Filters








75,039 Hits in 3.5 sec

On Basing Lower-Bounds for Learning on Worst-Case Assumptions

Benny Applebaum, Boaz Barak, David Xiao
2008 2008 49th Annual IEEE Symposium on Foundations of Computer Science  
We show that unless the Polynomial Hierarchy collapses, such a statement cannot be proven via a large class of reductions including Karp reductions, truth-table reductions, and a restricted form of non-adaptive  ...  These results are obtained by showing that lower bounds for improper learning are intimately related to the complexity of zero-knowledge arguments and to the existence of weak cryptographic primitives.  ...  We thank Tal Malkin and Parikshit Gopalan for fruitful discussions.  ... 
doi:10.1109/focs.2008.35 dblp:conf/focs/ApplebaumBX08 fatcat:44nkrplugncare53zwbzd26lwu

Learning from Data to Speed-up Sorted Table Search Procedures: Methodology and Practical Guidelines [article]

Domenico Amato, Giosué Lo Bosco, Raffaele Giancarlo
2020 arXiv   pre-print
Here we study to what extend Machine Learning Techniques can contribute to obtain such a speed-up via a systematic experimental comparison of known efficient implementations of Sorted Table Search procedures  ...  We characterize the scenarios in which those latter can be profitably used with respect to the former, accounting for both CPU and GPU computing.  ...  ACKNOWLEDGMENTS Many thanks to Giorgio Vinciguerra for helpful discusions and comments and for being so kind to run some of the experiments reported here on hardware available at his Institution.  ... 
arXiv:2007.10237v3 fatcat:hquh6i4r6bc6fklunzbqfdts6m

Active Clustering with Model-Based Uncertainty Reduction [article]

Caiming Xiong, David Johnson, Jason J. Corso
2014 arXiv   pre-print
Here, we propose a novel online framework for active semi-supervised spectral clustering that selects pairwise constraints as clustering proceeds, based on the principle of uncertainty reduction.  ...  Using a first-order Taylor expansion, we decompose the expected uncertainty reduction problem into a gradient and a step-scale, computed via an application of matrix perturbation theory and cluster-assignment  ...  ACKNOWLEDGEMENTS We are grateful for the support in part provided through the following grants: NSF CAREER IIS-0845282, ARO YIP W911NF-11-1-0090, DARPA Minds Eye W911NF-10-2-0062, DARPA CSSG D11AP00245  ... 
arXiv:1402.1783v2 fatcat:4z7mypvwyjgl7lba6pm52qy4na

Adaptive Dimensionality Reduction with Semi-Supervision (AdDReSS): Classifying Multi-Attribute Biomedical Data

George Lee, David Edmundo Romo Bucheli, Anant Madabhushi, Daoqiang Zhang
2016 PLoS ONE  
We present our novel methodology, AdDReSS (Adaptive Dimensionality Reduction with Semi-Supervision), to demonstrate that fewer labeled instances identified via AL in embedding space are needed for creating  ...  Currently, no approach that we are aware of has attempted to use active learning in the context of dimensionality reduction approaches for improving the construction of low dimensional representations.  ...  reduction and active learning.  ... 
doi:10.1371/journal.pone.0159088 pmid:27421116 pmcid:PMC4946789 fatcat:2mqamrce65f4botsloyn2ymrrm

Construction and learnability of canonical Horn formulas

Marta Arias, José L. Balcázar
2011 Machine Learning  
Using these tools, we provide a new, simpler validation of the classic Horn query learning algorithm of Angluin, Frazier, and Pitt, and we prove that this algorithm always outputs the GD basis regardless  ...  We extend the canonical representation to general Horn, by providing a reduction from definite to general Horn CNF.  ...  Positive results consist usually in algorithms that learn a given concept class, via a given protocol, within certain query or time resources.  ... 
doi:10.1007/s10994-011-5248-5 fatcat:a762c5jo7nawhdw5z7kjscq2sm

Active Clustering with Model-Based Uncertainty Reduction

Caiming Xiong, David M. Johnson, Jason J. Corso
2017 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Here, we propose a novel online framework for active semi-supervised spectral clustering that selects pairwise constraints as clustering proceeds, based on the principle of uncertainty reduction.  ...  Using a first-order Taylor expansion, we decompose the expected uncertainty reduction problem into a gradient and a step-scale, computed via an application of matrix perturbation theory and cluster-assignment  ...  ACKNOWLEDGEMENTS We are grateful for the support in part provided through the following grants: NSF CAREER IIS-0845282, ARO YIP W911NF-11-1-0090, DARPA Minds Eye W911NF-10-2-0062, DARPA CSSG D11AP00245  ... 
doi:10.1109/tpami.2016.2539965 pmid:26978555 fatcat:yvrb433525hipjmsy6nqx4qsdm

TLDR: Twin Learning for Dimensionality Reduction [article]

Yannis Kalantidis, Carlos Lassance, Jon Almazan, Diane Larlus
2021 arXiv   pre-print
for learning.  ...  In this paper, we unify these two families of approaches from the angle of manifold learning and propose TLDR, a dimensionality reduction method for generic input spaces that is porting the simple self-supervised  ...  Acknowledgements The authors want to sincerely thank Christopher Dance for his early comments that helped shape this work.  ... 
arXiv:2110.09455v1 fatcat:6ocuv5x2rfhh3n7tj5l57vn4ua

