Summarization in Sequential Data with Applications to Procedure Learning
|
|
- Gregory Gilmore
- 5 years ago
- Views:
Transcription
1 Summarization in Sequential Data with Applications to Procedure Learning Ehsan Elhamifar College of Computer and Information Science Northeastern University
2 Data deluge > 400hrs videos / minute > 300M photos / day 24/7/365
3 Big data opportunities Capturing modes and patterns unseen in small data - Increase learning and inference performance From Nearest Neighbors It takes a large amount of data for our method to succeed. We saw dramatic improvement when moving from ten thousand to two million images. [Hayes, Efros, SIGGRAPH 07] As we increase the size of the dataset from 10^5 to the 10^8 images, the quality of the retrieved set increases dramatically. [Torralba, Fergus, Freeman, PAMI 08]
4 Big data opportunities Capturing modes and patterns unseen in small data - Increase learning and inference performance To Deep Neural Networks Classification Error (%) Dataset Size From 100 to 100,000 training samples, classification error drops from 25% to less than 5%. [Srivastava-Hinton-Krizhevsky-Sutskever-Salakhutdinov 14]
5 Big data challenges Capturing modes and patterns unseen in small data - Increase learning and inference performance: from NNs to DNNs Limiting computational and memory resources - We have more data than capacities of our resources Economist, Feb 2010
6 Subset selection Find small data capturing statistical properties of large data Subset Selection
7 Subset selection components How to characterize informativeness of items? How to recover the most informative items? max S f(s) 1. What objective functions? 2. How to optimize?
8 This Talk max S f(s) Subset selection is NP-hard, hence, approximate algorithms - Better summarization: closer to global optimum Focus on one class of objective functions: Facility Location Develop efficient sparse optimization Show it can achieve global optimum (under certain condition)
9 This Talk Sequential data (video, text, signals, ): large part of Big Data (CNN) -- A Japanese rocket roared into orbit early Friday (Thursday afternoon ET) carrying what NASA calls its most precise instrument yet for measuring rain and snowfall. The Global Precipitation Measurement (GPM) satellite is the first of five earth science launches NASA has planned for The 4-ton spacecraft is the most sophisticated platform yet for measuring rainfall, capable of recording amounts as small as a hundredth of an inch an hour, said Gail Skofronick Jackson, GPM's deputy project scientist. The $900 million satellite is a joint project with the Japanese space agency JAXA, and it lifted off from Tanegashima Space Center at 3:37 a.m. Friday (1:37 p.m. Thursday ET). In a little over a half hour, it had reached orbit, deployed its solar panels and began beaming signals back to its controllers, NASA said. Also, once fully activated, GPM will use both radar and microwave instruments to detect falling snow for the first time. It will also combine data from other satellites with its own readings, beaming back a snapshot of worldwide precipitation every three hours, Sequential data have structured dependencies - Instructional videos: ordering of key steps Key steps
10 This Talk Sequential data (video, text, signals, ): large part of Big Data Sequential data have structured dependencies - Existing SS: data is a bag of randomly permutable items Generalize facility location to handle seq summarization x 1 x 7 x 4 x 8 x 3 x 5 x 2 max S f(s) x 9 x 6 Develop new objective func Develop efficient opt algs x 9 x 6 x 4 x 2
11 Subset selection from dissimilarities Given dissimilarities d : X Y! R 0, source target select a small subset of X that well represent Y w.r.t. d(, ) x 1 x 2 d 11 y 1 y 2 x 1 x 2 y 1 y 2 x M d MN y N x M y N d(x i, y j )=d ij X Y d(, ) : how well x i represents - = data, = data = Euclidean/geodesic distance X Y d(, ) - = models, = data = representation/coding error y j
12 Subset selection from dissimilarities Given dissimilarities d : X Y! R 0, source target select a small subset of X that well represent Y w.r.t. d(, ) x 1 x 2 d 11 y 1 y 2 x 1 x 2 y 1 y 2 x M d MN y N x M y N d(x i, y j )=d ij : how well x i represents y j Not necessarily metric: asymmetric, violate triangle ineq
13 Dissimilarity-based Sparse Subset Selection Let r j 2{1,...,M} be index of representative of Approach: define & maximize potential function (r 1,...,r N ), enc (r 1,...,r N ) card (r 1,...,r N ) y j x i x rj y j enc(r 1,...,r N ), NY j=1 e d r j,j Prefer lower encoding card(r 1,...,r N ), e {r 1,...,r N } Prefer small # representatives Max over {r 1,...,r N } 2 {1,...,M} N : combinatorial E. Elhamifar, G. Sapiro, S. Sastry, TPAMI, 2016
