14,138 Hits in 9.0 sec

On the Learnability of Rich Function Classes

Joel Ratsaby, Vitaly Maiorov
1999 Journal of computer and system sciences (Print)  
The probably approximately correct (PAC) model of learning and its extension to real-valued function classes sets a rigorous framework based upon which the complexity of learning a target from a function  ...  partial information.  ...  CONCLUSION We introduced a theoretical framework which extends the PAC model of learning to a scenario where a learner has general partial information about the target function, in addition to randomly  ... 
doi:10.1006/jcss.1998.1604 fatcat:apgtnu26zjeh3kgblgnpxj5xjm

Learning from Partial Observations

Loizos Michael
2007 International Joint Conference on Artificial Intelligence  
We present a general machine learning framework for modelling the phenomenon of missing information in data. We propose a masking process model to capture the stochastic nature of information loss.  ...  We extend the Probably Approximately Correct semantics to the case of learning from partial observations with arbitrarily hidden attributes.  ...  Acknowledgments The author is grateful to Leslie Valiant for his advice, and for valuable suggestions and remarks on this research.  ... 
dblp:conf/ijcai/Michael07 fatcat:rhqcsn2ufrgabhimggm3ousuya

Page 2072 of Mathematical Reviews Vol. , Issue 99c [page]

1991 Mathematical Reviews  
The error rates are calculated, which leads to a quantitative notion of the value of partial information for the paradigm of learning from examples.”  ...  A model for approximate testing of concepts is introduced by analogy to the PAC model of learning. A hierarchy of four levels of testability is discussed and illustrated by examples from geometry.  ... 

Foreseeing the Benefits of Incidental Supervision [article]

Hangfeng He, Mingyuan Zhang, Qiang Ning, Dan Roth
2021 arXiv   pre-print
We propose a unified PAC-Bayesian motivated informativeness measure, PABI, that characterizes the uncertainty reduction provided by incidental supervision signals.  ...  Experiments on named entity recognition (NER) and question answering (QA) show that PABI's predictions correlate well with learning performance, providing a promising way to determine, ahead of learning  ...  The views expressed are those of the authors and do not reflect the official policy or position of the Department of Defense or the U.S. Government.  ... 
arXiv:2006.05500v2 fatcat:e4yj4zu3ozfylibouota64mdhi

Learnability of Influence in Networks

Harikrishna Narasimhan, David C. Parkes, Yaron Singer
2015 Neural Information Processing Systems  
Our results for the LT model are based on interesting connections with neural networks; those for the IC model are based an interpretation of the influence function as an expectation over random draw of  ...  We show PAC learnability of influence functions for three common influence models, namely, the Linear Threshold (LT), Independent Cascade (IC) and Voter models, and present concrete sample complexity results  ...  We are now ready to provide the PAC learning algorithm for the partial observation setting with sample S = {(X 1 , Y 1 ), . . . , (X m , Y m )}; we shall sketch the proof here.  ... 
dblp:conf/nips/NarasimhanPS15 fatcat:lyaobxtahrewtek47xsmt6opw4

Page 8095 of Mathematical Reviews Vol. , Issue 98M [page]

1998 Mathematical Reviews  
Augustin) “Go with the winners” generators with applications to molecular modeling.  ...  On the other hand, under the distribution- free model, once a partial Occam algorithm is obtained for some concept class, we can use several remarkable techniques to boost it to a usual PAC learning algorithm  ... 

PAC: Assisted Value Factorisation with Counterfactual Predictions in Multi-Agent Reinforcement Learning [article]

Hanhan Zhou, Tian Lan, Vaneet Aggarwal
2022 arXiv   pre-print
Multi-agent reinforcement learning (MARL) has witnessed significant progress with the development of value function factorization methods.  ...  Empirical results demonstrate improved results of PAC over state-of-the-art value-based and policy-based multi-agent reinforcement learning algorithms on all benchmarks.  ...  MAR [39] learns the metarepresentation for generalization problems.  ... 
arXiv:2206.11420v2 fatcat:3cejhvnaszbl5pgg4tjme3hezy

A Fourier-based Framework for Domain Generalization [article]