Learning hidden semantic cues using support vector clustering

Jia-Wen Tung, Chiou-Ting Hsu
2005 IEEE International Conference on Image Processing 2005  
Given a dimensionreduced semantic space, we then perform the image query in terms of the semantic attributes instead of merely the visual features.  ...  In short-term learning, we apply the general 2-class SVM classification to initialize the semantic space.  ...  Dimension reduction via support vector clustering As discussed in Sec. 2.2, the semantic space expands with increased query sessions.  ... 
doi:10.1109/icip.2005.1529969 dblp:conf/icip/TungH05 fatcat:q4s4s3trpvdbjjddkdnzx7a4tm

Active Learning in CNNs via Expected Improvement Maximization [article]

Udai G. Nagpal, David A Knowles
2020 arXiv   pre-print
Pool-based active learning techniques have the potential to mitigate these issues, leveraging models trained on limited data to selectively query unlabeled data points from a pool in an attempt to expedite  ...  Here we present "Dropout-based Expected IMprOvementS" (DEIMOS), a flexible and computationally-efficient approach to active learning that queries points that are expected to maximize the model's improvement  ...  Acknowledgements We would like to thank the New York Genome Center and XSEDE for computational resources used in active learning experiments.  ... 
arXiv:2011.14015v1 fatcat:lainb3qfizcejai4trmjwcv4ma

Automatic Age Estimation from Face Images via Deep Ranking

Huei-Fang Yang, Bo-Yao Lin, Kuang-Yu Chang, Chu-Song Chen
2015 Procedings of the British Machine Vision Conference 2015  
For rank k, we separate X into two subsets, X + k and X − k , as follows: Next, we use the two subsets to learn a binary classifier from the network, which then conducts an answer to the query: "Is the  ...  We encode the age rank based on the reduction framework [5] .  ...  We encode the age rank based on the reduction framework [5] .  ... 
doi:10.5244/c.29.55 dblp:conf/bmvc/YangLCC15 fatcat:xf6nbpynuravha6rtju6ydz7iu

Learned Sorted Table Search and Static Indexes in Small Model Space [article]

Domenico Amato and Giosuè Lo Bosco and Raffaele Giancarlo
2021 arXiv   pre-print
Finally, our findings are of interest to designers, since they highlight the need for further studies regarding the time-space relation in Learned Indexes.  ...  Machine Learning Techniques, properly combined with Data Structures, have resulted in Learned Static Indexes, innovative and powerful tools that speed-up Binary Search, with the use of additional space  ...  Based on it, it is immediate to compute the reduction factor for that query.  ... 
arXiv:2107.09480v5 fatcat:j34mcxyvbbdkvdtrjza7xsbrxa

Human-guided Robot Behavior Learning: A GAN-assisted Preference-based Reinforcement Learning Approach [article]

Huixin Zhan, Feng Tao, Yongcan Cao
2020 arXiv   pre-print
To reduce and minimize the need for human queries, we propose a new GAN-assisted human preference-based reinforcement learning approach that uses a generative adversarial network (GAN) to actively learn  ...  Human demonstrations can provide trustful samples to train reinforcement learning algorithms for robots to learn complex behaviors in real-world environments.  ...  The existence of correct reward functions is crucial for the subsequent development of robotic control systems via reinforcement learning algorithms.  ... 
arXiv:2010.07467v1 fatcat:kscg6oykwvdchidrlt4iv5vrqa

Learning to Complement Humans

Bryan Wilder, Eric Horvitz, Ece Kamar
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
The goal is to focus machine learning on problem instances that are difficult for humans, while recognizing instances that are difficult for the machine and seeking human input on them.  ...  A rising vision for AI in the open world centers on the development of systems that can complement humans for perceptual, diagnostic, and reasoning tasks.  ...  Acknowledgments We thank Besmira Nushi for advice on characterizing error regions and insightful conversations throughout, as well as the CAMELYON team for providing data on pathologist panel responses  ... 
doi:10.24963/ijcai.2020/212 dblp:conf/ijcai/WilderHK20 fatcat:z7fvlqfbrvb2jonkt7natyypgq

Learning to Complement Humans [article]

Bryan Wilder, Eric Horvitz, Ece Kamar
2020 arXiv   pre-print
The goal is to focus machine learning on problem instances that are difficult for humans, while recognizing instances that are difficult for the machine and seeking human input on them.  ...  A rising vision for AI in the open world centers on the development of systems that can complement humans for perceptual, diagnostic, and reasoning tasks.  ...  Acknowledgments We thank Besmira Nushi for advice on characterizing error regions and insightful conversations throughout, as well as the CAMELYON team for providing data on pathologist panel responses  ... 
arXiv:2005.00582v1 fatcat:tr6q6y47ljdfbaijgmc7msg2ju

Private Data Release via Learning Thresholds [chapter]

Moritz Hardt, Guy N. Rothblum, Rocco A. Servedio
2012 Proceedings of the Twenty-Third Annual ACM-SIAM Symposium on Discrete Algorithms  
We instantiate this general reduction with algorithms for learning thresholds, obtaining new results for differentially private data release.  ...  Our primary contribution is a computationally efficient reduction from differentially private data release for a class of counting queries, to learning thresholded sums of predicates from a related class  ...  We especially thank an anonymous referee for pointing out the alternative simpler algorithm for releasing parity counting queries, outlined in Section 6.1.  ... 
doi:10.1137/1.9781611973099.15 dblp:conf/soda/HardtRS12 fatcat:qar27pwle5gorako5djckgskyi
« Previous Showing results 1 — 15 out of 75,039 results