14 DS3 formulation Develop a sparse binary optimization d ij z ij Define variables z i,j, I(r j = i) 2 {0, 1} x i MY NY = exp( d i,j z i,j ) exp I(k[z i,1...z i,n ]k p ) i=1 j=1 i=1 y j Minimizing log : Nonconvex min {z i,j } i=1 NX j=1 s. t. z i,j 2 {0, 1}, d i,j z i,j + z i,j =1, i=1 i=1 I(k[z i,1... z i,n ]k p ) 8 i, j ~ Facility location E. Elhamifar, G. Sapiro, S. Sastry, TPAMI, 2016
15 DS3 formulation Develop a sparse binary optimization d ij z ij Define variables z i,j, I(r j = i) 2 {0, 1} x i MY NY = exp( d i,j z i,j ) exp I(k[z i,1...z i,n ]k p ) i=1 j=1 i=1 y j Minimizing log : min {z i,j } NX d i,j z i,j + zi,1 z i,n i=1 j=1 i=1 Convex p 2{2, 1} s. t. z i,j 0, z i,j =1, i=1 DS3 8 i, j p E. Elhamifar, G. Sapiro, S. Sastry, TPAMI, 2016
16 Toy example Source = target = {data points} 0 data points data points 1 representatives 8 2 > 1 representatives 3 > 2 data points representatives Z matrix for λ = 0.01 λ max, 1 4 Z matrix for λ = 0.1 λ max, 1 Z matrix for λ = λ max, min {z i,j } i=1 NX d i,j z i,j + j=1 i=1 zi,1 z i,n p s. t. z i,j 0, z i,j =1, i=1 8 i, j
17 Theoretical analysis When is convex relaxation exact? Theorem: Let r j be representative of j, via non-convex optimization. The convex relaxation is exact if 1) + d rj,j <d j0,j N rj Clusters sufficiently separated j r j j 0 2) d j0,j apple Nrj + d rj,j j 0 r j j Sufficiently small radius E. Elhamifar, M. De Paolis Kaluza `17
18 Video summarization Comparison with greedy - Solve FL via Greedy and sparse coding (DS3) TVSum SumMe Cost function value Greedy DS3 Greedy DS3
19 Recognition via representatives Select fraction of training samples via DS Suburb Suburb Coast Coast Forest Forest Highway Highway Insidecity Insidecity Mountain Mountain Opencountry Opencountry Street Street Tallbuilding Tallbuilding Office Office Bedroom Industrial Kitchen Livingroom Store Store Livingroom Kitchen Industrial Bedroom Office Tallbuilding Street Opencountry Mountain Insidecity Highway Forest Coast Suburb Bedroom Industrial Kitchen Livingroom Store Store Livingroom Kitchen Industrial Bedroom Office Tallbuilding Street Opencountry Mountain Insidecity Highway Forest Coast Suburb DS3 selecting 35% samples Trained via all samples E. Elhamifar, G. Sapiro, S. Sastry, TPAMI, 2016
20 Recognition via representatives Select fraction of training samples via DS Suburb Suburb Coast Coast Forest Forest Highway Highway Insidecity Insidecity Mountain Mountain Opencountry Opencountry Street Street Tallbuilding Tallbuilding Office Office Bedroom Industrial Kitchen Livingroom Store Store Livingroom Kitchen Industrial Bedroom Office Tallbuilding Street Opencountry Mountain Insidecity Highway Forest Coast Suburb Bedroom Industrial Kitchen Livingroom Store Store Livingroom Kitchen Industrial Bedroom Office Tallbuilding Street Opencountry Mountain Insidecity Highway Forest Coast Suburb DS3 selecting 35% samples Trained via all samples E. Elhamifar, G. Sapiro, S. Sastry, TPAMI, 2016
21 Recognition via representatives Select fraction of training samples via DS3 Suburb Suburb Coast Forest Highway Coast Forest Highway Insidecity Insidecity Mountain Mountain Opencountry Opencountry Representatives improve computational time and Street Street Tallbuilding can improve Tallbuilding performance Office Office Bedroom Industrial Kitchen Livingroom Store Store Livingroom Kitchen Industrial Bedroom Office Tallbuilding Street Opencountry Mountain Insidecity Highway Forest Coast Suburb Bedroom Industrial Kitchen Livingroom Store Suburb Store Livingroom Kitchen Industrial Bedroom Office Tallbuilding Street Opencountry Mountain Insidecity Highway Forest Coast E. Elhamifar, G. Sapiro, S. Sastry, TPAMI, 2016
22 Sequential subset selection Incorporate dynamics of sequential data into SS - Source set: with transition dynamics - Target set: sequential data y 1 x 1 y T x M Goal: find (r 1,...,r T ) with high transition probability that well encodes the data and has small cardinality y 1 y 2 y 3 y 1 y 2 y 3 x 1 x 2 x 3 x 1 x 2 x 3 (r 1,r 2,r 3 )=(1, 1, 2) (r 1,r 2,r 3 )=(1, 1, 3) E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
23 Sequential facility location Let r t be index of representative of y t y 1 y T x 1 x M Maximize global potential function Prefer a sequence of compatible reps max (r 1,...,r T ) enc(r 1,...,r N ) card (r 1,...,r N ) dyn (r 1,...,r N ) - Dynamic potential: n-th order Markov model dyn(r 1,...,r N ), Y t p t (x rt x rt 1,...,x rt n )! E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
24 Sequential facility location SeqFL: Solve optimization on assignment variables {z i,t } max {z i,t } TX t=1 + ( i=1 z i,t d i,t M X i=1 z i,1 log p 1 (x i )+ i=1 s. t. z i,t 2 {0, 1}, k z i,1 z i,t k1 TX t=2 i,i 0 =1 z i,t =1, 8 i, t. i=1 z i,t 1 z i 0,t log p(x i 0 x i )) x 1 y 1 z i,t 1 z i 0,t =1 y t 1 y t x M y T p(x i 0 x i ) should be large! x i x i 0 E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
