Zobrazeno 1 - 10
of 835
pro vyhledávání: '"McAllester, David"'
Autor:
McAllester, David
This paper gives direct derivations of the differential equations and likelihood formulas of diffusion models assuming only knowledge of Gaussian distributions. A VAE analysis derives both forward and backward stochastic differential equations (SDEs)
Externí odkaz:
http://arxiv.org/abs/2301.11108
Autor:
Levine, Victoria Lindsay, Sercombe, Laurel
Publikováno v:
The Grove Dictionary of American Music, 2 ed., 2013.
Externí odkaz:
http://www.oxfordreference.com/view/10.1093/acref/9780195314281.001.0001/acref-9780195314281-e-5150
Autor:
McAllester, David
This paper develops a version of dependent type theory in which isomorphism is handled through a direct generalization of the 1939 definitions of Bourbaki. More specifically we generalize the Bourbaki definition of structure from simple type signatur
Externí odkaz:
http://arxiv.org/abs/2104.08958
We study image segmentation from an information-theoretic perspective, proposing a novel adversarial method that performs unsupervised segmentation by partitioning images into maximally independent sets. More specifically, we group image pixels into
Externí odkaz:
http://arxiv.org/abs/2012.07287
Autor:
Wang, Hai, McAllester, David
Here we experiment with the use of information retrieval as an augmentation for pre-trained language models. The text corpus used in information retrieval can be viewed as form of episodic memory which grows over time. By augmenting GPT 2.0 with info
Externí odkaz:
http://arxiv.org/abs/2007.01528
Autor:
McAllester, David
AlphaZero learns to play go, chess and shogi at a superhuman level through self play given only the rules of the game. This raises the question of whether a similar thing could be done for mathematics -- a MathZero. MathZero would require a formal fo
Externí odkaz:
http://arxiv.org/abs/2005.05512
Autor:
McAllester, David
Isomorphism is central to the structure of mathematics and has been formalized in various ways within dependent type theory. All previous treatments have done this by replacing quantification over sets with quantification over groupoids of some form
Externí odkaz:
http://arxiv.org/abs/1912.02885
From a simplified analysis of adaptive methods, we derive AvaGrad, a new optimizer which outperforms SGD on vision tasks when its adaptability is properly tuned. We observe that the power of our method is partially explained by a decoupling of learni
Externí odkaz:
http://arxiv.org/abs/1912.01823
Remarkable success has been achieved in the last few years on some limited machine reading comprehension (MRC) tasks. However, it is still difficult to interpret the predictions of existing MRC models. In this paper, we focus on extracting evidence s
Externí odkaz:
http://arxiv.org/abs/1902.08852
Autor:
Rice, Timothy
Publikováno v:
Yearbook for Traditional Music, 1986 Jan 01. 18, 181-183.
Externí odkaz:
https://www.jstor.org/stable/768532