The Experts below are selected from a list of 49572 Experts worldwide ranked by ideXlab platform
A C M Fong - One of the best experts on this subject based on the ideXlab platform.
-
cs nlp team at semeval 2020 Task 4 evaluation of state of the art nlp deep learning architectures on commonsense Reasoning Task
International Conference on Computational Linguistics, 2020Co-Authors: Sirwe Saeedi, Aliakbar Panahi, Seyran Saeedi, A C M FongAbstract:In this paper, we investigate a commonsense inference Task that unifies natural language understanding and commonsense Reasoning. We describe our attempt at SemEval-2020 Task 4 competition: Commonsense Validation and Explanation (ComVE) challenge. We discuss several state-of-the-art deep learning architectures for this challenge. Our system uses prepared labeled textual datasets that were manually curated for three different natural language inference subTasks. The goal of the first subTask is to test whether a model can distinguish between natural language statements that make sense and those that do not make sense. We compare the performance of several language models and fine-tuned classifiers. Then, we propose a method inspired by question/answering Tasks to treat a classification problem as a multiple choice question Task to boost the performance of our experimental results (96.06%), which is significantly better than the baseline. For the second subTask, which is to select the reason why a statement does not make sense, we stand within the first six teams (93.7%) among 27 participants with very competitive results. Our result for last subTask of generating reason against the nonsense statement shows many potentials for future researches as we applied the most powerful generative model of language (GPT-2) with 6.1732 BLEU score among first four teams. .
Katarzyna Blinowska - One of the best experts on this subject based on the ideXlab platform.
-
p27 t common pattern of information transfer in the brain for different modalities during a Reasoning Task
Clinical Neurophysiology, 2019Co-Authors: Maciej Kamiński, Aneta Brzezicka, Jan Kamiński, Katarzyna BlinowskaAbstract:Background Information processing in the brain involves the synchronization of brain structures realized by the transmission of rhythmic activity. The aim of our work was to find connectivity patterns between brain structures during the same mental activity performed during Tasks with different stimuli modality. Material and methods The Task: the paradigm of linear syllogism was performed with visual (12 subjects) and auditory stimuli (29 subjects). The EEG activity transmission was analyzed by means of the short-time Directed Transfer Function (sDTF), a frequency dependent estimator of directed coupling between time series based on the Granger causality principle. Results and conclusions The obtained patterns showed similar coupling for both modalities involving EEG propagation mainly between the frontal and posterior areas which communicated intermittently. Our results indicate that the fronto-parietal network operated mainly in the theta frequency range as well as in other frequency ranges, especially gamma. By means of the assortative mixing approach, the strengths of coupling between the regions of interests (ROIs) were determined. For both modalities, the results showed stronger coupling within the ROIs than between them in agreement with the theories considering information transfer efficiency and metabolic energy savings. The patterns differences were minor and concerned stronger propagation from the posterior electrodes towards the frontal ones during the visual Task and from the temporal sites to the frontal ones during the auditory Task, which can be explained as bottom-up communication from specific sensory sites. Our results support a modality-free process of information retrieval and integration in a Reasoning Task.
-
Information Transfer During a Transitive Reasoning Task
Brain Topography, 2011Co-Authors: Aneta Brzezicka, Maciej Kamiński, Jan Kamiński, Katarzyna BlinowskaAbstract:For about two decades now, the localization of the brain regions involved in Reasoning processes is being investigated through fMRI studies, and it is known that for a transitive form of Reasoning the frontal and parietal regions are most active. In contrast, less is known about the information exchange during the performance of such complex Tasks. In this study, the propagation of brain activity during a transitive Reasoning Task was investigated and compared to the propagation during a simple memory Task. We studied EEG transmission patterns obtained for physiological indicators of brain activity and determined whether there are frequency bands specifically related to this type of cognitive operations. The analysis was performed by means of the directed transfer function. The transmission patterns were determined in the theta, alpha and gamma bands. The results show stronger transmissions in theta and alpha bands from frontal to parietal as well as within frontal regions in Reasoning trials comparing to memory trials. The increase in theta and alpha transmissions was accompanied by flows in gamma band from right posterior to left posterior and anterior sites. These results are consistent with previous neuroimaging (fMRI) data concerning fronto-parietal regions involvement in Reasoning and working memory processes and also provide new evidence for the executive role of frontal theta waves in organizing the cognition.
Sirwe Saeedi - One of the best experts on this subject based on the ideXlab platform.
-
cs nlp team at semeval 2020 Task 4 evaluation of state of the art nlp deep learning architectures on commonsense Reasoning Task
International Conference on Computational Linguistics, 2020Co-Authors: Sirwe Saeedi, Aliakbar Panahi, Seyran Saeedi, A C M FongAbstract:In this paper, we investigate a commonsense inference Task that unifies natural language understanding and commonsense Reasoning. We describe our attempt at SemEval-2020 Task 4 competition: Commonsense Validation and Explanation (ComVE) challenge. We discuss several state-of-the-art deep learning architectures for this challenge. Our system uses prepared labeled textual datasets that were manually curated for three different natural language inference subTasks. The goal of the first subTask is to test whether a model can distinguish between natural language statements that make sense and those that do not make sense. We compare the performance of several language models and fine-tuned classifiers. Then, we propose a method inspired by question/answering Tasks to treat a classification problem as a multiple choice question Task to boost the performance of our experimental results (96.06%), which is significantly better than the baseline. For the second subTask, which is to select the reason why a statement does not make sense, we stand within the first six teams (93.7%) among 27 participants with very competitive results. Our result for last subTask of generating reason against the nonsense statement shows many potentials for future researches as we applied the most powerful generative model of language (GPT-2) with 6.1732 BLEU score among first four teams. .