25 Sequential facility location SeqFL: Solve optimization on assignment variables max {z i,t } TX t=1 + ( i=1 Non-convex: does not make sense (here) to relax! - Solve via z i,t d i,t M X i=1 z i,1 log p 1 (x i )+ i=1 s. t. z i,t 2 {0, 1}, k z i,1 z i,t k1 TX t=2 i,i 0 =1 z i,t =1, 8 i, t. i=1 z i,t 1 z i 0,t log p(x i 0 x i )) Max-Sum Message Passing {z i,t } x 1 x M y 1 y T E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
26 SeqFL via message passing C 1 11 C 2 12 C 3 13 R 1 C t it R i z 11 D 11;12 z 12 D 12;13 z 13 it it it 21 z 21. D 21;12 22 z 22. D 22;13 23 z 23 R 2 D 1(t 1);it. 0 it,1(t+1) 0 it,i 0 (t+1) z it it,1(t+1) it,i 0 (t+1). D it;1(t+1) R 3 D i 0 (t 1);it. 0 it,m(t+1) it,m(t+1). D it;i 0 (t+1) z 31. D 31;12. z 32. D 32,13. z 33. D M(t 1);it D it;m(t+1) Encoding factor: it (z i,t ) = d i,t z i,t Cardinality factor: R i (z i,:) =, kz i,: k 1 > 0 0, otherwise Dynamic factor: D it 1;i 0,t (z i,t 1,z i 0,t) = log p(x i 0 x i )z i,t 1 z i 0,t Constraint factor: C t (z :,t ) = 0, P M i=1 z i,t =1 1, otherwise. E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
27 Synthetic experiments Data from HMM with M = 50 states and T = 100 Dynamic cost kdpp M-kDPP Seq-kDPP Greedy DS3 SeqFL Encoding cost kdpp M-kDPP Seq-kDPP Greedy DS3 SeqFL Number of representatives Number of representatives Sequence of reps in SeqFL: compatible with dynamic model Encoding + *Dynamic cost kdpp M-kDPP Seq-kDPP Greedy DS3 SeqFL Number of representatives
28 Procedure learning We learn how to do tasks using instructional videos How to perform CPR How to change a flat tire How to make a carrot cake How to assemble a bike How to install chrome cast Large # videos
29 Procedure learning We learn how to do tasks using instructional videos How to perform CPR How to change a flat tire How to make a carrot cake How to assemble a bike How to install chrome cast Goal: Learn grammar of complex tasks using video data procedure:
30 Procedure learning via SeqFL 1) Extract superframes (Gygli et al 14) from each video 2) Extract deep features for each superframe 3) Build a dynamic model: 1st order HMM x 7 x 2 CNN- LSTM h (1) h (1) 1 2 h (1) h (1) 3 T 1 x 1 x 4 x 9 x 6 x 8 x 5 CNN- LSTM h (K) 1 h (K) 2 h (K) 3 h (K) T K x 3 E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
31 Procedure learning via SeqFL 1) Extract superframes (Gygli et al 14) from each video 2) Extract deep features for each superframe 3) Build a dynamic model: 1st order HMM 4) SeqFL (X = HMM, Y = test videos) Video 1 x 7 x 2 x 1 x 4 r x x 5 6 x SeqFL 8 x 9 x 3 h (1) h (1) h (1) 1 2 h (1) 3 T 1 (1) 1, r (1) 2, r (1) 3,, r (1) T 1 Video K SeqFL h (K) 1 h (K) 2 h (K) 3 h (K),,,, T K r (K) 1 r (K) 2 r (K) 3 r (K) T K E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
32 Procedure learning via SeqFL 1) Extract superframes (Gygli et al 14) from each video 2) Extract deep features for each superframe 3) Build a dynamic model: 1st order HMM 4) SeqFL (X = HMM, Y = test videos) 5) Align sequences of representatives 6) Map representatives to actions x 7 x 2,,,, r (1) 1 r (1) 2 r (1) 3 r (1) T 1 Video 1 summary x 1 x 9 x 6 x 4 x 8 x 3 x 5,,,, r (K) 1 r (K) 2 r (K) 3 r (K) T K Video K summary x 5! x 2! x 9! x 3! x 8! x 3! x 8 E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
33 Experimental results Instructional video dataset (Alayrac et al CVPR 16) - 5 tasks, 30 videos/task, labels of key steps given No dynamics with dynamics E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
34 How to perform CPR Experimental results SeqFL Give Compression Check Breathing Give Breath Give Compression Give Breath Give Compression Ground Truth Check Response Open Airway Check Breathing Give Breath Give Compression Give Breath Give Compression How to repot a plant SeqFL Put Soil Add Top Loosen Root Place Plant Add Top Ground Truth Put Soil Tap Pot Take Plant Loosen Root Place Plant Add Top E. Elhamifar, C. De Paolis Kaluza, NIPS, 2017
35 Online summarization Sequential data: grow incrementally, collecting all and running SS infeasible Approach: incremental subset selection D (0) n D (1) n D (2) n SS E (1) o SS E (2) o t =0 t =1 t =2 SS [ - Instead of running SS on n, run SS on E o (t) [ D n (t) at - Proved exactness of the online SS t D (t) t E. Elhamifar, C. De Paolis Kaluza, CVPR 17
36 Summary and ongoing work Important to have summarization close to global optimum - Developed efficient sparse optimization for facility location - Showed exactness under appropriate condition Important to handle sequential data: compatible seq of reps - Developed SeqFL for sequential summarization - Addressed procedure learning from instructional videos
37 Thanks! Codes:
HIGH RESOLUTION DEPTH IMAGE RECOVERY ALGORITHM USING GRAYSCALE IMAGE.
