Share This Author
Improving Language Understanding by Generative Pre-Training
The general task-agnostic model outperforms discriminatively trained models that use architectures speciﬁcally crafted for each task, improving upon the state of the art in 9 out of the 12 tasks studied.
Hierarchical Deep Reinforcement Learning: Integrating Temporal Abstraction and Intrinsic Motivation
- Tejas D. Kulkarni, Karthik Narasimhan, Ardavan Saeedi, J. Tenenbaum
- Computer ScienceNIPS
- 20 April 2016
h-DQN is presented, a framework to integrate hierarchical value functions, operating at different temporal scales, with intrinsically motivated deep reinforcement learning, and allows for flexible goal specifications, such as functions over entities and relations.
Language Understanding for Text-based Games using Deep Reinforcement Learning
This paper employs a deep reinforcement learning framework to jointly learn state representations and action policies using game rewards as feedback to map text descriptions into vector representations that capture the semantics of the game states.
Neural Generation of Regular Expressions from Natural Language with Minimal Domain Knowledge
- N. Locascio, Karthik Narasimhan, E. DeLeon, Nate Kushman, R. Barzilay
- Computer ScienceEMNLP
- 9 August 2016
This paper proposes a methodology for collecting a large corpus of regular expression, natural language pairs, and achieves a performance gain of 19.6% over previous state-of-the-art models.
An Unsupervised Method for Uncovering Morphological Chains
This work proposes a model for unsupervised morphological analysis that integrates orthographic and semantic views of words, and model word formation in terms of morphological chains, from base words to the observed words, breaking the chains into parent-child relations.
Improving Information Extraction by Acquiring External Evidence with Reinforcement Learning
This work explores the task of acquiring and incorporating external evidence to improve extraction accuracy in domains where the amount of training data is scarce, and employs a deep Q-network, trained to optimize a reward function that reflects extraction accuracy while penalizing extra effort.
A Generalized Algorithm for Multi-Objective Reinforcement Learning and Policy Adaptation
A generalized version of the Bellman equation is proposed to learn a single parametric representation for optimal policies over the space of all possible preferences in MORL, with the goal of enabling few-shot adaptation to new tasks.
Projection-Based Constrained Policy Optimization
- Karthik Narasimhan
- Computer ScienceICLR
- 30 April 2020
This paper proposes a new algorithm - Projection Based ConstrainedPolicy Optimization (PCPO), an iterative method for optimizing policies in a two-step process - the first step performs an unconstrained update while the second step reconciles the constraint violation by projection the policy back onto the constraint set.
Keep CALM and Explore: Language Models for Action Generation in Text-based Games
This paper proposes the Contextual Action Language Model (CALM) to generate a compact set of action candidates at each game state, and combines CALM with a reinforcement learning agent which re-ranks the generated action candidates to maximize in-game rewards.
Nonparametric Spherical Topic Modeling with Word Embeddings
This paper uses a Hierarchical Dirichlet Process for the base topic model and proposes an efficient inference algorithm based on Stochastic Variational Inference that enables it to naturally exploit the semantic structures of word embeddings while flexibly discovering the number of topics.