Filters








2,047 Hits in 5.0 sec

Sequence-to-Sequence Language Grounding of Non-Markovian Task Specifications

Nakul Gopalan, Dilip Arumugam, Lawson Wong, Stefanie Tellex
2018 Robotics: Science and Systems XIV  
Due to the non-Markovian nature of the objective, approaches in the former category must map to potentially unbounded action sequences whereas approaches in the latter category would require folding the  ...  While demonstrating that standard neural sequence-to-sequence learning models can successfully ground language to this semantic representation, we also provide analysis that highlights generalization to  ...  Since we want to ground language for non-Markovian specifications in potentially novel and stochastic environments, we adopt LTL as our target specification in this paper, and will only review grounding  ... 
doi:10.15607/rss.2018.xiv.067 dblp:conf/rss/GopalanAWT18 fatcat:mlhrku6zrzfbbjbco62wbmnsva

Is Semantics Physical?! [article]

Maria K. Koleva
2010 arXiv   pre-print
The boundedness sets an exclusive two-fold representation of a semantic unit: as a specific sequence of letters and as a performance of a specific engine so that their interplay serves as grounds for building  ...  The robustness of the hierarchical organization of semantic structures is maintained by new generic form of non-local feedback that appears as a result of the necessary for sustaining boundeness matter  ...  Thus, our task now is to show that under the constraint of boundedness, any sequence of symbols and the corresponding sequence of words (viewed as a non-uniform coarse-graining of the sequence of symbols  ... 
arXiv:1009.1470v1 fatcat:3s7pca6geveo7f6xsygauc2huu

Generalized Inverse Planning: Learning Lifted non-Markovian Utility for Generalizable Task Representation [article]

Sirui Xie and Feng Gao and Song-Chun Zhu
2020 arXiv   pre-print
shift, the utility also needs to be lifted to abstract out specific grounding objects.  ...  In searching for a generalizable representation of temporally extended tasks, we spot two necessary constituents: the utility needs to be non-Markovian to transfer temporal relations invariant to a probability  ...  Yixin Zhu of UCLA Statistics Department, Prof. Guy Van den Broeck of UCLA Computer Science Department for useful discussions.  ... 
arXiv:2011.09854v1 fatcat:ttukjs6urzb5pahuxtpr3x33lu

A Tale of Two DRAGGNs: A Hybrid Approach for Interpreting Action-Oriented and Goal-Oriented Instructions

Siddharth Karamcheti, Edward Clem Williams, Dilip Arumugam, Mina Rhee, Nakul Gopalan, Lawson L.S. Wong, Stefanie Tellex
2017 Proceedings of the First Workshop on Language Grounding for Robotics  
Our robotsimulation results demonstrate that a system successfully interpreting both goaloriented and action-oriented task specifications brings us closer to robust natural language understanding for human-robot  ...  We introduce a new hybrid approach, the Deep Recurrent Action-Goal Grounding Network (DRAGGN), for task grounding and execution that handles natural language from either category as input, and generalizes  ...  In Conference on Empirical Methods in Natural Language Processing. Yoav Artzi and Luke Zettlemoyer. 2013. Weakly supervized learning of semantic parsers for mapping  ... 
doi:10.18653/v1/w17-2809 dblp:conf/acl/KaramchetiWARGW17 fatcat:aqv5k2cd6nazxbce6r5ig46wdq

A Tale of Two DRAGGNs: A Hybrid Approach for Interpreting Action-Oriented and Goal-Oriented Instructions [article]

Siddharth Karamcheti, Edward C. Williams, Dilip Arumugam, Mina Rhee, Nakul Gopalan, Lawson L. S. Wong, Stefanie Tellex
2017 arXiv   pre-print
Our robot-simulation results demonstrate that a system successfully interpreting both goal-oriented and action-oriented task specifications brings us closer to robust natural language understanding for  ...  We introduce a new hybrid approach, the Deep Recurrent Action-Goal Grounding Network (DRAGGN), for task grounding and execution that handles natural language from either category as input, and generalizes  ...  In Conference on Empirical Methods in Natural Language Processing. Yoav Artzi and Luke Zettlemoyer. 2013. Weakly supervized learning of semantic parsers for mapping instructions to actions.  ... 
arXiv:1707.08668v1 fatcat:ifbsnjm3pvdqzd7nedvs55qi2e

ElGolog: A High-Level Programming Language with Memory of the Execution History

Giuseppe De Giacomo, Yves Lespérance, Eugenia Ternovska
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Most programming languages only support tests that refer exclusively to the current state. This applies even to high-level programming languages based on the situation calculus such as Golog.  ...  The result is that additional variables/fluents/data structures must be introduced to track conditions that the program uses in tests to make decisions.  ...  part by the European Research Council under the European Union's Horizon 2020 programme through the ERC Advanced Grant WhiteMec (No. 834228), and the National Science and Engineering Research Council of  ... 
doi:10.1609/aaai.v34i03.5669 fatcat:hxje6o4dvbhqtg7cleiwwpxcfi

On Extractive and Abstractive Neural Document Summarization with Transformer Language Models [article]

Sandeep Subramanian, Raymond Li, Jonathan Pilault, Christopher Pal
2020 arXiv   pre-print
We perform a simple extractive step before generating a summary, which is then used to condition the transformer language model on relevant information before being tasked with generating a summary.  ...  We present a method to produce abstractive summaries of long documents that exceed several thousand words via neural abstractive summarization.  ...  Our results are based on a set of eleven task specific tasks that are benchmarks for a large suite of nlp tasks.  ... 
arXiv:1909.03186v2 fatcat:4t33csrjzvdhlai2fkzw5e5hdi