HIGH RESOLUTION DEPTH IMAGE RECOVERY ALGORITHM USING GRAYSCALE IMAGE Kazunori Uruma 1, Katsumi Konishi 2, Tomohiro Takahashi 1 and Toshihiro Furukawa 1 1 Graduate School of Engineering, Tokyo University
More informationBiomechanics and Models of Locomotion
Physics-Based Models for People Tracking: Biomechanics and Models of Locomotion Marcus Brubaker 1 Leonid Sigal 1,2 David J Fleet 1 1 University of Toronto 2 Disney Research, Pittsburgh Biomechanics Biomechanics
More informationEE 364B: Wind Farm Layout Optimization via Sequential Convex Programming
EE 364B: Wind Farm Layout Optimization via Sequential Convex Programming Jinkyoo Park 1 Introduction In a wind farm, the wakes formed by upstream wind turbines decrease the power outputs of downstream
More informationEvaluating and Classifying NBA Free Agents
Evaluating and Classifying NBA Free Agents Shanwei Yan In this project, I applied machine learning techniques to perform multiclass classification on free agents by using game statistics, which is useful
More informationMixture Models & EM. Nicholas Ruozzi University of Texas at Dallas. based on the slides of Vibhav Gogate
Mixture Models & EM Nicholas Ruozzi University of Texas at Dallas based on the slides of Vibhav Gogate Previously We looked at -means and hierarchical clustering as mechanisms for unsupervised learning
More informationIntroduction to Pattern Recognition
Introduction to Pattern Recognition Jason Corso SUNY at Buffalo 12 January 2009 J. Corso (SUNY at Buffalo) Introduction to Pattern Recognition 12 January 2009 1 / 28 Pattern Recognition By Example Example:
More informationAttacking and defending neural networks. HU Xiaolin ( 胡晓林 ) Department of Computer Science and Technology Tsinghua University, Beijing, China
Attacking and defending neural networks HU Xiaolin ( 胡晓林 ) Department of Computer Science and Technology Tsinghua University, Beijing, China Outline Background Attacking methods Defending methods 2 AI
More informationknn & Naïve Bayes Hongning Wang
knn & Naïve Bayes Hongning Wang CS@UVa Today s lecture Instance-based classifiers k nearest neighbors Non-parametric learning algorithm Model-based classifiers Naïve Bayes classifier A generative model
More information1.1 The size of the search space Modeling the problem Change over time Constraints... 21
Introduction : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 1 I What Are the Ages of My Three Sons? : : : : : : : : : : : : : : : : : 9 1 Why Are Some Problems Dicult to Solve? : : :
More informationA computer program that improves its performance at some task through experience.
1 A computer program that improves its performance at some task through experience. 2 Example: Learn to Diagnose Patients T: Diagnose tumors from images P: Percent of patients correctly diagnosed E: Pre
More informationPlanning and Acting in Partially Observable Stochastic Domains
Planning and Acting in Partially Observable Stochastic Domains Leslie Pack Kaelbling and Michael L. Littman and Anthony R. Cassandra (1998). Planning and Acting in Partially Observable Stochastic Domains,
More information5th Symposium on Integrating CFD and Experiments in Aerodynamics (Integration 2012) th Symposium on Integrating CFD and Experiments in Aerodynam
5th Symposium on Integrating CFD and Experiments in Aerodynamics (Integration 202) 36 Multi-objective Optimization of Airfoil of Mars Exploration Aircraft using Evolutionary Algorithm Gaku Sasaki Tomoaki
More informationPolynomial DC decompositions
Polynomial DC decompositions Georgina Hall Princeton, ORFE Joint work with Amir Ali Ahmadi Princeton, ORFE 7/31/16 DIMACS Distance geometry workshop 1 Difference of convex (dc) programming Problems of
More informationOptimizing Cyclist Parking in a Closed System
Optimizing Cyclist Parking in a Closed System Letu Qingge, Killian Smith Gianforte School of Computing, Montana State University, Bozeman, MT 59717, USA Abstract. In this paper, we consider the two different
More informationNCSS Statistical Software
Chapter 256 Introduction This procedure computes summary statistics and common non-parametric, single-sample runs tests for a series of n numeric, binary, or categorical data values. For numeric data,
More information2 When Some or All Labels are Missing: The EM Algorithm
CS769 Spring Advanced Natural Language Processing The EM Algorithm Lecturer: Xiaojin Zhu jerryzhu@cs.wisc.edu Given labeled examples (x, y ),..., (x l, y l ), one can build a classifier. If in addition
More informationPrediction Market and Parimutuel Mechanism
Prediction Market and Parimutuel Mechanism Yinyu Ye MS&E and ICME Stanford University Joint work with Agrawal, Peters, So and Wang Math. of Ranking, AIM, 2 Outline World-Cup Betting Example Market for
More informationNaïve Bayes. Robot Image Credit: Viktoriya Sukhanova 123RF.com
Naïve Bayes These slides were assembled by Eric Eaton, with grateful acknowledgement of the many others who made their course materials freely available online. Feel free to reuse or adapt these slides
