21,779 Hits in 3.8 sec

Accurate intelligible models with pairwise interactions

Yin Lou, Rich Caruana, Johannes Gehrke, Giles Hooker
2013 Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining - KDD '13  
Thus this paper postulates that for many problems, GA 2 M-models can yield models that are both intelligible and accurate.  ...  The resulting models, which we call GA 2 M-models, for Generalized Additive Models plus Interactions, consist of univariate terms and a small number of pairwise interaction terms.  ...  CONCLUSIONS We present a framework called GA 2 M for building intelligible models with pairwise interactions.  ... 
doi:10.1145/2487575.2487579 dblp:conf/kdd/LouCGH13 fatcat:jegjajhztjb5fof3skvru2w7xa

Modelling antibody side chain conformations using heuristic database search

D W Ritchie, G J Kemp
1997 Proceedings. International Conference on Intelligent Systems for Molecular Biology  
Even with a simple model, we find that a large proportion of side chain conformations are modelled accurately.  ...  Construction and placement of side chain clusters is guided by a heuristic cost function based on a simple model of side chain packing interactions.  ...  The dimensionless pairwise score, Fij, increases with good volumetric packing and good chemical interactions but decreases in tile presence of steric clashes.  ... 
pmid:9322043 fatcat:bcxpgwgmh5aohppzhwkpjycgni

Interactive Image Segmentation via Pairwise Likelihood Learning

Tao Wang, Quansen Sun, Qi Ge, Zexuan Ji, Qiang Chen, Guiyu Xia
2017 Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence  
This paper presents an interactive image segmentation approach where the segmentation problem is formulated as a probabilistic estimation manner.  ...  The unary prior probability of each pixel belonging to the foreground F and background B can be effectively estimated based on the similarities with label pairs (F, F),(F, B),(B, F) and (B, B).  ...  Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence 2962  ... 
doi:10.24963/ijcai.2017/412 dblp:conf/ijcai/WangSGJCX17 fatcat:z4fwwpj5abeendntciik2wyz2a

Adaptive Polling for Information Aggregation

Thomas Pfeiffer, Xi Gao, Yiling Chen, Andrew Mao, David Rand
At each step, our method estimates the strength parameters from the collected pairwise comparison data and adaptively chooses another pairwise comparison question for the next recruited worker.  ...  elicitation and aggregation over a combinatorial space can be achieved by integrating small pieces of potentially imprecise information, gathered from a large number of workers through simple, one-shot interactions  ...  A fair concern with our model is whether the Thurstone-Mosteller model accurately characterizes the participants' information in our setting.  ... 
doi:10.1609/aaai.v26i1.8099 fatcat:xjurwdumn5c7rb5kck5e56kgku

Preference learning for affective modeling

Georgios N. Yannakakis
2009 2009 3rd International Conference on Affective Computing and Intelligent Interaction and Workshops  
This paper outlines the use of preference learning for modeling emotional preferences and shows the methodology's promise for constructing accurate computational models of affect. 978-1-4244-4799-2/09/  ...  There is an increasing trend towards personalization of services and interaction.  ...  The class of entertainment (or "fun") preferences has been accurately predicted through affective models in those studies using feature data derived from multiple modes of interaction: interaction between  ... 
doi:10.1109/acii.2009.5349491 dblp:conf/acii/Yannakakis09a fatcat:eccgyzvwx5dnhn6pqrun7x2oii

Action-Guided Attention Mining and Relation Reasoning Network for Human-Object Interaction Detection

Xue Lin, Qi Zou, Xixia Xu
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
Extensive experiments on V-COCO and HICO-DET datasets demonstrate the effectiveness of the proposed model compared with the state-of-the-art approaches.  ...  Human-object interaction (HOI) detection is important to understand human-centric scenes and is challenging due to subtle difference between fine-grained actions, and multiple co-occurring interactions  ...  Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  ... 
doi:10.24963/ijcai.2020/154 dblp:conf/ijcai/LinZX20 fatcat:rgwl6ums2fgkteb6vqdoh7whg4

Intelligent Website Evolution of Public Sector Based on Data Mining Tools [chapter]

Jang Hee Lee, Gye Hang Hong
2005 Lecture Notes in Computer Science  
In this study we present an intelligent evolution model of public sector's website based on data mining tools in order to improve the whole users' satisfaction and the effects of using it, especially the  ...  users with lower performances by continuously redesigning and complementing the current key web pages.  ...  Intelligent Website Evolution Model of Public Sector We propose a new intelligent evolution model for public sector's website.  ... 
doi:10.1007/11531371_37 fatcat:hwe5trskf5hvnf37xnba6sosze

InterpretML: A Unified Framework for Machine Learning Interpretability [article]

Harsha Nori and Samuel Jenkins and Paul Koch and Rich Caruana
2019 arXiv   pre-print
InterpretML also includes the first implementation of the Explainable Boosting Machine, a powerful, interpretable, glassbox model that can be as accurate as many blackbox models.  ...  InterpretML exposes two types of interpretability - glassbox models, which are machine learning models designed for interpretability (ex: linear models, rule lists, generalized additive models), and blackbox  ...  Second, EBM can automatically detect and include pairwise interaction terms of the form: g(E[y]) = β 0 + f j (x j ) + f i j (x i , x j ) which further increases accuracy while maintaining intelligibility  ... 
arXiv:1909.09223v1 fatcat:uapsreh465cftf4u25rbamxxje

