Filters








11,314 Hits in 4.7 sec

Learning Explanatory Rules from Noisy Data [article]

Richard Evans, Edward Grefenstette
2018 arXiv   pre-print
data distribution of the domain we wish to test on.  ...  cannot address, while providing data efficiency and generalisation beyond what neural networks on their own can achieve.  ...  This approach is very similar to ours in its aim: we both want to learn first-order rules from noisy data by maximising a conditional probability.  ... 
arXiv:1711.04574v2 fatcat:lcakzuirdbeudgdolnwrkiei3i

Learning Explanatory Rules from Noisy Data

Richard Evans, Edward Grefenstette
2018 The Journal of Artificial Intelligence Research  
data distribution of the domain we wish to test on.  ...  cannot address, while providing data efficiency and generalisation beyond what neural networks on their own can achieve.  ...  This approach is very similar to ours in its aim: we both want to learn first-order rules from noisy data by maximising a conditional probability.  ... 
doi:10.1613/jair.5714 fatcat:u4bzq4dz4jaalmv2p4eskrv6ku

Learning Explanatory Rules from Noisy Data (Extended Abstract)

Richard Evans, Edward Grefenstette
2018 Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence  
Although mitigated by a variety of model regularisation methods, the common cure is to seek large amounts of training data—which is not necessarily easily obtained—that sufficiently approximates the data  ...  cannot address, while providing data efficiency and generalisation beyond what neural networks on their own can achieve.  ...  This allows us to apply gradient descent to learn which clauses to turn on and off, even in the presence of noisy or ambiguous data.  ... 
doi:10.24963/ijcai.2018/792 dblp:conf/ijcai/EvansG18 fatcat:f7skf5hnbjgcfbjbleid3j7qda

Methodologies from Machine Learning in Data Analysis and Software

D. Michie
1991 Computer journal  
new data sampled from the same source but also possess the quality of clear explanatory structure.  ...  New developments in the computer induction of decision rules have contributed to two areas, multivariate data analysis and computer assisted software engineering.  ...  Quinlan adapted his C4 algorithm for inducing trees from noisy data so as to generate solutions in the form of compact sets of logic rules. 26 After pooling the branches harvested from the trees separately  ... 
doi:10.1093/comjnl/34.6.559 fatcat:fw7usy456jcl3fq4th4p3fxgby

Investigating ADR mechanisms with Explainable AI: a feasibility study with knowledge graph mining

Emmanuel Bresso, Pierre Monnin, Cédric Bousquet, François-Elie Calvier, Ndeye-Coumba Ndiaye, Nadine Petitpain, Malika Smaïl-Tabbone, Adrien Coulet
2021 BMC Medical Informatics and Decision Making  
Aside from clinical trials, many elements of knowledge about drug ingredients are available in open-access knowledge graphs, such as their properties, interactions, or involvements in pathways.  ...  In an Explainable AI perspective, we explore simple classification techniques such as Decision Trees and Classification Rules because they provide human-readable models, which explain the classification  ...  reaction; AI:: Artificial intelligence; AUC:: Area under the curve, usually under the ROC curve; DILI:: Drug-induced liver injury; GCN:: Graph convolutional networks; GO:: Gene ontology; LOD:: Linked open data  ... 
doi:10.1186/s12911-021-01518-6 pmid:34039343 fatcat:ay76ba6npzdjfmsfryz2cg4xaa

Weakly Supervised Multi-task Learning for Concept-based Explainability [article]

Catarina Belém, Vladimir Balayan, Pedro Saleiro, Pedro Bizarro
2021 arXiv   pre-print
To address both, we propose to: i) use expert rules to generate a large dataset of noisy concept labels, and ii) apply two distinct multi-task learning strategies combining noisy and golden labels.  ...  To obtain faithful concept-based explanations, we leverage multi-task learning to train a neural network that jointly learns to predict a decision task based on the predictions of a precedent explainability  ...  Based on a few rule-based predictors available off-the-shelf in historical production data, we are able to automatically generate noisy concept labels for datasets with millions of instances.  ... 
arXiv:2104.12459v1 fatcat:grah66ofqnf7dhltd6esqa25hq

An information-fusion method to identify pattern of spatial heterogeneity for improving the accuracy of estimation

Lianfa Li, Jinfeng Wang, Zhidong Cao, Ershun Zhong
2007 Stochastic environmental research and risk assessment (Print)  
Different from many stratification schemes which just uses the goal variable to stratify which is too simplified, information from multiple sources can be fused to identify pattern of spatial heterogeneity  ...  Data mining is major analysis components in our method: multivariate statistics, association analysis, decision tree and rough set are used in data filter, identification of contributing factors, and examination  ...  Acknowledgments This research has been done in support of the grants 40601077/D0120 and 40471111/D0120 from the Natural Science Foundation of China, and the grant 2007AA12Z233 from Hi-tech Research and  ... 
doi:10.1007/s00477-007-0179-1 fatcat:qidqux46rvabjb46dyhhpgqmn4

Interleaved Inductive-Abductive Reasoning for Learning Complex Event Models [chapter]