More informationFun Neural Net Demo Site. CS 188: Artificial Intelligence. N-Layer Neural Network. Multi-class Softmax Σ >0? Deep Learning II
Fun Neural Net Demo Site CS 188: Artificial Intelligence Demo-site: http://playground.tensorflow.org/ Deep Learning II Instructors: Pieter Abbeel & Anca Dragan --- University of California, Berkeley [These
More informationReprocessed QuikSCAT (V04) Wind Vectors with Ku-2011 Geophysical Model Function
Reprocessed QuikSCAT (V04) Wind Vectors with Ku-2011 Geophysical Model Function Lucrezia Ricciardulli and Frank Wentz Introduction In April 2011, we reprocessed the QuikSCAT ocean wind vectors using a
More informationCharacterizing Ireland s wave energy resource
Snapshots of Doctoral Research at University College Cork 2011 Characterizing Ireland s wave energy resource Brendan Cahill Hydraulics & Maritime Research Centre, UCC Introduction In theory, the energy
More informationMinimum Mean-Square Error (MMSE) and Linear MMSE (LMMSE) Estimation
Minimum Mean-Square Error (MMSE) and Linear MMSE (LMMSE) Estimation Outline: MMSE estimation, Linear MMSE (LMMSE) estimation, Geometric formulation of LMMSE estimation and orthogonality principle. Reading:
More informationNaïve Bayes. Robot Image Credit: Viktoriya Sukhanova 123RF.com
Naïve Bayes These slides were assembled by Byron Boots, with only minor modifications from Eric Eaton s slides and grateful acknowledgement to the many others who made their course materials freely available
More informationPhrase-based Image Captioning
Phrase-based Image Captioning Rémi Lebret, Pedro O. Pinheiro, Ronan Collobert Idiap Research Institute / EPFL ICML, 9 July 2015 Image Captioning Objective: Generate descriptive sentences given a sample
More informationBayesian Methods: Naïve Bayes
Bayesian Methods: Naïve Bayes Nicholas Ruozzi University of Texas at Dallas based on the slides of Vibhav Gogate Last Time Parameter learning Learning the parameter of a simple coin flipping model Prior
More informationSMAP Radiometer-Only Tropical Cyclone Size and Strength
SMAP Radiometer-Only Tropical Cyclone Size and Strength Alexander Fore, Simon Yueh, Wenqing Tang, Bryan SCles, Akiko Hayashi Jet Propulsion Laboratory, California InsCtute of Technology 218 California
More informationIntroduction to Pattern Recognition
Introduction to Pattern Recognition Jason Corso SUNY at Buffalo 19 January 2011 J. Corso (SUNY at Buffalo) Introduction to Pattern Recognition 19 January 2011 1 / 32 Examples of Pattern Recognition in
More informationArtificial Intelligence for the EChO Mission Scheduler
Artificial Intelligence for the EChO Mission Scheduler Álvaro García Piquer Ignasi Ribas Josep Colomé Institute of Space Sciences (CSIC/IEEC), Barcelona, Spain SCIOPS 2013 September 10 13, 2013 Introduction
More informationScheduling the Brazilian Soccer Championship. Celso C. Ribeiro* Sebastián Urrutia
Scheduling the Brazilian Soccer Championship Celso C. Ribeiro* Sebastián Urrutia Motivation Problem statement Solution approach Summary Phase 1: create all feasible Ps Phase 2: assign Ps to elite teams
More informationOn the Impact of Garbage Collection on flash-based SSD endurance
On the Impact of Garbage Collection on flash-based SSD endurance Robin Verschoren and Benny Van Houdt Dept. Mathematics and Computer Science University of Antwerp Antwerp, Belgium INFLOW 16 Robin Verschoren
More informationEstimating the Probability of Winning an NFL Game Using Random Forests
Estimating the Probability of Winning an NFL Game Using Random Forests Dale Zimmerman February 17, 2017 2 Brian Burke s NFL win probability metric May be found at www.advancednflstats.com, but the site
More informationImperfectly Shared Randomness in Communication
Imperfectly Shared Randomness in Communication Madhu Sudan Harvard Joint work with Clément Canonne (Columbia), Venkatesan Guruswami (CMU) and Raghu Meka (UCLA). 11/16/2016 UofT: ISR in Communication 1
More informationTHEORY OF TRAINING, THEORETICAL CONSIDERATIONS WOMEN S RACE WALKING
THEORY OF TRAINING, THEORETICAL CONSIDERATIONS WOMEN S RACE WALKING Prof.Corina ȚIFREA Ph.D General characteristics of race walking Sport performance is multiply determined, but we can t definitely settle
More informationi) Linear programming
Problem statement Sailco Corporation must determine how many sailboats should be produced during each of the next four quarters (one quarter = three months). The demand during each of the next four quarters
More informationBuilding a bike friendly Chicago
CASE STUDY Miovision Scout Building a bike friendly Chicago How video data is moving Chicago forward Chicago 2020 Becoming the Most Bike-Friendly City in the United States In 2011, Mayor Rahm Emanuel set
More informationA Bag-of-Gait Model for Gait Recognition
A Bag-of-Gait Model for Gait Recognition Jianzhao Qin, T. Luo, W. Shao, R. H. Y. Chung and K. P. Chow The Department of Computer Science, The University of Hong Kong, Hong Kong, China Abstract In this
More informationLecture 10. Support Vector Machines (cont.)
