on the cooperation relationships between agents, and then uses Shapley values to assess the contribution of each agent in the collaboration graph. Additionally, we propose a method based on self-attention mechanisms to dynamically construct implicit interaction structures. Experimental results demonstrate that, compared to several state-of-the-art multi-agent collaboration algorithms, the two collaboration methods we proposed achieve significant performance advantages across multiple complex collaboration scenarios.