Filters








75 Hits in 7.3 sec

Specifying and Interpreting Reinforcement Learning Policies through Simulatable Machine Learning [article]

Pradyumna Tambwekar, Andrew Silva, Nakul Gopalan, Matthew Gombolay
2022 arXiv   pre-print
Finally, to close the loop on human-specification, we produce explanations of the final learned policy, in multiple modalities, to provide the user a final depiction about the learned policy of the agent  ...  Human-AI collaborative policy synthesis is a procedure in which (1) a human initializes an autonomous agent's behavior, (2) Reinforcement Learning improves the human specified behavior, and (3) the agent  ...  Policy Explanation User Study: Additional Details All participants were recruited on mechanical turk.  ... 
arXiv:2101.07140v3 fatcat:h5a2ltgz3vdllix4blr55uumxu

Bottom-up learning of explicit knowledge using a Bayesian algorithm and a new Hebbian learning rule

Sébastien Hélie, Robert Proulx, Bernard Lefebvre
2011 Neural Networks  
In TELECAST, implicit processing is modeled using an unsupervised connectionist network (the Joint Probability EXtractor: JPEX) while explicit (causal) knowledge is implemented using a Bayesian belief  ...  This phenomenon has recently received much attention in empirical research that was not accompanied by a corresponding work effort in cognitive modeling.  ...  The CLARION model was composed of two feedforward connectionist networks.  ... 
doi:10.1016/j.neunet.2010.12.002 pmid:21239141 fatcat:slaaekszhbak7fznyf42k2ysbe

Deep Learning in Science [article]

Stefano Bianchini, Moritz Müller, Pierre Pelletier
2020 arXiv   pre-print
Therefore, we empirically assess how DL adoption relates to re-combinatorial novelty and scientific impact in the health sciences.  ...  Much of the recent success of Artificial Intelligence (AI) has been spurred on by impressive achievements within a broader family of machine learning methods, commonly referred to as Deep Learning (DL)  ...  of the empirical evidence on the nexus between re-combinatorial novelty and impact comes from bibliometric studies on scientific publications and patents.  ... 
arXiv:2009.01575v2 fatcat:4ttqgjdjfjbydp7flnhcgg5p7m

Implicit assumptions about implicit learning

Keith J. Holyoak, Merideth Gattis
1994 Behavioral and Brain Sciences  
The assumption that learning in some of these tasks (e.g., artificial grammar learning) is predominantly based on rule abstraction is questionable.  ...  A number of ways of taxonomizing human learning have been proposed. We examine the evidence for one such proposal, namely, that there exist independent explicit and implicit learning systems.  ...  We would like to express our considerable gratitude to colleagues who have commented on this work, especially Axel  ... 
doi:10.1017/s0140525x00035159 fatcat:fqtxatogcnbdndgdn7vujkausq

Word learning as Bayesian inference

Fei Xu, Joshua B. Tenenbaum
2007 Psychological review  
The main alternatives to hypothesis elimination are based on some form of associative learning, such as connectionist networks (using internal layers of "hidden" units and appropriately designed input  ...  While both hypothesis elimination and associative learning models offer certain important insights, we will argue that neither approach provides an adequate framework for explaining how people learn the  ...  Author Note The authors contributed equally to this work and they are listed in order of birth. This  ... 
doi:10.1037/0033-295x.114.2.245 pmid:17500627 fatcat:duhdlg5cbfa2tfjnxyjqezqjfq

Deep Learning for Text Style Transfer: A Survey [article]

Di Jin, Zhijing Jin, Zhiting Hu, Olga Vechtomova, Rada Mihalcea
2021 arXiv   pre-print
Text style transfer is an important task in natural language generation, which aims to control certain attributes in the generated text, such as politeness, emotion, humor, and many others.  ...  It has a long history in the field of natural language processing, and recently has re-gained significant attention thanks to the promising performance brought by deep neural models.  ...  In Proceedings of the learning style instance supported latent 2018 Conference on Empirical Methods in space.  ... 
arXiv:2011.00416v5 fatcat:wfw3jfh2mjfupbzrmnztsqy4ny

A neurocomputational account of taxonomic responding and fast mapping in early word learning

Julien Mayor, Kim Plunkett
2010 Psychological review  
The model demonstrates how an established constraint on lexical learning which has often been regarded as domain-specific can emerge from domain-general learning principles that are simultaneously biologically  ...  We present a neuro-computational model using self-organising maps that accounts for the emergence of taxonomic responding and fast mapping in early word learning, as well as a rapid increase in the rate  ...  Classical backpropagation networks need constant supervision provided by an error-correction learning algorithm in order to generate meaningful patterns of generalisation.  ... 
doi:10.1037/a0018130 pmid:20063962 fatcat:gqjmauc4hvgm5kfr2jfz4hbo64

Spatial Learning and Action Planning in a Prefrontal Cortical Network Model

Louis-Emmanuel Martinet, Denis Sheynikhovich, Karim Benchenane, Angelo Arleo, Olaf Sporns
2011 PLoS Computational Biology  
Specific connectivity and synaptic adaptation principles shape the recurrent dynamics of the network arranged in cortical minicolumns.  ...  We model a prefrontal network mediating distributed information processing for spatial learning and action planning.  ...  Adding borders or barriers would result in the ''recruitment'' of new C 2 units preferentially active on either one side or the other of the barriers.  ... 
doi:10.1371/journal.pcbi.1002045 pmid:21625569 pmcid:PMC3098199 fatcat:7gaw35j4jjg3hmetpv4vlyduiu