Metaphysics of Planning Domain Descriptions

Siddharth Srivastava, Stuart Russell, Alessandro Pinto
2016 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
to produce correct solutions from initially incorrect or non-Markovian abstract models.  ...  There is some evidence that such limitations have restricted the applicability of AI planning technology in the real world, as is apparent in the case of task and motion planning in robotics.  ...  Non-Markovian Abstractions In this section we derive conditions under an abstraction will be non-Markovian. A non-Markovian residue constitutes the evidence for an abstraction's being non-Markovian.  ... 
doi:10.1609/aaai.v30i1.10118 fatcat:hooh4ajjpferjehukig4vbw3zu

Variational Latent-State GPT for Semi-supervised Task-Oriented Dialog Systems [article]

Hong Liu, Yucheng Cai, Zhenru Lin, Zhijian Ou, Yi Huang, Junlan Feng
2022 arXiv   pre-print
The inference model in VLS-GPT is non-Markovian due to the use of the Transformer architecture.  ...  Recently, two approaches, fine-tuning large pre-trained language models and variational training, have attracted significant interests, separately, for semi-supervised end-to-end task-oriented dialog (  ...  language processing tasks.  ... 
arXiv:2109.04314v2 fatcat:j6toboql2bezjl67zibl5bcaey

DeepSynth: Automata Synthesis for Automatic Task Segmentation in Deep Reinforcement Learning [article]

Mohammadhosein Hasanbeig, Natasha Yogananda Jeppu, Alessandro Abate, Tom Melham, Daniel Kroening
2021 arXiv   pre-print
The proposed approach is able to cope with both high-dimensional, low-level features and unknown sparse non-Markovian rewards.  ...  achieving an unknown sequence of high-level objectives.  ...  Specifically, we relate the black-box MDP and the automaton by synchronising them on-the-fly to create a new structure that breaks down a non-Markovian task into a set of Markovian, history-independent  ... 
arXiv:1911.10244v5 fatcat:c2r47b7aafdbbkadkltxocclde

Efficient Exploration of Hamiltonian Parameter Space for Optimal Control of Non-Markovian Open Quantum Systems

Gerald E. Fux, Eoin P. Butler, Paul R. Eastham, Brendon W. Lovett, Jonathan Keeling
2021 Physical Review Letters  
We present a general method to efficiently design optimal control sequences for non-Markovian open quantum systems, and illustrate it by optimizing the shape of a laser pulse to prepare a quantum dot in  ...  a specific state.  ...  -We have shown that the PT-TEMPO method makes optimal control of non-Markovian open quantum systems a feasible task.  ... 
doi:10.1103/physrevlett.126.200401 pmid:34110219 fatcat:ufan2of2vvayjeaghts6qawx4a

Neural Physicist: Learning Physical Dynamics from Image Sequences [article]

Baocheng Zhu, Shijun Wang, James Zhang
2020 arXiv   pre-print
state at each time step, neural process (NP) to extract the global system parameters, and a non-linear non-recurrent stochastic state space model to learn the physical dynamic transition.  ...  We present a novel architecture named Neural Physicist (NeurPhy) to learn physical dynamics directly from image sequences using deep neural networks.  ...  Note that learning a manifold of using one dimension to encode m is non-trivial, as the pendulum's mass m can only be extracted from the time evolution of the image sequence.  ... 
arXiv:2006.05044v1 fatcat:dn2lzqfpmzaazmjzql35onyare

Formal grammar and information theory: together again?

F. Pereira
2000 Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences  
of language by counting occurrences of linguistic events in very large bodies of text and speech.  ...  Zellig Harris had advocated a close alliance between grammatical and information-theoretic principles in the analysis of natural language, and early formal-language theory provided another strong link  ...  in the study of language that most of my work derives from.  ... 
doi:10.1098/rsta.2000.0583 fatcat:pfm2cu4eargphjxrdieln5auti

Renormalization of Hierarchy and Semantic Computing

Maria K. Koleva
2018 Journal of Modern Physics  
non-extensive reduction of computation costs and non-extensive speeding up of computing.  ...  The fundamental novelty of that renormalization is provided by a highly non-trivial interplay between structural and functional properties.  ...  However, the problem of non-ambiguous convergence to a single state is left open. One of the most important cases is that of Markovian process.  ... 
doi:10.4236/jmp.2018.93024 fatcat:bu46fkjkkbfyvh4hw65wdixgau

Deep Variational Bayes Filters: Unsupervised Learning of State Space Models from Raw Data [article]

Maximilian Karl, Maximilian Soelch, Justin Bayer, Patrick van der Smagt
2017 arXiv   pre-print
We introduce Deep Variational Bayes Filters (DVBF), a new method for unsupervised learning and identification of latent Markovian state space models.  ...  Thus, it can handle highly nonlinear input data with temporal and spatial dependencies such as image sequences without domain knowledge.  ...  We would like to thank Jost Tobias Springenberg, Adam Kosiorek, Moritz Münst, and anonymous reviewers for valuable input.  ... 
arXiv:1605.06432v3 fatcat:fpcorlx3z5dffilzrix2iorwyy
« Previous Showing results 1 — 15 out of 2,047 results