20,535 Hits in 7.0 sec

A discriminative approach for finding and characterizing positivity violations using decision trees [article]

Ehud Karavani, Peter Bak, Yishai Shimoni
2019 arXiv   pre-print
To overcome these issues, we suggest to harness decision trees for detecting violations.  ...  By dividing the covariate space into mutually exclusive regions, each with maximized homogeneity of treatment groups, decision trees can be used to automatically detect subspaces violating positivity.  ...  Specifically, to Chen Yanover and Tal El-Hay for their thoughtful insights and help with background research.  ... 
arXiv:1907.08127v1 fatcat:ljnip65atfh4nlcspz3mxealla

Mammographic Classification of Breast Cancer Microcalcifications through Extreme Gradient Boosting

Haobang Liang, Jiao Li, Hejun Wu, Li Li, Xinrui Zhou, Xinhua Jiang
2022 Electronics  
This paper finds a way to select the best discriminative features as a collection to improve the accuracy.  ...  Then, we compared other machine learning techniques, including k-nearest neighbor (kNN), adaboostM1, decision tree, random decision forest (RDF), and gradient boosting decision tree (GBDT), with XGBoost  ...  Cancer Center, who have worked in obtaining the tissue specimens and provided them for the current study.  ... 
doi:10.3390/electronics11152435 fatcat:ffyb5sqtvrh7thmc3yndqq2q24

Proxy Non-Discrimination in Data-Driven Systems [article]

Anupam Datta, Matt Fredrikson, Gihyuk Ko, Piotr Mardziel, Shayak Sen
2017 arXiv   pre-print
We evaluate an implementation on a corpus of social datasets, demonstrating how to validate systems against these properties and to repair violations where they occur.  ...  We formalize proxy discrimination in data-driven systems, a class of properties indicative of bias, as the presence of protected class correlates that have causal influence on the system's output.  ...  The views, opinions, and/or findings expressed are those of the author(s) and should not be interpreted as representing the official views or policies of DARPA, the Air Force Research Laboratory, the National  ... 
arXiv:1707.08120v1 fatcat:foscoggsffhithhsnyzipqwqei

Exploring Business Process Deviance with Sequential and Declarative Patterns [article]

Giacomo Bergami, Chiara Di Francescomarino, Chiara Ghidini, Fabrizio Maria Maggi, Joonas Puura
2021 arXiv   pre-print
Using real-life logs from multiple domains, a range of feature types and different forms of decision rules are evaluated in terms of their ability to accurately discriminate between non-deviant and deviant  ...  The explanations characterizing the deviances are then extracted by direct and indirect methods for rule induction.  ...  using decision trees.  ... 
arXiv:2111.12454v1 fatcat:5ndkla4fenezzjfroybadhatde

Information Forests

Zhao Yi, Stefano Soatto, Maneesh Dewan, Yiqiang Zhan
2012 2012 Information Theory and Applications Workshop  
We describe Information Forests, an approach to classification that generalizes Random Forests by replacing the splitting criterion of non-leaf nodes from a discriminative onebased on the entropy of the  ...  Our approach relates to active learning, semi-supervised learning, mixed generative/discriminative learning.  ...  The continuation of this research was sponsored by DARPA under the MSEE program FA8650-11-1-7156, and by ARO under a MURI program W911NF-11-1-0391.  ... 
doi:10.1109/ita.2012.6181810 dblp:conf/ita/YiSDZ12 fatcat:z3hcjdzg2jhdxnzae6nqhk2l3i

Information Forests [article]

Zhao Yi, Stefano Soatto, Maneesh Dewan, Yiqiang Zhan
2012 arXiv   pre-print
We describe Information Forests, an approach to classification that generalizes Random Forests by replacing the splitting criterion of non-leaf nodes from a discriminative one -- based on the entropy of  ...  Our approach relates to active learning, semi-supervised learning, mixed generative/discriminative learning.  ...  The continuation of this research was sponsored by DARPA under the MSEE program FA8650-11-1-7156, and by ARO under a MURI program W911NF-11-1-0391.  ... 
arXiv:1202.1523v1 fatcat:r6qf7wlx6bct7ixvxubrmfyzbm

Discovering Data-Aware Declarative Process Models from Event Logs [chapter]

Fabrizio Maria Maggi, Marlon Dumas, Luciano García-Bañuelos, Marco Montali
2013 Lecture Notes in Computer Science  
However, the bulk of these techniques yield procedural process models that may be useful for detailed analysis, but do not necessarily provide a comprehensible picture of the process.  ...  This paper contributes to filling these gaps by proposing a technique to automatically discover declarative process models that incorporate both control-flow dependencies and data conditions.  ...  Given these two sets, classification techniques, such as decision tree learning, can be used to discover a condition on the data attributes that discriminates between fulfillments and violations.  ... 
doi:10.1007/978-3-642-40176-3_8 fatcat:ixzhcirp3fb2tcs3qufrrgnt3i

Defect classification of highly noisy NDE data using classifier ensembles

Kai F. Goebel, Weizhong Yan, Neil H. W. Eklund, Xiao Hu, Viswanath Avasarala, Jose Celaya, Daniele Inaudi, Wolfgang Ecke, Brian Culshaw, Kara J. Peters, Eric Udd
2006 Smart Structures and Materials 2006: Smart Sensor Monitoring Systems and Applications  
In this paper, we present a feature selection and classification approach that was used to assess highly noisy sensor data from a NDE field study.  ...  Results are shown from a number of different classifiers and classifier ensembles that were tuned to a set true positive rate using the Neyman-Pearson criterion.  ...  Single classifier results A leave-one-out (LOO) approach was employed, where all except one data points were used for training. The one remaining data point was used for testing.  ... 
doi:10.1117/12.659704 fatcat:cw4epzimsvbwfgwry3ky2aap7u

