Cas-GANs: An Approach of Dialogue Policy Learning based on GAN and RL Techniques

Abstract

Dialogue management systems are commonly applied in daily life, such as online shopping, hotel booking, and driving booking. Efficient dialogue management policy helps systems to respond to the user in an effective way. Policy learning is a complex task to build a dialogue system. There are different approaches have been proposed in the last decade to build a goal-oriented dialogue agent to train the systems with an efficient policy. The Generative adversarial network (GAN) is used in the dialogue generation, in previous works to build dialogue agents by selecting the optimal policy learning. The efficient dialogue policy learning aims to improve the quality of fluency and diversity for generated dialogues. Reinforcement learning (RL) algorithms are used to optimize the policies because the sequence is discrete. In this study, we have proposed a new technique called Cascade Generative Adversarial Network (Cas-GAN) that is combination of the GAN and RL for dialog generation. The Cas-GAN can model the relations between the dialogues (sentences) by using Graph Convolutional Networks (GCN). The graph nodes are consisting of different high level and low-level nodes representing the vertices and edges of the graph. Then, we use the maximum log-likelihood (MLL) approach to train the parameters and choose the best nodes. The experimental results compared with the HRL, RL agents and we got state-of-the-art results.

SBAG: A Hybrid Deep Learning Model for Large Scale Traffic Speed Prediction

“…In the future, further improvements can be made based on the proposed study and will improve more significant towards the graph structure to interpret and learn the spatial features or to improve by hybridizing with GANs [27] to make further amendments.…”

Reinforcement learning for content's customization: a first step of experimentation in Skyscanner

“…In this contest, reinforcement learning, a field of machine learning, is becoming of growing interest and has shown great potential in tackling automation tasks, e.g. it is used in a number of industries to define, for example, market forecasting (Moriyama et al, 2008) and can be applied in electronic marketplace transactions (Bandyopadhyay et al, 2006), in behaviour science to determine and analyse the associative mechanism as a tool for habit formation in healthy individuals (Gillan et al, 2015) or in combination with the generative adversarial network (GAN) can be used to build a goal-oriented dialogue agent to train systems with efficient policy (Nabeel et al, 2019).…”

“…In the tourism sector, there are few contribution on reinforcement learning; however, it can be used in a powerful way, thanks to its great potential in tackling automation tasks (Moriyama et al, 2008) and can be adapted from other industries exploiting its function to determine and forecast human behaviour and to reinforce dialogues (Gillan et al, 2015;Nabeel et al, 2019). From this point of view, some studies focussed their attention on the online bookings and the way to optimize them; in fact, in a research realized by Bondoux et al (2020) has been experimented the use of reinforcement learning in the management of the airlines revenue systems, and in previous research, it was used to set a correct price for airline tickets considering specific time horizon for booking (Kulkarny et al, 2011).…”