Dialog Policy Learning for Joint Clarification and Active Learning Queries

  title={Dialog Policy Learning for Joint Clarification and Active Learning Queries},
  author={Aishwarya Padmakumar and Raymond J. Mooney},
Intelligent systems need to be able to recover from mistakes, resolve uncertainty, and adapt to novel concepts not seen during training. Dialog interaction can enable this by the use of clarifications for correction and resolving uncertainty, and active learning queries to learn new concepts encountered during operation. Prior work on dialog systems has either focused on exclusively learning how to perform clarification/ information seeking, or to perform active learning. In this work, we train… 

Figures and Tables from this paper

Deciding Whether to Ask Clarifying Questions in Large-Scale Spoken Language Understanding

A neural self-attentive model is proposed that leverages the hypotheses with ambiguities and contextual signals to trigger clarifying questions only when necessary for the user satisfaction.

What does the User Want? Information Gain for Hierarchical Dialogue Policy Optimisation

This work proposes the usage of an intrinsic reward based on information gain that enables the policy to learn how to retrieve the users' needs efficiently, which is an integral aspect in every task-oriented conversation.

Part2Whole: Iteratively Enrich Detail for Cross-Modal Retrieval with Partial Query

This work introduces the partialquery problem and extensively analyze its influence on textbased image retrieval, and proposes an interactive retrieval framework called Part2Whole to tackle this problem by iteratively enriching the missing details.

Ask&Confirm: Active Detail Enriching for Cross-Modal Retrieval with Partial Query

This work introduces the partial-query problem and extensively analyze its influence on text-based image retrieval, and proposes a novel retrieval framework that conducts the interactive process in an Ask-and-Confirm fashion, where AI actively searches for discriminative details missing in the current query, and users only need to confirm AI’s proposal.



Answerer in Questioner's Mind for Goal-Oriented Visual Dialogue

This work proposes "Answerer in Questioner's Mind" (AQM), a novel algorithm for goal-oriented dialogue that outperforms comparative algorithms and makes human-like dialogue.

Active One-shot Learning

A recurrent neural network based action-value function is presented, and its ability to learn how and when to request labels is demonstrated, and the model can achieve a higher prediction accuracy than a similar model on a purely supervised task, or trade prediction accuracy for fewer label requests.

Asynchronous Methods for Deep Reinforcement Learning

A conceptually simple and lightweight framework for deep reinforcement learning that uses asynchronous gradient descent for optimization of deep neural network controllers and shows that asynchronous actor-critic succeeds on a wide variety of continuous motor control problems as well as on a new task of navigating random 3D mazes using a visual input.

POMDP-Based Statistical Spoken Dialog Systems: A Review

This review article provides an overview of the current state of the art in the development of POMDP-based spoken dialog systems.

Opportunistic Active Learning for Grounding Natural Language Descriptions

It is demonstrated that inquisitive behavior—asking users important questions about the meanings of words that may be off-topic for the current dialog—leads to identifying the correct object more often over time in an object identification setting.

Dialog-based Interactive Image Retrieval

A new approach to interactive image search that enables users to provide feedback via natural language, allowing for more natural and effective interaction, and achieves better accuracy than other supervised and reinforcement learning baselines.

Can I Help You? - The Acceptance of Intelligent Personal Assistants

This study investigates how the perceived advantages and disadvantages of using an IPA affect its acceptance and shows that the advantages have a higher impact on acceptance than the disadvantages.

The iMaterialist Fashion Attribute Dataset

This work contributes to the community a new dataset called iMaterialist Fashion Attribute (iFashion-Attribute), constructed from over one million fashion images with a label space that includes 8 groups of 228 fine-grained attributes in total, which is the first known million-scale multi-label and fine- grained image dataset.

Multimodal Dialog for Browsing Large Visual Catalogs using Exploration-Exploitation Paradigm in a Joint Embedding Space

A slightly asymmetric version of a complete MMD system that can understand both text and image queries, but responds only in images is presented, to assist online customers in visually browsing through large catalogs.

Learning a Policy for Opportunistic Active Learning

This work uses reinforcement learning for an object retrieval task, to learn a policy that effectively trades off task completion with model improvement that would benefit future tasks.