Few-Shot Structured Policy Learning for Multi-Domain and Multi-Task Dialogues

Reinforcement learning has been widely adopted to model dialogue managers in task-oriented dialogues. However, the user simulator provided by state-of-the-art dialogue frameworks are only rough approximations of human behaviour. The ability to learn from a small number of human interactions is hence crucial, especially on multi-domain and multi-task environments where the action space is large. We therefore propose to use structured policies to improve sample efficiency when learning on these kinds of environments. We also evaluate the impact of learning from human vs simulated experts. Among the different levels of structure that we tested, the graph neural networks (GNNs) show a remarkable superiority by reaching a success rate above 80% with only 50 dialogues, when learning from simulated experts. They also show superiority when learning from human experts, although a performance drop was observed, indicating a possible difficulty in capturing the variability of human strategies. We therefore suggest to concentrate future research efforts on bridging the gap between human data, simulators and automatic evaluators in dialogue frameworks.

Graph Neural Network Policies and Imitation Learning for Multi-Domain Task-Oriented Dialogues

Task-oriented dialogue systems are designed to achieve specific goals while conversing with humans. In practice, they may have to handle simultaneously several domains and tasks. The dialogue manager must therefore be able to take into account domain changes and plan over different domains/tasks in order to deal with multidomain dialogues. However, learning with reinforcement in such context becomes difficult because the state-action dimension is larger while the reward signal remains scarce. Our experimental results suggest that structured policies based on graph neural networks combined with different degrees of imitation learning can effectively handle multi-domain dialogues. The reported experiments underline the benefit of structured policies over standard policies.

Denoising Pre-Training and Data Augmentation Strategies for Enhanced RDF Verbalization with Transformers

The task of verbalization of RDF triples has known a growth in popularity due to the rising ubiquity of Knowledge Bases (KBs). The formalism of RDF triples is a simple and efficient way to store facts at a large scale. However, its abstract representation makes it difficult for humans to interpret. For this purpose, the WebNLG challenge aims at promoting automated RDF-to-text generation. We propose to leverage pre-trainings from augmented data with the Transformer model using a data augmentation strategy. Our experiment results show a minimum relative increases of 3.73%, 126.05% and 88.16% in BLEU score for seen categories, unseen entities and unseen categories respectively over the standard training.

Diluted Near-Optimal Expert Demonstrations for Guiding Dialogue Stochastic Policy Optimisation

A learning dialogue agent can infer its behaviour from interactions with the users. These interactions can be taken from either human-to-human or human-machine conversations. However, human interactions are scarce and costly, making learning from few interactions essential. One solution to speedup the learning process is to guide the agent's exploration with the help of an expert. We present in this paper several imitation learning strategies for dialogue policy where the guiding expert is a near-optimal handcrafted policy. We incorporate these strategies with state-of-the-art reinforcement learning methods based on Q-learning and actor-critic. We notably propose a randomised exploration policy which allows for a seamless hybridisation of the learned policy and the expert. Our experiments show that our hybridisation strategy outperforms several baselines, and that it can accelerate the learning when facing real humans.

Scaling up budgeted reinforcement learning

Can we learn a control policy able to adapt its behaviour in real time so as to take any desired amount of risk? The general Reinforcement Learning framework solely aims at optimising a total reward in expectation, which may not be desirable in critical applications. In stark contrast, the Budgeted Markov Decision Process (BMDP) framework is a formalism in which the notion of risk is implemented as a hard constraint on a failure signal. Existing algorithms solving BMDPs rely on strong assumptions and have so far only been applied to toy-examples. In this work, we relax some of these assumptions and demonstrate the scalability of our approach on two practical problems: a spoken dialogue system and an autonomous driving task. On both examples, we reach similar performances as Lagrangian Relaxation methods with a significant improvement in sample and memory efficiency.

Corrupt Bandits for Preserving Local Privacy

We study a variant of the stochastic multi-armed bandit (MAB) problem in which the rewards are corrupted. In this framework, motivated by privacy preservation in online recommender systems, the goal is to maximize the sum of the (unobserved) rewards, based on the observation of transformation of these rewards through a stochastic corruption process with known parameters. We provide a lower bound on the expected regret of any bandit algorithm in this corrupted setting. We devise a frequentist algorithm, KLUCB-CF, and a Bayesian algorithm, TS-CF and give upper bounds on their regret. We also provide the appropriate corruption parameters to guarantee a desired level of local privacy and analyze how this impacts the regret. Finally, we present some experimental results that confirm our analysis.

Random Forest for the Contextual Bandit Problem - extended version

To address the contextual bandit problem, we propose an online random forest algorithm. The analysis of the proposed algorithm is based on the sample complexity needed to find the optimal decision stump. Then, the decision stumps are assembled in a random collection of decision trees, Bandit Forest. We show that the proposed algorithm is optimal up to logarithmic factors. The dependence of the sample complexity upon the number of contextual variables is logarithmic. The computational cost of the proposed algorithm with respect to the time horizon is linear. These analytical results allow the proposed algorithm to be efficient in real applications, where the number of events to process is huge, and where we expect that some contextual variables, chosen from a large set, have potentially non- linear dependencies with the rewards. In the experiments done to illustrate the theoretical analysis, Bandit Forest obtain promising results in comparison with state-of-the-art algorithms.

Bandit Structured Prediction for Learning from Partial Feedback in Statistical Machine Translation

We present an approach to structured prediction from bandit feedback, called Bandit Structured Prediction, where only the value of a task loss function at a single predicted point, instead of a correct structure, is observed in learning. We present an application to discriminative reranking in Statistical Machine Translation (SMT) where the learning algorithm only has access to a 1-BLEU loss evaluation of a predicted translation instead of obtaining a gold standard reference translation. In our experiment bandit feedback is obtained by evaluating BLEU on reference translations without revealing them to the algorithm. This can be thought of as a simulation of interactive machine translation where an SMT system is personalized by a user who provides single point feedback to predicted translations. Our experiments show that our approach improves translation quality and is comparable to approaches that employ more informative feedback in learning.

A Relative Exponential Weighing Algorithm for Adversarial Utility-based Dueling Bandits

We study the K-armed dueling bandit problem which is a variation of the classical Multi-Armed Bandit (MAB) problem in which the learner receives only relative feedback about the selected pairs of arms. We propose a new algorithm called Relative Exponential-weight algorithm for Exploration and Exploitation (REX3) to handle the adversarial utility-based formulation of this problem. This algorithm is a non-trivial extension of the Exponential-weight algorithm for Exploration and Exploitation (EXP3) algorithm. We prove a finite time expected regret upper bound of order O(sqrt(K ln(K)T)) for this algorithm and a general lower bound of order omega(sqrt(KT)). At the end, we provide experimental results using real data from information retrieval applications.

Utility-based Dueling Bandits as a Partial Monitoring Game

Partial monitoring is a generic framework for sequential decision-making with incomplete feedback. It encompasses a wide class of problems such as dueling bandits, learning with expect advice, dynamic pricing, dark pools, and label efficient prediction. We study the utility-based dueling bandit problem as an instance of partial monitoring problem and prove that it fits the time-regret partial monitoring hierarchy as an easy - i.e. Theta (sqrt{T})- instance. We survey some partial monitoring algorithms and see how they could be used to solve dueling bandits efficiently. Keywords: Online learning, Dueling Bandits, Partial Monitoring, Partial Feedback, Multiarmed Bandits