The Experts below are selected from a list of 360 Experts worldwide ranked by ideXlab platform
Timothy Lillicrap - One of the best experts on this subject based on the ideXlab platform.
-
a simple neural network module for relational reasoning
arXiv: Computation and Language, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamic physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Our work shows how a deep learning architecture equipped with an RN module can implicitly discover and learn to reason about entities and their relations.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.
Adam Santoro - One of the best experts on this subject based on the ideXlab platform.
-
a simple neural network module for relational reasoning
arXiv: Computation and Language, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamic physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Our work shows how a deep learning architecture equipped with an RN module can implicitly discover and learn to reason about entities and their relations.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.
Mateusz Malinowski - One of the best experts on this subject based on the ideXlab platform.
-
a simple neural network module for relational reasoning
arXiv: Computation and Language, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamic physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Our work shows how a deep learning architecture equipped with an RN module can implicitly discover and learn to reason about entities and their relations.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.
Peter William Battaglia - One of the best experts on this subject based on the ideXlab platform.
-
a simple neural network module for relational reasoning
arXiv: Computation and Language, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamic physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Our work shows how a deep learning architecture equipped with an RN module can implicitly discover and learn to reason about entities and their relations.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.
Razvan Pascanu - One of the best experts on this subject based on the ideXlab platform.
-
a simple neural network module for relational reasoning
arXiv: Computation and Language, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamic physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Our work shows how a deep learning architecture equipped with an RN module can implicitly discover and learn to reason about entities and their relations.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.
-
a simple neural network module for relational reasoning
Neural Information Processing Systems, 2017Co-Authors: Adam Santoro, Peter William Battaglia, David Raposo, Razvan Pascanu, Mateusz Malinowski, David G. T. Barrett, Timothy LillicrapAbstract:Relational reasoning is a central component of generally Intelligent Behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation Networks (RNs) as a simple plug-and-play module to solve problems that fundamentally hinge on relational reasoning. We tested RN-augmented networks on three tasks: visual question answering using a challenging dataset called CLEVR, on which we achieve state-of-the-art, super-human performance; text-based question answering using the bAbI suite of tasks; and complex reasoning about dynamical physical systems. Then, using a curated dataset called Sort-of-CLEVR we show that powerful convolutional networks do not have a general capacity to solve relational questions, but can gain this capacity when augmented with RNs. Thus, by simply augmenting convolutions, LSTMs, and MLPs with RNs, we can remove computational burden from network components that are not well-suited to handle relational reasoning, reduce overall network complexity, and gain a general ability to reason about the relations between entities and their properties.