David C. Reutens - One of the best experts on this subject based on the ideXlab platform.
-
Age-related differences in structural and functional prefrontal networks during a logical Reasoning Task
Brain Imaging and Behavior, 2020Co-Authors: Maryam Ziaei, Mohammad Reza Bonyadi, David C. ReutensAbstract:In logical Reasoning, difficulties in inhibition of currently-held beliefs may lead to unwarranted conclusions, known as belief bias. Aging is associated with difficulties in inhibitory control, which may lead to deficits in inhibition of currently-held beliefs. No study to date, however, has investigated the underlying neural substrates of age-related differences in logical Reasoning and the impact of belief load. The aim of the present study was to delineate age differences in brain activity during a syllogistic logical Reasoning Task while the believability load of logical inferences was manipulated. Twenty-nine, healthy, younger and thirty, healthy, older adults (males and females) completed a functional magnetic resonance imaging experiment in which they were asked to determine the logical validity of conclusions. Unlike younger adults, older adults engaged a large-scale network including anterior cingulate cortex and inferior frontal gyrus during conclusion stage. Our functional connectivity results suggest that while older adults engaged the anterior cingulate network to overcome their intuitive responses for believable inferences, the inferior frontal gyrus network contributed to higher control over responses during both believable and unbelievable conditions. Our functional results were further supported by structure-function-behavior analyses indicating the importance of cingulum bundle and uncinate fasciculus integrity in rejection of believable statements. These novel findings lend evidence for age-related differences in belief bias, with potentially important implications for decision making where currently-held beliefs and given assumptions are in conflict.
-
age related differences in structural and functional prefrontal networks during a logical Reasoning Task
bioRxiv, 2019Co-Authors: Maryam Ziaei, Mohammad Reza Bonyadi, David C. ReutensAbstract:In logical Reasoning, difficulties in inhibition of currently-held beliefs may lead to unwarranted conclusions, known as belief bias. Aging is associated with difficulties in inhibitory control, which may lead to deficits in inhibition of currently-held beliefs. But, no study, to date, has investigated the underlying neural substrates of age-related differences in logical Reasoning and the impact of belief loads. The aim of the present study was to delineate age differences in brain activity during a syllogistic logical Reasoning Task while the believability load of the logical inferences was manipulated. Twenty-nine, healthy, younger and thirty, healthy, older adults (males and females) completed a functional magnetic resonance imaging experiment in which they were asked to determine the logical validity of conclusions. Unlike younger adults, older adults engaged a large-scale network including anterior cingulate cortex (ACC) and inferior frontal gyrus (IFG) during conclusion stage. Our functional connectivity results suggest that while older adults engaged the ACC network to overcome their intuitive responses for believable inferences, the IFG network contributed to higher control over responses during both believable and unbelievable conditions. This result was further supported by mediation analysis indicating the role of Uncinate Fasciculus tract as a mediator for a relationship between age and rejection of believable statements. These novel findings lend evidence for age-related differences in belief bias, with potentially important implications for socially-relevant decision making where currently-held beliefs and given assumptions are in conflict.
Raymond J Mooney - One of the best experts on this subject based on the ideXlab platform.
-
plan recognition using statistical relational models
Plan Activity and Intent Recognition#R##N#Theory and Practice, 2014Co-Authors: Sindhu Raghavan, Parag Singla, Raymond J MooneyAbstract:Plan recognition is the Task of predicting an agent’s top-le vel plans based on its observed actions. It is an abductive Reasoning Task that involves inferring plans that best explain observed actions. Most existing approaches to plan recognition and other abductive Reasoning Tasks either use first-order l ogic (or subsets of it) or probabilistic graphical models. While the former cannot handle uncertainty in the data, the latter cannot handle structured representati ons. To overcome these limitations, we explore the application of statistical rel ational models that combine the strengths of both first-order logic and probabilistic gr aphical models to plan recognition. Specifically, we introduce two new approaches to abductive plan recognition using Bayesian Logic Programs (BLPs) and Markov Logic Networks (MLNs). Neither of these formalisms is suited for abductive Reasoning because of the deductive nature of the underlying logical inference. I n this work, we propose approaches to adapt both these formalisms for abductive plan recognition. We present an extensive evaluation of our approaches on three benchmark datasets on plan recognition, comparing them with existing state-of-the-art methods.
-
abductive plan recognition by extending bayesian logic programs
European conference on Machine Learning, 2011Co-Authors: Sindhu Raghavan, Raymond J MooneyAbstract:Plan recognition is the Task of predicting an agent's top-level plans based on its observed actions. It is an abductive Reasoning Task that involves inferring cause from effect. Most existing approaches to plan recognition use either first-order logic or probabilistic graphical models. While the former cannot handle uncertainty, the latter cannot handle structured representations. In order to overcome these limitations, we develop an approach to plan recognition using Bayesian Logic Programs (BLPs), which combine first-order logic and Bayesian networks. Since BLPs employ logical deduction to construct the networks, they cannot be used effectively for plan recognition. Therefore, we extend BLPs to use logical abduction to construct Bayesian networks and call the resulting model Bayesian Abductive Logic Programs (BALPs). We learn the parameters in BALPs using the Expectation Maximization algorithm adapted for BLPs. Finally, we present an experimental evaluation of BALPs on three benchmark data sets and compare its performance with the state-of-the-art for plan recognition.