Lecture 10. Support Vector Machines (cont.) COMP90051 Statistical Machine Learning Semester 2, 2017 Lecturer: Andrey Kan Copyright: University of Melbourne This lecture Soft margin SVM Intuition and problem
More informationObject Recognition. Selim Aksoy. Bilkent University
Image Classification and Object Recognition Selim Aksoy Department of Computer Engineering Bilkent University saksoy@cs.bilkent.edu.tr Image classification Image (scene) classification is a fundamental
More informationEnergy Output. Outline. Characterizing Wind Variability. Characterizing Wind Variability 3/7/2015. for Wind Power Management
Energy Output for Wind Power Management Spring 215 Variability in wind Distribution plotting Mean power of the wind Betz' law Power density Power curves The power coefficient Calculator guide The power
More informationPredicting Horse Racing Results with TensorFlow
Predicting Horse Racing Results with TensorFlow LYU 1703 LIU YIDE WANG ZUOYANG News CUHK Professor, Gu Mingao, wins 50 MILLIONS dividend using his sure-win statistical strategy. News AlphaGO defeats human
More informationCS 7641 A (Machine Learning) Sethuraman K, Parameswaran Raman, Vijay Ramakrishnan
CS 7641 A (Machine Learning) Sethuraman K, Parameswaran Raman, Vijay Ramakrishnan Scenario 1: Team 1 scored 200 runs from their 50 overs, and then Team 2 reaches 146 for the loss of two wickets from their
More informationDESIGN AND ANALYSIS OF ALGORITHMS (DAA 2017)
DESIGN AND ANALYSIS OF ALGORITHMS (DAA 2017) Veli Mäkinen 12/05/2017 1 COURSE STRUCTURE 7 weeks: video lecture -> demo lecture -> study group -> exercise Video lecture: Overview, main concepts, algorithm
More informationOpen Research Online The Open University s repository of research publications and other research outputs
Open Research Online The Open University s repository of research publications and other research outputs Developing an intelligent table tennis umpiring system Conference or Workshop Item How to cite:
More informationSiła-Nowicka, K. (2018) Analysis of Actual Versus Permitted Driving Speed: a Case Study from Glasgow, Scotland. In: 26th Annual GIScience Research UK Conference (GISRUK 2018), Leicester, UK, 17-20 Apr
More informationProduct Decomposition in Supply Chain Planning
Mario R. Eden, Marianthi Ierapetritou and Gavin P. Towler (Editors) Proceedings of the 13 th International Symposium on Process Systems Engineering PSE 2018 July 1-5, 2018, San Diego, California, USA 2018
More informationAppendix A Bridge Deck Section Images
Appendix A Bridge Deck Section Images 127 128 Figure A1. Bridge deck R13, fabricated at the FHWA NDE Validation Center. Figure A2. Bridge deck R12, fabricated at the FHWA NDE Validation Center. 129 Figure
More informationRecognition of Tennis Strokes using Key Postures
ISSC 2010, UCC, Cork, June 23 24 Recognition of Tennis Strokes using Key Postures Damien Connaghan, Ciarán Ó Conaire, Philip Kelly, Noel E. O Connor CLARITY: Centre for Sensor Web Technologies Dublin City
More informationParsimonious Linear Fingerprinting for Time Series
Parsimonious Linear Fingerprinting for Time Series Lei Li, B. Aditya Prakash, Christos Faloutsos School of Computer Science Carnegie Mellon University VLDB 2010 1 L. Li, 2010 VLDB2010, 36 th International
More informationTIMETABLING IN SPORTS AND ENTERTAINMENT
TIMETABLING IN SPORTS AND ENTERTAINMENT Introduction More complicated applications of interval scheduling and timetabling: scheduling of games in tournaments scheduling of commercials on network television
More information[CROSS COUNTRY SCORING]
2015 The Race Director Guide [CROSS COUNTRY SCORING] This document describes the setup and scoring processes employed when scoring a cross country race with Race Director. Contents Intro... 3 Division
More informationUsing Spatio-Temporal Data To Create A Shot Probability Model
Using Spatio-Temporal Data To Create A Shot Probability Model Eli Shayer, Ankit Goyal, Younes Bensouda Mourri June 2, 2016 1 Introduction Basketball is an invasion sport, which means that players move
More informationLinear Compressor Suction Valve Optimization
Purdue University Purdue e-pubs International Compressor Engineering Conference School of Mechanical Engineering 2016 Linear Compressor Suction Valve Optimization Rinaldo Puff Embraco, Brazil, rinaldo.puff@embraco.com
More informationNYC Marathon Quarter 1 Review Task. The New York City Marathon was last weekend. Over 50,000 people ran 26.2 miles around the city!
