Filters








17,883 Hits in 6.1 sec

Reliable Visual Question Answering: Abstain Rather Than Answer Incorrectly [article]

Spencer Whitehead, Suzanne Petryk, Vedaad Shakib, Joseph Gonzalez, Trevor Darrell, Anna Rohrbach, Marcus Rohrbach
2022 arXiv   pre-print
This motivates us to utilize a multimodal selection function to directly estimate the correctness of the predicted answers, which we show can increase the coverage by, for example, 2.4x from 6.8% to 16.3%  ...  ., abstain from answering a question), such ability has been largely neglected in multimodal research, despite the importance of this problem to the usage of VQA in real settings.  ...  Authors, as part of their affiliation with UC Berkeley, were supported in part by the NSF CISE Expeditions Award CCF-1730628; DoD, including DARPA's LwLL, PTG, and/or SemaFor programs; the Berkeley Artificial  ... 
arXiv:2204.13631v2 fatcat:3khp63qfmbgcpmb6ej7qwjjwfy

Subsumption-free bottom-up evaluation of logic programs with partially instantiated data structures [chapter]

Zoltan Somogyi, David Kemp, James Harland, Kotagiri Ramamohanarao
1994 Lecture Notes in Computer Science  
This class is based on the notion of modes, which generalise adornments by not requiring that \bound" be synonymous with \ground".  ...  We study a class of programs that can handle terms with both variables and function symbols but for which subsumption tests are not necessary.  ...  We can use this information to separate the input and output parts via the argument splitting transformation, and then make all arguments output via the magic set transformation.  ... 
doi:10.1007/3-540-57818-8_41 fatcat:ydzh5eintngnlcpx2ojo4jd42y

Multi-Task Learning with Generative Adversarial Training for Multi-Passage Machine Reading Comprehension

Qiyu Ren, Xiang Cheng, Sen Su
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
the answer candidate set and its representation.  ...  MG-MRC adopts the extract-then-select framework, where an extractor is first used to predict answer candidates, then a selector is used to choose the final answer.  ...  This work was supported by the National Natural Science Foundation of China (Grant No. 61872045), the Foundation for Innovative Research Groups of the National Natural Science Foundation of China (Grant  ... 
doi:10.1609/aaai.v34i05.6396 fatcat:7w6xlgv6zfeetax5j3xdth3lwi

Generative Context Pair Selection for Multi-hop Question Answering [article]

Dheeru Dua, Cicero Nogueira dos Santos, Patrick Ng, Ben Athiwaratkun, Bing Xiang, Matt Gardner, Sameer Singh
2021 arXiv   pre-print
While being comparable to the state-of-the-art answering performance, our proposed generative passage selection model has a better performance (4.9% higher than baseline) on adversarial held-out set which  ...  In this work, we propose a generative context selection model for multi-hop question answering that reasons about how the given question could have been generated given a context pair.  ...  Our generative passage selector shows an improvement of 4.9% in Top-1 accuracy as compared to discriminatively trained passage selector on the adversarial dev set.  ... 
arXiv:2104.08744v1 fatcat:mrm6ucfjerhsvek4a4x5ih77qa

Optimizing scoring functions and indexes for proximity search in type-annotated corpora

Soumen Chakrabarti, Kriti Puniyani, Sujatha Das
2006 Proceedings of the 15th international conference on World Wide Web - WWW '06  
Nearness is defined as a flexible, trainable parameterized aggregation function of the selectors, their frequency in the corpus, and their distance from the candidate answer.  ...  First, we propose a new algorithm that estimates a scoring function from past logs of queries and answer spans.  ...  , i.e., strongly activated by selectors.  ... 
doi:10.1145/1135777.1135882 dblp:conf/www/ChakrabartiPD06 fatcat:fzgzxrakmffpni67r56plfn5aq

Abstract Preference Frameworks — a Unifying Perspective on Separability and Strong Equivalence

Wolfgang Faber, Mirosław Truszczyński, Stefan Woltran
2013 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
We introduce abstract preference frameworks to study general properties common across a variety of preference formalisms.  ...  We identify abstract postulates on preference frameworks, satisfied by most of the currently studied preference formalisms, that lead to characterizations of both properties of interest.  ...  The second author was supported by the NSF grant IIS-0913459.  ... 
doi:10.1609/aaai.v27i1.8593 fatcat:vhegqbdjwfhj3filxh2a5ib3fq

Estimate and Replace: A Novel Approach to Integrating Deep Neural Networks with Existing Applications [article]

Guy Hadash, Einat Kermany, Boaz Carmeli, Ofer Lavi, George Kour and Alon Jacovi
2018 arXiv   pre-print
Using this approach, we estimate each application's functionality with an estimator, which is implemented as a deep neural network (DNN).  ...  At inference time, we replace each estimator with its existing application counterpart and let the base network solve the task by interacting with the existing application.  ...  One direction is to dynamically learn the interface signature of a given functionality. Another is to learn to orchestrate a set of interfaces to solve a complex task.  ... 
arXiv:1804.09028v1 fatcat:6zuknbntpvbafabswp5scsjojm