Krishna Dubba, Mehul Bhatt, Frank Dylla, David C. Hogg, Anthony G. Cohn
2012 Lecture Notes in Computer Science  
Typed Inductive Logic Programming (Typed-ILP) is used as a basis for learning the domain theory by generalising from observation data, whereas abductive reasoning is used for noisy data correction by scenario  ...  We apply the model to an airport domain consisting of video data for 15 turn-arounds from six cameras simultaneously monitoring logistical processes concerned with aircraft arrival, docking, departure  ...  We show how well-fitted, semantically meaningful event models can be learned from noisy data by interleaving induction and abduction.  ... 
doi:10.1007/978-3-642-31951-8_14 fatcat:7q3tpjo2z5afrpudv63d6mvg6e

Abductive Plan Recognition by Extending Bayesian Logic Programs [chapter]

Sindhu Raghavan, Raymond J. Mooney
2011 Lecture Notes in Computer Science  
We learn the parameters in BALPs using the Expectation Maximization algorithm adapted for BLPs.  ...  Finally, we present an experimental evaluation of BALPs on three benchmark data sets and compare its performance with the state-of-the-art for plan recognition.  ...  Acknowledgements We would like to thank Nate Blaylock for sharing Linux and Monroe data sets, Vibhav Gogate for helping us modify SampleSearch algorithm for our experiments, and Parag Singla for his valuable  ... 
doi:10.1007/978-3-642-23783-6_40 fatcat:s4m6l6545farbonobi4xkuvo7a

Explanatory Interactive Machine Learning

Stefano Teso, Kristian Kersting
2019 Proceedings of the 2019 AAAI/ACM Conference on AI, Ethics, and Society - AIES '19  
We demonstrate that this can boost the predictive and explanatory powers of, and the trust into, the learned model, using text (e.g.  ...  Consequently, we propose the novel framework of explanatory interactive learning where, in each step, the learner explains its query to the user, and the user interacts by both answering the query and  ...  Explanatory Interactive Learning (XIL) In XIL, a learner is able to interactively query the user (or some other information source) to obtain the desired outputs at data points.  ... 
doi:10.1145/3306618.3314293 dblp:conf/aies/TesoK19 fatcat:4cswu3bzubeqrc4t2egydmt6ri

Data-intensive analytics for predictive modeling

C. V. Apte, S. J. Hong, R. Natarajan, E. P. D. Pednault, F. A. Tipu, S. M. Weiss
2003 IBM Journal of Research and Development  
The Data Abstraction Research Group was formed in the early 1990s, to bring focus to the Mathematical Sciences department's work in the emerging area of Knowledge Discovery and Data Mining (KD&DM).  ...  Most activities in this group have been performed in the technical area of predictive modeling, roughly at the intersection of machine learning, statistical modeling, and database technology.  ...  Acknowledgment We thank the many individuals at IBM, both within Research, as well as in IBM's Industry Sectors, Global Services, and the Software Group, from whose collaborations we have immensely benefited  ... 
doi:10.1147/rd.471.0017 fatcat:yf4uiodrurglnftkllu6nyzcle

"Why Should I Trust Interactive Learners?" Explaining Interactive Queries of Classifiers to Users [article]

Stefano Teso, Kristian Kersting
2018 arXiv   pre-print
We demonstrate that this can boost the predictive and explanatory powers of and the trust into the learned model, using text (e.g.  ...  Consequently, we propose the novel framework of explanatory interactive learning: in each step, the learner explains its interactive query to the user, and she queries of any active classifier for visualizing  ...  KK acknowledges the support by the German Science Foundation project "CAML: Argumentative Machine Learning" (KE1686/3-1) as part of the SPP 1999 (RATIO).  ... 
arXiv:1805.08578v1 fatcat:66fi77hoqbah5imkuyxbzgjuta

Mining parasite data using genetic programming

J BARRETT, A KOSTADINOVA, J RAGA
2005 Trends in Parasitology  
Application of genetic programming to a problem using parasites as biological tags demonstrates its potential for developing explanatory models using data that are both complex and noisy.  ...  in data that are both complex and noisy [6] [7] [8] [9] .  ...  Application of genetic programming to a problem using parasites as biological tags demonstrates its potential for developing explanatory models using data that are both complex and noisy.  ... 
doi:10.1016/j.pt.2005.03.007 pmid:15837607 fatcat:dyuf45suhzaxfo25qsvriowdre

The impact of measurement scale and correlation structure on classification performance of inductive learning and statistical methods

Ingoo Han, John S. Chandler, Ting-Peng Liang
1996 Expert systems with applications  
The purpose of this study is to investigate the impact of measurement scale of explanatory variables on the relative performance of the statistical method (probit) and the inductive learning method (1D3  ...  This is a comparative study of inductive learning and statistical methods using the simulation approach to provide a generalizable results.  ...  Arinze & Subbanarasimha (1994) chose the data set which were noisy and non-normally distributed (skewed) to test the relative efficiency of rule based induction to regression model.  ... 
doi:10.1016/0957-4174(95)00047-x fatcat:yfolg6jabbaszoushvarek5lje

An Entity Resolution Approach to Isolate Instances of Human Trafficking Online

Chirag Nagpal, Kyle Miller, Benedikt Boecking, Artur Dubrawski
2017 Proceedings of the 3rd Workshop on Noisy User-generated Text  
Most of the features are self explanatory.  ...  Rule Learning We extract clusters and identify records that are associated with human trafficking using domain knowledge from experts.  ... 
doi:10.18653/v1/w17-4411 dblp:conf/aclnut/NagpalMBD17 fatcat:jrer24uxmbehdblyjljg4ggufq
« Previous Showing results 1 — 15 out of 11,314 results