NYC Marathon Quarter 1 Review Task Name: Alg: The New York City Marathon was last weekend. Over 50,000 people ran 26.2 miles around the city! 1. The 2015 male winner was Stanely Biwott, his time was 2
More informationPlayer Availability Rating (PAR) - A Tool for Quantifying Skater Performance for NHL General Managers
Player Availability Rating (PAR) - A Tool for Quantifying Skater Performance for NHL General Managers Shuja Khalid 1 1 Department of Computer Science, University of Toronto arxiv:1811.02885v1 [cs.cy] 15
More informationUniversal Style Transfer via Feature Transforms
Universal Style Transfer via Feature Transforms Yijun Li, Chen Fang, Jimei Yang, Zhaowen Wang, Xin Lu, Ming-Hsuan Yang UC Merced, Adobe Research, NVIDIA Research Presented: Dong Wang (Refer to slides by
More informationDATA MINING SAMPLE RESEARCH: ACTIVITY RECOGNITION CLASSIFICATION IN ACTION
DATA MINING SAMPLE RESEARCH: ACTIVITY RECOGNITION CLASSIFICATION IN ACTION 1 Mobile Activity Recognition Mobile devices like smartphones and smartwatches have many sensors Some sensors measure motion Tri-axial
More informationBASKETBALL PREDICTION ANALYSIS OF MARCH MADNESS GAMES CHRIS TSENG YIBO WANG
BASKETBALL PREDICTION ANALYSIS OF MARCH MADNESS GAMES CHRIS TSENG YIBO WANG GOAL OF PROJECT The goal is to predict the winners between college men s basketball teams competing in the 2018 (NCAA) s March
More informationPerformance of Fully Automated 3D Cracking Survey with Pixel Accuracy based on Deep Learning
Performance of Fully Automated 3D Cracking Survey with Pixel Accuracy based on Deep Learning Kelvin C.P. Wang Oklahoma State University and WayLink Systems Corp. 2017-10-19, Copenhagen, Denmark European
More informationMarathon Performance Prediction of Amateur Runners based on Training Session Data
Marathon Performance Prediction of Amateur Runners based on Training Session Data Daniel Ruiz-Mayo, Estrella Pulido, and Gonzalo Martínez-Muñoz Escuela Politécnica Superior, Universidad Autónoma de Madrid,
More informationSEA SURFACE TEMPERATURE RETRIEVAL USING TRMM MICROWAVE IMAGER SATELLITE DATA IN THE SOUTH CHINA SEA
SEA SURFACE TEMPERATURE RETRIEVAL USING TRMM MICROWAVE IMAGER SATELLITE DATA IN THE SOUTH CHINA SEA Mohd Ibrahim Seeni Mohd and Mohd Nadzri Md. Reba Faculty of Geoinformation Science and Engineering Universiti
More informationComparison of the Dispersant Mission Planner 2 (DMP2) and Estimated Dispersant System Potential (EDSP) Calculators
Comparison of the Dispersant Mission Planner 2 (DMP2) and Estimated Dispersant System Potential (EDSP) Calculators The DMP2 was developed by Genwest Systems, Inc. in 2006 for the NOAA Office of Response
More informationDiscriminative Feature Selection for Uncertain Graph Classification
Discriminative Feature Selection for Uncertain Graph Classification Xiangnan Kong University of Illinois at Chicago joint work with Philip S. Yu (Univ. Illinois at Chicago) Xue Wang & Ann B. Ragin (Northwestern
More informationPart 1: Decimals. The decimal point separates the whole numbers from the fractional part of a
Part 1: Decimals Decimal Place Values The decimal point separates the whole numbers from the fractional part of a number. 1328. 1095 In a whole number the decimal point is all the way to the right, even
More informationLogistic Regression. Hongning Wang
Logistic Regression Hongning Wang CS@UVa Today s lecture Logistic regression model A discriminative classification model Two different perspectives to derive the model Parameter estimation CS@UVa CS 6501:
More informationInferring land use from mobile phone activity
Inferring land use from mobile phone activity Jameson L. Toole (MIT) Michael Ulm (AIT) Dietmar Bauer (AIT) Marta C. Gonzalez (MIT) UrbComp 2012 Beijing, China 1 The Big Questions Can land use be predicted
More informationDecision Trees. Nicholas Ruozzi University of Texas at Dallas. Based on the slides of Vibhav Gogate and David Sontag
Decision Trees Nicholas Ruozzi University of Texas at Dallas Based on the slides of Vibhav Gogate and David Sontag Announcements Course TA: Hao Xiong Office hours: Friday 2pm-4pm in ECSS2.104A1 First homework
More informationFast Floating Point Compression on the Cell BE Processor
Fast Floating Point Compression on the Cell BE Processor Ajith Padyana, T.V. Siva Kumar, P.K.Baruah Sri Satya Sai University Prasanthi Nilayam - 515134 Andhra Pradhesh, India ajith.padyana@gmail.com, tvsivakumar@gmail.com,
More informationTHe rip currents are very fast moving narrow channels,
1 Rip Current Detection using Optical Flow Shweta Philip sphilip@ucsc.edu Abstract Rip currents are narrow currents of fast moving water that are strongest near the beach. These type of currents are dangerous
More informationUpdate June 2018 OUR 2017 PUBLIC ENGAGEMENT
Update June 2018 Hi, Let s Get Wellington Moving (LGWM) is developing a Recommended Programme of Investment (RPI) that meets our programme objectives and supports the city s and region s growth. As part
More informationA IMPROVED VOGEL S APPROXIMATIO METHOD FOR THE TRA SPORTATIO PROBLEM. Serdar Korukoğlu 1 and Serkan Ballı 2.
