Neuro-Symbolic Concept Learner (NS-CL) - Revision history

BPeat at 18:43, 16 September 2023

2023-09-16T18:43:47Z

BPeat at 15:48, 9 May 2023

2023-05-09T15:48:36Z

BPeat at 14:43, 9 May 2023

2023-05-09T14:43:58Z

BPeat at 03:17, 4 May 2020

2020-05-04T03:17:42Z

BPeat at 02:16, 9 April 2019

2019-04-09T02:16:31Z

BPeat at 02:14, 9 April 2019

2019-04-09T02:14:44Z

BPeat at 02:13, 9 April 2019

2019-04-09T02:13:10Z

BPeat at 01:52, 9 April 2019

2019-04-09T01:52:03Z

BPeat at 01:46, 9 April 2019

2019-04-09T01:46:44Z

BPeat at 01:37, 9 April 2019

2019-04-09T01:37:00Z

@@ Line 12: / Line 12: @@
 * [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child | Will Knight - MIT Technology Review]
-a model that learns visual concepts, words, and semantic parsing of sentences without explicit supervision on any of them; instead, our model learns by simply looking at images and reading paired questions and answers. Our model builds an object-based scene representation and translates sentences into executable, symbolic programs. To bridge the learning of two modules, we use a neuro-symbolic reasoning module that executes these programs on the latent scene representation. Analogical to human concept learning, the perception module learns visual concepts based on the language description of the object being referred to. Meanwhile, the learned visual concepts facilitate learning new words and parsing new sentences. We use curriculum learning to guide the searching over the large compositional space of images and language. Extensive experiments demonstrate the accuracy and efficiency of our model on learning visual concepts, word representations, and semantic parsing of sentences. Further, our method allows easy generalization to new object attributes, compositions, language concepts, scenes and questions, and even new program domains. It also empowers applications including visual question answering and bidirectional image-text retrieval. [http://deepmind.com/research/publications/neuro-symbolic-concept-learner-interpreting-scenes-words-and-sentences-natural-supervision/ The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision | J. Mao, C. Gan, P. Kohli, J. Tenenbaum, and J. Wu]
+a model that learns visual concepts, words, and semantic parsing of sentences without explicit supervision on any of them; instead, our model learns by simply looking at images and reading paired questions and answers. Our model builds an object-based scene representation and translates sentences into executable, symbolic programs. To bridge the learning of two modules, we use a neuro-symbolic reasoning module that executes these programs on the [[latent]] scene representation. Analogical to human concept learning, the perception module learns visual concepts based on the language description of the object being referred to. Meanwhile, the learned visual concepts facilitate learning new words and parsing new sentences. We use curriculum learning to guide the searching over the large compositional space of images and language. Extensive experiments demonstrate the accuracy and efficiency of our model on learning visual concepts, word representations, and semantic parsing of sentences. Further, our method allows easy generalization to new object attributes, compositions, language concepts, scenes and questions, and even new program domains. It also empowers applications including visual question answering and bidirectional image-text retrieval. [http://deepmind.com/research/publications/neuro-symbolic-concept-learner-interpreting-scenes-words-and-sentences-natural-supervision/ The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision | J. Mao, C. Gan, P. Kohli, J. Tenenbaum, and J. Wu]
 http://cdn.technologyreview.com/i/images/tenenbaumpaper.jpg

@@ Line 8: / Line 8: @@
 [http://www.google.com/search?q=Neuro+Symbolic+Concept+Learner+NS+CL+machine+learning+ML+artificial+intelligence ...Google search]
-* [[Neural-Symbolic]] ... [[Symbolic Artificial Intelligence]]
+* [[Neuro-Symbolic]] ... [[Symbolic Artificial Intelligence]]
 * [http://pathmind.com/wiki/symbolic-reasoning Symbolic Reasoning (Symbolic AI) and Machine Learning | Chris Nicholson - A.I. Wiki pathmind]
 * [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child | Will Knight - MIT Technology Review]

@@ Line 8: / Line 8: @@
 [http://www.google.com/search?q=Neuro+Symbolic+Concept+Learner+NS+CL+machine+learning+ML+artificial+intelligence ...Google search]
-* [[Neuro-Symbolic]]
+* [[Neural-Symbolic]] ... [[Symbolic Artificial Intelligence]]
 * [http://pathmind.com/wiki/symbolic-reasoning Symbolic Reasoning (Symbolic AI) and Machine Learning | Chris Nicholson - A.I. Wiki pathmind]
 * [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child | Will Knight - MIT Technology Review]