Explainable Global Fairness Verification of Tree-Based Classifiers [article]

Stefano Calzavara, Lorenzo Cazzaro, Claudio Lucchese, Federico Marcuzzi
2022 arXiv   pre-print
Given a tree-based classifier and a set of sensitive features potentially leading to discrimination, our analysis synthesizes sufficient conditions for fairness, expressed as a set of traditional propositional  ...  We present a new approach to the global fairness verification of tree-based classifiers.  ...  , v ∈ R is a threshold for the feature, and t l , t r are decision trees (left and right respectively).  ... 
arXiv:2209.13179v1 fatcat:65orehth6jbejilz4sncz4wr7m

Taking uncertainty seriously: simplicity versus complexity in financial regulation

David Aikman, Mirta Galesic, Gerd Gigerenzer, Sujit Kapadia, Konstantinos Katsikopoulos, Amit Kothiyal, Emma Murphy, Tobias Neumann
2021 Industrial and Corporate Change  
"fast-and-frugal" decision trees can perform comparably to standard, but more information-intensive, regressions.  ...  We find that: simple methods can sometimes dominate more complex modeling approaches for calculating banks' capital requirements, especially when data are limited or underlying risks are fat-tailed; simple  ...  Acknowledgment We would like to thank Mervyn King for instigating this research collaboration.  ... 
doi:10.1093/icc/dtaa024 fatcat:2nefkjplwzc27dn3nv73zlwyjq

Who initiates punishment, who joins punishment? Disentangling types of third-party punishers by neural traits

Thomas Baumgartner, Jan Hausfeld, Miguel Dos Santos, Daria Knoch
2021 Human Brain Mapping  
By using a neural trait approach, we were able to differentiate these three types clearly based on their neural signatures, allowing us to shed light on the underlying psychological mechanisms.  ...  We use anatomical brain characteristics in combination with stated motives to characterize these types.  ...  brain characteristics) and using stated motives for further characterization.  ... 
doi:10.1002/hbm.25648 pmid:34523772 pmcid:PMC8559474 fatcat:74jkdlli2vdfnmn2atyv4f2esu

Pattern classification by convex analysis

Miles N. Wernick
1991 Optical Society of America. Journal A: Optics, Image Science, and Vision  
A useful discriminant vector for pattern classification is one that maximizes the minimum separation of discriminant function values for two pattern classes.  ...  Digitized images are used to demonstrate application of the approach to two-class and multipleclass image classification tasks.  ...  Michael Morris for his support and encouragement of this research. The author also wishes to acknowledge Lennart Saaf for many helpful discussions and for providing the images shown in Fig. 5  ... 
doi:10.1364/josaa.8.001874 fatcat:mmsmn74bb5d6dhstlczbrmn5si

Correlating Activation and Target Conditions in Data-Aware Declarative Process Discovery [chapter]

Volodymyr Leno, Marlon Dumas, Fabrizio Maria Maggi
2018 Lecture Notes in Computer Science  
The approach has been implemented as a proofof-concept prototype and tested on both synthetic and real-life logs.  ...  However, the bulk of approaches for automated discovery of declarative process models are focused on the control-flow perspective of business processes and do not cover other perspectives, e.g., the data  ...  Violation and fulfillment feature vectors are used to train a decision tree. This procedure allows for finding correlations between the activation payloads and the target payloads.  ... 
doi:10.1007/978-3-319-98648-7_11 fatcat:shjzcldslzbwdkum564nu3favi


Maksim Koptelov, Albrecht Zimmermann, Pascal Bonnet, Ronan Bureau, Bruno Crémilleux
2018 Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining - KDD '18  
We evaluate the quality of the approach using benchmark data sets from the literature and show that we correct several short-comings of existing PAINS alerts that have recently been pointed out.  ...  To address this short-coming, we are in the process of developing a tool, PrePeP, that predicts PAINS, and allows experts to visually explore the reasons for the prediction.  ...  By using the result presented in [32] , which exploits the convexity of certain quality measures, such approaches can be used to find discriminative substructures for the (Q)SAR problem [15, 46] .  ... 
doi:10.1145/3219819.3219849 dblp:conf/kdd/KoptelovZBBC18 fatcat:jmgxyf5fzncvjdb7khkmeo67ea

YASENN: Explaining Neural Networks via Partitioning Activation Sequences [article]

Yaroslav Zharov, Denis Korzhenkov, Pavel Shvechikov, Alexander Tuzhilin
2018 arXiv   pre-print
Each partition may be described in a variety of ways, including examination of an interpretable model (e.g. a logistic regression or a decision tree) trained to discriminate between objects of those partitions  ...  Technically, YASENN distills the network with an ensemble of layer-wise gradient boosting decision trees and encodes the sequences of neuron activations with leaf indices.  ...  Decision Tree Decision tree T with K leaves and C-dimensional output is a function T (x) = W q(x) , where • W ∈ R K×C is a matrix of scores -W ij is a score for the j-th output of the i-th leaf; • q :  ... 
arXiv:1811.02783v1 fatcat:qrzmdxcmpbfp3n3i3ug3x4emwm
« Previous Showing results 1 — 15 out of 20,535 results