Machine Learning with Known Input Data Uncertainty Measure
| Autor: | Wojciech Czarnecki, Igor T. Podolak |
|---|---|
| Přispěvatelé: | Faculty of Mathematics and Computer Science [Poznan], Adam Mickiewicz University in Poznań (UAM), Faculty of Mathematics and Computer Science of the Jagiellonian University, Uniwersytet Jagielloński w Krakowie = Jagiellonian University (UJ), Khalid Saeed, Rituparna Chaki, Agostino Cortesi, Sławomir Wierzchoń, TC 8 |
| Jazyk: | angličtina |
| Rok vydání: | 2013 |
| Předmět: |
jitter
Computer science Active learning (machine learning) [SHS.INFO]Humanities and Social Sciences/Library and information sciences 02 engineering and technology Overfitting computer.software_genre Machine learning Measure (mathematics) Tikhonov regularization 020204 information systems 0202 electrical engineering electronic engineering information engineering [INFO]Computer Science [cs] Cluster analysis uncertainty Uncertainty analysis Artificial neural network business.industry Online machine learning neural networks random variables machine learning classification 020201 artificial intelligence & image processing Data mining Artificial intelligence business computer clustering |
| Zdroj: | Lecture Notes in Computer Science 12th International Conference on Information Systems and Industrial Management (CISIM) 12th International Conference on Information Systems and Industrial Management (CISIM), Sep 2013, Krakow, Poland. pp.379-388, ⟨10.1007/978-3-642-40925-7_35⟩ Computer Information Systems and Industrial Management ISBN: 9783642409240 CISIM |
| DOI: | 10.1007/978-3-642-40925-7_35⟩ |
| Popis: | Part 7: Algorithms; International audience; Uncertainty of the input data is a common issue in machine learning. In this paper we show how one can incorporate knowledge on uncertainty measure regarding particular points in the training set. This may boost up models accuracy as well as reduce overfitting. We show an approach based on the classical training with jitter for Artificial Neural Networks (ANNs). We prove that our method, which can be applied to a wide class of models, is approximately equivalent to generalised Tikhonov regularisation learning. We also compare our results with some alternative methods. In the end we discuss further prospects and applications. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|