Qinwei Xu, Ruipeng Zhang, Ya Zhang, Yanfeng Wang, Qi Tian
2021 arXiv   pre-print
To force the model to capture phase information, we develop a novel Fourier-based data augmentation strategy called amplitude mix which linearly interpolates between the amplitude spectrums of two images  ...  Extensive experiments on three benchmarks have demonstrated that the proposed method is able to achieve state-of-the-arts performance for domain generalization.  ...  For Digits-DG and PACS, we train the model for 50 epochs. For Office-Home, we train the model for 30 epochs. The initial learning rate for Digits-DG is 0.05 and decayed by 0.1 every 20 epochs.  ... 
arXiv:2105.11120v1 fatcat:fmvnko2pfzcfnnxsm6erfa3rwa

Page 8124 of Mathematical Reviews Vol. , Issue 2000k [page]

2000 Mathematical Reviews  
We relate the query model and the simple-PAC model with PACS.  ...  Summary: “We study a distribution dependent form of PAC learn- ing that uses probability distributions related to Kolmogorov com- plexity: the PACS model.  ... 

Page 6247 of Mathematical Reviews Vol. , Issue 96j [page]

1996 Mathematical Reviews  
(English summary) Inform. Process. Lett. 57 (1996), no.4, 189-195. Summary: “In this paper, we further characterize the complexity of noise-tolerant learning in the PAC model.  ...  Finally, we note that our general lower bound compares favorably with various general upper bounds for PAC learning in the presence of classification noise. 96j:68153 68T05 Kinber, Efim (1-DE-C; Newark  ... 

Learning Implicitly with Noisy Data in Linear Arithmetic [article]

Alexander Philipp Rader, Ionela G. Mocanu, Vaishak Belle, Brendan Juba
2020 arXiv   pre-print
In this work, we extend implicit learning in PAC-Semantics to handle noisy data in the form of intervals and threshold uncertainty in the language of linear arithmetic.  ...  However, recent work on so-called "implicit" learning has shown tremendous promise in terms of obtaining polynomial-time results for fragments of first-order logic.  ...  A model ρ ρ ρ can be seen simply as an element of Σ n , where Σ is the universe of the model. A partial model will be written using the regular font as ρ vs the bold font for a full model as ρ ρ ρ.  ... 
arXiv:2010.12619v1 fatcat:hicqarx54vedzpc4nrkefp3qfe

Machine learning based models for prediction of subtype diagnosis of primary aldosteronism using blood test

Hiroki Kaneko, Hironobu Umakoshi, Masatoshi Ogata, Norio Wada, Norifusa Iwahashi, Tazuru Fukumoto, Maki Yokomoto-Umakoshi, Yui Nakano, Yayoi Matsuda, Takashi Miyazawa, Ryuichi Sakamoto, Yoshihiro Ogawa
2021 Scientific Reports  
The aim of this retrospective cross-sectional study was to develop a predictive model for subtype diagnosis of PA based on machine learning methods using clinical data available in general practice.  ...  Machine learning models developed using blood test can help predict subtype diagnosis of PA in general practice.  ...  Therefore, these observations could partially explain the robustness of our machine learning-based model.  ... 
doi:10.1038/s41598-021-88712-8 pmid:33947886 fatcat:fhdsviuwsnbnvgyfdxyc6eyheq

Realizable Learning is All You Need [article]

Max Hopkins, Daniel Kane, Shachar Lovett, Gaurav Mahajan
2022 arXiv   pre-print
learning, partial learning, fair learning, and the statistical query model.  ...  This includes models with no known characterization of learnability such as learning with arbitrary distributional assumptions or general loss, as well as a host of other popular settings such as robust  ...  We also thank anonymous referees for constructive feedback, and especially for pointing out the notion of probabilistic representations and that prior work discussed in Section 5 falls under the general  ... 
arXiv:2111.04746v2 fatcat:wjjc672e35df7k4gtolunybfwm

Page 7822 of Mathematical Reviews Vol. , Issue 97M [page]

1997 Mathematical Reviews  
Summary: “In this paper we study a new view of the PAC-learning model in which the examples are more complicated than in the standard model.  ...  Summary: “We present a PAC-learning algorithm with member- ship queries for learning any multivariate polynomial over any finite field # under the uniform distribution.  ... 

Partial observability and learnability

Loizos Michael
2010 Artificial Intelligence  
previous learning models that deal with missing information.  ...  For this to happen, however, learning algorithms need to be developed that can deal with missing information in the learning examples in a principled manner, and without the need for external supervision  ...  Acknowledgements The author is grateful to Leslie Valiant for his advice, and for valuable suggestions and remarks on this research.  ... 
doi:10.1016/j.artint.2010.03.004 fatcat:lkqw47or3jfvzpdswf44tu6grm
« Previous Showing results 1 — 15 out of 14,138 results