← Older revision		Revision as of 03:17, 4 May 2020
Line 9:		Line 9:

	* [[Neuro-Symbolic]]		* [[Neuro-Symbolic]]
		+	* [http://pathmind.com/wiki/symbolic-reasoning Symbolic Reasoning (Symbolic AI) and Machine Learning \| Chris Nicholson - A.I. Wiki pathmind]
	* [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child \| Will Knight - MIT Technology Review]		* [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child \| Will Knight - MIT Technology Review]

← Older revision		Revision as of 02:16, 9 April 2019
Line 12:		Line 12:

	a model that learns visual concepts, words, and semantic parsing of sentences without explicit supervision on any of them; instead, our model learns by simply looking at images and reading paired questions and answers. Our model builds an object-based scene representation and translates sentences into executable, symbolic programs. To bridge the learning of two modules, we use a neuro-symbolic reasoning module that executes these programs on the latent scene representation. Analogical to human concept learning, the perception module learns visual concepts based on the language description of the object being referred to. Meanwhile, the learned visual concepts facilitate learning new words and parsing new sentences. We use curriculum learning to guide the searching over the large compositional space of images and language. Extensive experiments demonstrate the accuracy and efficiency of our model on learning visual concepts, word representations, and semantic parsing of sentences. Further, our method allows easy generalization to new object attributes, compositions, language concepts, scenes and questions, and even new program domains. It also empowers applications including visual question answering and bidirectional image-text retrieval. [http://deepmind.com/research/publications/neuro-symbolic-concept-learner-interpreting-scenes-words-and-sentences-natural-supervision/ The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision \| J. Mao, C. Gan, P. Kohli, J. Tenenbaum, and J. Wu]		a model that learns visual concepts, words, and semantic parsing of sentences without explicit supervision on any of them; instead, our model learns by simply looking at images and reading paired questions and answers. Our model builds an object-based scene representation and translates sentences into executable, symbolic programs. To bridge the learning of two modules, we use a neuro-symbolic reasoning module that executes these programs on the latent scene representation. Analogical to human concept learning, the perception module learns visual concepts based on the language description of the object being referred to. Meanwhile, the learned visual concepts facilitate learning new words and parsing new sentences. We use curriculum learning to guide the searching over the large compositional space of images and language. Extensive experiments demonstrate the accuracy and efficiency of our model on learning visual concepts, word representations, and semantic parsing of sentences. Further, our method allows easy generalization to new object attributes, compositions, language concepts, scenes and questions, and even new program domains. It also empowers applications including visual question answering and bidirectional image-text retrieval. [http://deepmind.com/research/publications/neuro-symbolic-concept-learner-interpreting-scenes-words-and-sentences-natural-supervision/ The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision \| J. Mao, C. Gan, P. Kohli, J. Tenenbaum, and J. Wu]
		+
		+	http://cdn.technologyreview.com/i/images/tenenbaumpaper.jpg

← Older revision		Revision as of 02:14, 9 April 2019
Line 12:		Line 12:

	a model that learns visual concepts, words, and semantic parsing of sentences without explicit supervision on any of them; instead, our model learns by simply looking at images and reading paired questions and answers. Our model builds an object-based scene representation and translates sentences into executable, symbolic programs. To bridge the learning of two modules, we use a neuro-symbolic reasoning module that executes these programs on the latent scene representation. Analogical to human concept learning, the perception module learns visual concepts based on the language description of the object being referred to. Meanwhile, the learned visual concepts facilitate learning new words and parsing new sentences. We use curriculum learning to guide the searching over the large compositional space of images and language. Extensive experiments demonstrate the accuracy and efficiency of our model on learning visual concepts, word representations, and semantic parsing of sentences. Further, our method allows easy generalization to new object attributes, compositions, language concepts, scenes and questions, and even new program domains. It also empowers applications including visual question answering and bidirectional image-text retrieval. [http://deepmind.com/research/publications/neuro-symbolic-concept-learner-interpreting-scenes-words-and-sentences-natural-supervision/ The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision \| J. Mao, C. Gan, P. Kohli, J. Tenenbaum, and J. Wu]		a model that learns visual concepts, words, and semantic parsing of sentences without explicit supervision on any of them; instead, our model learns by simply looking at images and reading paired questions and answers. Our model builds an object-based scene representation and translates sentences into executable, symbolic programs. To bridge the learning of two modules, we use a neuro-symbolic reasoning module that executes these programs on the latent scene representation. Analogical to human concept learning, the perception module learns visual concepts based on the language description of the object being referred to. Meanwhile, the learned visual concepts facilitate learning new words and parsing new sentences. We use curriculum learning to guide the searching over the large compositional space of images and language. Extensive experiments demonstrate the accuracy and efficiency of our model on learning visual concepts, word representations, and semantic parsing of sentences. Further, our method allows easy generalization to new object attributes, compositions, language concepts, scenes and questions, and even new program domains. It also empowers applications including visual question answering and bidirectional image-text retrieval. [http://deepmind.com/research/publications/neuro-symbolic-concept-learner-interpreting-scenes-words-and-sentences-natural-supervision/ The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision \| J. Mao, C. Gan, P. Kohli, J. Tenenbaum, and J. Wu]
−
−	~~<youtube>x_QjJry0hTc</youtube>~~