Intelligible Models for HealthCare

Rich Caruana, Yin Lou, Johannes Gehrke, Paul Koch, Marc Sturm, Noemie Elhadad
2015 Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining - KDD '15  
We present two case studies where high-performance generalized additive models with pairwise interactions (GA 2 Ms) are applied to real healthcare problems yielding intelligible models with state-of-the-art  ...  More accurate models such as boosted trees, random forests, and neural nets usually are not intelligible, but more intelligible models such as logistic regression, naive-Bayes, and single decision trees  ...  We thank Eric Horvitz, MD, PhD, Microsoft Research Redmond for help with the 30-day hospital readmission data and model.  ... 
doi:10.1145/2783258.2788613 dblp:conf/kdd/CaruanaLGKSE15 fatcat:6g275vrryfgz3hr2zn4pt5jsvy

A Batch Learning Framework for Scalable Personalized Ranking

Kuan Liu, Prem Natarajan
In this work we point out that these methods do not scale well in a large-scale setting, and this is partly due to the inaccurate pointwise or pairwise rank estimation.  ...  This new batch learning framework leads to more stable and accurate rank approximations compared to previous work. Moreover, it enables explicit use of parallel computation to speed up training.  ...  Pairwise algorithms convert the Copyright c 2018, Association for the Advancement of Artificial Intelligence (  ... 
doi:10.1609/aaai.v32i1.11608 fatcat:z65qee4mvvdrvfnkwppymdlv3i

ANOLEA: a www server to assess protein structures

F Melo, D Devos, E Depiereux, E Feytmans
1997 Proceedings. International Conference on Intelligent Systems for Molecular Biology  
High energy zones (HEZs) in the profile correlate with errors or with potential interacting zones of proteins.  ...  The calculations involve the non-local interactions between all the heavy atoms of the twenty standard amino acids in the molecule.  ...  Acknowledgments We thank Andrej Sali from the Rockefeller University, NY, USA, for providing coordinates of protein models in order to test our MFP.  ... 
pmid:9322034 fatcat:s24mdvu5jzaozlgg3ffdjvyhx4

StarNet: Pedestrian Trajectory Prediction using Deep Neural Network in Star Topology [article]

Yanliang Zhu, Deheng Qian, Dongchun Ren, Huaxia Xia
2020 arXiv   pre-print
Previous methods model only the pairwise interactions between pedestrians, which not only oversimplifies the interactions among pedestrians but also is computationally inefficient.  ...  In this paper, we propose a novel model StarNet to deal with these issues. StarNet has a star topology which includes a unique hub network and multiple host networks.  ...  First, the representation r is able to describe not only pairwise interactions but also collective ones. Such a comprehensive representation enables StarNet to make accurate predictions.  ... 
arXiv:1906.01797v2 fatcat:tsfts2n6zfgmjiis3bn6nugxt4

Part IV: Business and Financial Analytics Preface

David Enke
2014 Procedia Computer Science  
In "Demand Forecasting Based on Pairwise Item Associations", one researcher has utilized linear regression, along with pairwise item associations found in transactional data, to develop a framework for  ...  As the aforementioned papers highlight, adaptive systems and computational intelligence are providing the models and approaches necessary to offer such explanations, even in environments with increasing  ... 
doi:10.1016/j.procs.2014.09.082 fatcat:ntgwlk6t7ncohj66gmje35bnlu

Inverse Ising inference by combining Ornstein-Zernike theory with deep learning [article]

Soma Turi, Alpha A. Lee
2018 arXiv   pre-print
The neural network is able to learn from synthetic data, which can be generated with relative ease, to give accurate predictions on real world datasets.  ...  It is well-known that the Ising model is the maximum entropy model for binary variables which reproduces the sample mean and pairwise correlations.  ...  Molecules interact with a pairwise additive potential v(r 12 ), where r 12 is the distance between particles 1 and 2.  ... 
arXiv:1706.08466v2 fatcat:wpek3e24vrhjxnljiiumlpznym

Interactive Prior Elicitation of Feature Similarities for Small Sample Size Prediction

Homayun Afrabandpey, Tomi Peltola, Samuel Kaski
2017 Proceedings of the 25th Conference on User Modeling, Adaptation and Personalization - UMAP '17  
Furthermore, elicitation with an interactive scatterplot display outperforms straightforward elicitation where the users choose feature pairs from a feature list.  ...  Specifically, for learning to predict a target variable with Bayesian linear regression, the feature relationships are used to construct a Gaussian prior with a full covariance matrix for the regression  ...  Our simulation results and preliminary user study demonstrate that when collecting pairwise similarity knowledge using the proposed interactive intelligent interface, users are able to provide more informative  ... 
doi:10.1145/3079628.3079698 dblp:conf/um/AfrabandpeyPK17 fatcat:kzenfnqub5fvff7twxpalbgnwa
« Previous Showing results 1 — 15 out of 21,779 results