Priority based computation: "An Initial study result of paradigm shift on real time computation"

Sukemi Sukemi, Riyanto Riyanto
2017 Computer Engineering and Applications Journal  
Each processor will be separated by selector/arbiter priority of a task.  ...  This research is purposed to increase computer function into a time driven to support real time system.  ...  Interestingly, until now, there is no mechanism that can play the selector function represented by the hardware or software assigned to it.  ... 
doi:10.18495/comengapp.v6i1.198 fatcat:ykgovxhtyfeytkcq72cluhydl4

Analysis of English Multitext Reading Comprehension Model Based on Deep Belief Neural Network

Qiaohui Tang, Syed Hassan Ahmed
2021 Computational Intelligence and Neuroscience  
Firstly, the paragraph selector in the multitext reading comprehension model is constructed.  ...  Secondly, the text reader is designed, and the deep belief neural network is introduced to predict the question answering probability.  ...  Similarly, the loss value of the multicategory paragraph selector in the test set is the minimum loss value in the test set.  ... 
doi:10.1155/2021/5100809 pmid:34567102 pmcid:PMC8457992 fatcat:ry34jyzbnfgxfa7566rxfwc2ze

Infer XPath [article]

Michał J. Gajda, Hai Nguyen Quang, Do Ngoc Khanh, Vuong Hai Thanh
2020 arXiv   pre-print
We propose reformulation of discovery of data structure within a web page as relations between sets of document nodes.  ...  We start by reformulating web page analysis as finding expressions in extension of XPath. Then we propose to automatically discover these XPath expressions with InferXPath meta-language.  ...  Acknowledgments Implementation of this proposal is a joint work in progress by Migamake and KikaiTech to automate laborious scraping tasks. Bibliography  ... 
arXiv:2011.03538v1 fatcat:xfxderjclfhyhig76fkefsu7na

Learning a Natural Language Interface with Neural Programmer [article]

Arvind Neelakantan, Quoc V. Le, Martin Abadi, Andrew McCallum, Dario Amodei
2017 arXiv   pre-print
We enhance the objective function of Neural Programmer, a neural network with built-in discrete operations, and apply it on WikiTableQuestions, a natural language question-answering dataset.  ...  The task is often approached by mapping natural language queries to logical forms or programs that provide the desired response when executed on the database.  ...  Arvind Neelakantan is supported by a Google PhD fellowship in machine learning.  ... 
arXiv:1611.08945v4 fatcat:th73j7sqtzgfrl6ogf72dj2lc4

Identifying an Honest EXP^NP Oracle Among Many

Shuichi Hirahara, Marc Herbstritt
2015 Computational Complexity Conference  
We provide a general framework to remove short advice by formulating the following computational task for a function f : given two oracles at least one of which is honest (i.e. correctly computes f on  ...  We characterize the languages for which short advice can be removed by the notion of selector: a paddable language has a selector if and only if short advice of a probabilistic machine that accepts the  ...  Every PSPACE-complete language has a deterministic selector. More generally, any downward self-reducible language has a deterministic selector. 2.  ... 
doi:10.4230/lipics.ccc.2015.244 dblp:conf/coco/Hirahara15 fatcat:sa7egfeuf5hnzojijde7ysqhwq

A Deep Reinforcement Learning Based Multi-Step Coarse to Fine Question Answering (MSCQA) System

Yu Wang, Hongxia Jin
2019 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
In this paper, we present a multi-step coarse to fine question answering (MSCQA) system which can efficiently processes documents with different lengths by choosing appropriate actions.  ...  The system is designed using an actor-critic based deep reinforcement learning model to achieve multistep question answering.  ...  In this work, the author uses the sentence selector as a coarse selector, and generate the answer based on the selected sentences.  ... 
doi:10.1609/aaai.v33i01.33017224 fatcat:ztgzufto7jbpjoz2a4tfirbstq

Identifying an Honest EXP^ NP Oracle Among Many [article]

Shuichi Hirahara
2015 arXiv   pre-print
We provide a general framework to remove short advice by formulating the following computational task for a function f: given two oracles at least one of which is honest (i.e. correctly computes f on all  ...  We characterize the languages for which short advice can be removed by the notion of selector: a paddable language has a selector if and only if short advice of a probabilistic machine that accepts the  ...  Indeed, the YES player wins by setting A 1 = L; similarly, if x ∈ L, then the NO player wins by setting A 0 = L.  ... 
arXiv:1502.07258v2 fatcat:5c53cfywdvffhoosqasu4yxvfy

Disjunctive Programs with Set Constraints [chapter]

Victor W. Marek, Jeffrey B. Remmel
2012 Lecture Notes in Computer Science  
We introduce new class of models called selector stable models which represent all models which can be computed by an analogue the Gelfond-Lifschitz transform.  ...  head and arbitrary monotone and antimonotone set constraints in their body.  ...  By that same argument the selector functions generalize the approach of [13] .  ... 
doi:10.1007/978-3-642-30743-0_32 fatcat:mk7ng3yjsrhejfzf7qbdhnpexa
« Previous Showing results 1 — 15 out of 17,883 results