@@ Line 11: / Line 11: @@
 * [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child | Will Knight - MIT Technology Review]
-a model that learns visual concepts, words, and semantic parsing of sentences without explicit supervision on any of them; instead, our model learns by simply looking at images and reading paired questions and answers. Our model builds an object-based scene representation and translates sentences into executable, symbolic programs. To bridge the learning of two modules, we use a neuro-symbolic reasoning module that executes these programs on the latent scene representation. Analogical to human concept learning, the perception module learns visual concepts based on the language description of the object being referred to. Meanwhile, the learned visual concepts facilitate learning new words and parsing new sentences. We use curriculum learning to guide the searching over the large compositional space of images and language. Extensive experiments demonstrate the accuracy and efficiency of our model on learning visual concepts, word representations, and semantic parsing of sentences. Further, our method allows easy generalization to new object attributes, compositions, language concepts, scenes and questions, and even new program domains. It also empowers applications including visual question answering and bidirectional image-text retrieval. [http://openreview.net/forum?id=rJgMlhRctm The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision | J. Mao, C. Gan, P. Kohli, J. Tenenbaum, and J. Wu]
+a model that learns visual concepts, words, and semantic parsing of sentences without explicit supervision on any of them; instead, our model learns by simply looking at images and reading paired questions and answers. Our model builds an object-based scene representation and translates sentences into executable, symbolic programs. To bridge the learning of two modules, we use a neuro-symbolic reasoning module that executes these programs on the latent scene representation. Analogical to human concept learning, the perception module learns visual concepts based on the language description of the object being referred to. Meanwhile, the learned visual concepts facilitate learning new words and parsing new sentences. We use curriculum learning to guide the searching over the large compositional space of images and language. Extensive experiments demonstrate the accuracy and efficiency of our model on learning visual concepts, word representations, and semantic parsing of sentences. Further, our method allows easy generalization to new object attributes, compositions, language concepts, scenes and questions, and even new program domains. It also empowers applications including visual question answering and bidirectional image-text retrieval. [http://deepmind.com/research/publications/neuro-symbolic-concept-learner-interpreting-scenes-words-and-sentences-natural-supervision/ The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision | J. Mao, C. Gan, P. Kohli, J. Tenenbaum, and J. Wu]
 <youtube>x_QjJry0hTc</youtube>

← Older revision		Revision as of 01:52, 9 April 2019
Line 8:		Line 8:
	[http://www.google.com/search?q=Neuro+Symbolic+Concept+Learner+NS+CL+machine+learning+ML+artificial+intelligence ...Google search]		[http://www.google.com/search?q=Neuro+Symbolic+Concept+Learner+NS+CL+machine+learning+ML+artificial+intelligence ...Google search]

		+	* [[Neuro-Symbolic]]
	* [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child \| Will Knight - MIT Technology Review]		* [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child \| Will Knight - MIT Technology Review]

← Older revision		Revision as of 01:37, 9 April 2019
Line 9:		Line 9:

	* [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child \| Will Knight - MIT Technology Review]		* [http://www.technologyreview.com/s/613270/two-rival-ai-approaches-combine-to-let-machines-learn-about-the-world-like-a-child/ Two rival AI approaches combine to let machines learn about the world like a child \| Will Knight - MIT Technology Review]
		+	* [http://openreview.net/forum?id=rJgMlhRctm The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision

	<youtube>x_QjJry0hTc</youtube>		<youtube>x_QjJry0hTc</youtube>