The exploitation of distributional information in syllable processing

Pierre Perruchet, Ronald Peereman
2004 Journal of Neurolinguistics  
We then analyze the ability of computational models to account for these results, successively considering a connectionist model based on the automatic computation of statistical regularities (SRN) [Cogn  ...  In order to address this issue, we first propose an overview of some basic measures of association, going from the simple co-occurrence frequency to the normative measure of contingency, r w : We then  ...  Acknowledgements We would like to thank Padraic Monoghan and Zoltan Dienes for their valuable comments on an earlier version of this manuscript, and Aurelien Perruchet for his contribution to Appendix  ... 
doi:10.1016/s0911-6044(03)00059-9 fatcat:vgmvm3my7rd7jneqx7ggnkpygm

Deep Learning for Text Style Transfer: A Survey

Di Jin, Zhijing Jin, Zhiting Hu, Olga Vechtomova, Rada Mihalcea
2021 Computational Linguistics  
Text style transfer is an important task in natural language generation, which aims to control certain attributes in the generated text, such as politeness, emotion, humor, and many others.  ...  It has a long history in the field of natural language processing, and recently has re-gained significant attention thanks to the promising performance brought by deep neural models.  ...  In Proceedings of the learning style instance supported latent 2018 Conference on Empirical Methods in space.  ... 
doi:10.1162/coli_a_00426 fatcat:v7vmb62ckfcu5k5mpu2pydnrxy

Modeling language acquisition in atypical phenotypes

Michael S. C. Thomas, Annette Karmiloff-Smith
2003 Psychological review  
An increasing number of connectionist models have been proposed to explain behavioral deficits in developmental disorders.  ...  The authors examine these issues in depth with respect to a series of new simulations investigating past-tense formation in Williams syndrome.  ...  Connectionist Models of Past-Tense Formation Connectionist theories of past-tense formation have converged on construing this domain in terms of an associative system that learns the relationship between  ... 
doi:10.1037/0033-295x.110.4.647 pmid:14599237 fatcat:5p5kiv252rhudg4sf2xqupqfda

Hocus-Socus: An Error Catastrophe for Complex Hebbian Learning Implies Neocortical Proofreading [article]

Kingsley J.A. Cox, Paul R. Adams
2010 arXiv   pre-print
Because output firing depends on input firing via the relevant connection strengths, Hebbian learning provides, in a feedback manner, sensitivity to input correlations.  ...  Thus if 3 pixels correlate, they may give an "edge".  ...  Acknowledgements We thank the following for comments on a draft of this paper: H. Barlow, P. Dayan  ... 
arXiv:1012.0946v1 fatcat:pr7zjimj4ja6bpqvarz2dhjcrq

Are imaging and lesioning convergent methods for assessing functional specialisation? Investigations using an artificial neural network

Michael S.C. Thomas, Harry R.M. Purser, Simon Tomlinson, Denis Mareschal
2012 Brain and Cognition  
SBI therefore distinguishes the ways in which different processes (e.g., producing regular vs. irregular verbs) exploit the same fixed network connectivity structure to drive behaviour (see also Sanger  ...  The model comprised two processing 'routes': one a direct route between layers of input and output units, while the other, indirect, route featured an intermediate layer of processing units.  ...  Acknowledgements This work was supported by European Commission grant NEST-029088(ANALOGY), ESRC grant RES-062-23-2721, and a Leverhulme Study Abroad Fellowship to MT.  ... 
doi:10.1016/j.bandc.2011.10.003 pmid:22088777 fatcat:m32iatirq5dcjatgov7rz3liam

Critical periods and catastrophic interference effects in the development of self-organizing feature maps

Fiona M. Richardson, Michael S.C. Thomas
2008 Developmental Science  
We explored the relative contribution of network parameters (for example, whether learning rate and neighbourhood reduce across training), the representational resources available to the network, and the  ...  We argue that the impact of map organization on behaviour must be interpreted in terms of the cognitive processes that the map is driving.  ...  For example, at least one popular computational methodology for studying development -backpropagation connectionist networks -has indicated that catastrophic interference may be a serious problem for the  ... 
doi:10.1111/j.1467-7687.2008.00682.x pmid:18466371 fatcat:nkkx4ke22raitmg5c7sl25k4nu

Unsupervised learning for vascular heterogeneity assessment of glioblastoma based on magnetic resonance imaging: The Hemodynamic Tissue Signature [article]

Javier Juan-Albarracín
2020 arXiv   pre-print
This thesis focuses on the research and development of the Hemodynamic Tissue Signature (HTS) method: an unsupervised machine learning approach to describe the vascular heterogeneity of glioblastomas by  ...  The HTS builds on the concept of habitats. An habitat is defined as a sub-region of the lesion with a particular MRI profile describing a specific physiological behavior.  ...  ) o 1 = 0.0582 → (vs. t 2 = 0.05) concluding the training of the network.  ... 
arXiv:2009.06288v1 fatcat:dum2y7fuuve73lxbb2any6iak4
« Previous Showing results 1 — 15 out of 75 results