Mathematical and Computational Applications, Vol. 16, No. 2, pp. 370-381, 2011. Association for Scientific Research A IMPROVED VOGEL S APPROXIMATIO METHOD FOR THE TRA SPORTATIO PROBLEM Serdar Korukoğlu
More informationA Game Theoretic Study of Attack and Defense in Cyber-Physical Systems
The First International Workshop on Cyber-Physical Networking Systems A Game Theoretic Study of Attack and Defense in Cyber-Physical Systems ChrisY.T.Ma Advanced Digital Sciences Center Illinois at Singapore
More informationWeatherTRAK ET plus Series Quick Reference Guide
For 6, 9, 12, 18 and 24 station models plus Quick Start Guide WeatherTRAK plus Series Quick Reference Guide 3. 4. 5. 6. Upper line knob. Controls the upper line of the display. Function knob. Provides
More informationSpatio-temporal analysis of team sports Joachim Gudmundsson
Spatio-temporal analysis of team sports Joachim Gudmundsson The University of Sydney Page 1 Team sport analysis Talk is partly based on: Joachim Gudmundsson and Michael Horton Spatio-Temporal Analysis
More informationKathleen Dohan. Wind-Driven Surface Currents. Earth and Space Research, Seattle, WA
Updates to OSCAR and challenges with capturing the wind-driven currents. Wind-Driven Surface Currents Kathleen Dohan Earth and Space Research, Seattle, WA ENSO OSCAR Surface currents from satellite fields
More informationMobility Detection Using Everyday GSM Traces
Mobility Detection Using Everyday GSM Traces Timothy Sohn et al Philip Cootey pcootey@wpi.edu (03/22/2011) Mobility Detection High level activity discerned from course Grained GSM Data provides immediate
More informationAssessing Compliance with United States Government Orbital Debris Mitigation Guidelines
Assessing Compliance with United States Government Orbital Debris Mitigation Guidelines R. L. Kelley 1, D. R. Jarkey 2 1. Jacobs, NASA Johnson Space Center, Houston, TX 77058, USA 2. HX5 - Jacobs JETS
More informationSmart Water Application Technologies (SWAT)
Smart Water Application Technologies (SWAT) Turf and Landscape Irrigation Equipment PRESSURE REGULATING SPRAY HEAD SPRINKLERS Equipment Functionality Test Testing Protocol Version 3.0 (May 2012) Developed
More informationTime and synchronization
Time and synchronization ( There s never enough time ) Today s outline Global Time Time in distributed systems A baseball example Synchronizing real clocks Cristian s algorithm The Berkeley Algorithm Network
More informationTime and synchronization. ( There s never enough time )
Time and synchronization ( There s never enough time ) Today s outline Global Time Time in distributed systems A baseball example Synchronizing real clocks Cristian s algorithm The Berkeley Algorithm Network
More informationApplication of Bayesian Networks to Shopping Assistance
Application of Bayesian Networks to Shopping Assistance Yang Xiang, Chenwen Ye, and Deborah Ann Stacey University of Guelph, CANADA Abstract. We develop an on-line shopping assistant that can help a e-shopper
More informationOptimizing positional scoring rules for rank aggregation
Optimizing positional scoring rules for rank aggregation Ioannis Caragiannis Xenophon Chatzigeorgiou George Krimpas Alexandros Voudouris University of Patras AAAI 2017 An application Find a pizza place
More informationLOCOMOTION CONTROL CYCLES ADAPTED FOR DISABILITIES IN HEXAPOD ROBOTS
LOCOMOTION CONTROL CYCLES ADAPTED FOR DISABILITIES IN HEXAPOD ROBOTS GARY B. PARKER and INGO CYLIAX Department of Computer Science, Indiana University, Bloomington, IN 47405 gaparker@cs.indiana.edu, cyliax@cs.indiana.edu
More informationCOST ESTIMATION. Fully Allocated Causal Factor Models Temporal Variation Models Incremental Fixed Variable Cost Models
COST ESTIMATION Outline 1. Roles for Cost Models 2. Conventional Model Types Fully Allocated Causal Factor Models Temporal Variation Models Incremental Fixed Variable Cost Models 3. New Approaches HASTUS
More informationNext Generation Quartz Pressure Gauges
Next Generation Quartz Pressure Gauges Rick Puccio, Ph.D. Senior Scientist Quartzdyne Houston, TX March 2, 2016 Outline Quartz Pressure Gauge Introduction Gauge Size Reduction Long Term Pressure Drift
More informationChapter 20. Planning Accelerated Life Tests. William Q. Meeker and Luis A. Escobar Iowa State University and Louisiana State University
Chapter 20 Planning Accelerated Life Tests William Q. Meeker and Luis A. Escobar Iowa State University and Louisiana State University Copyright 1998-2008 W. Q. Meeker and L. A. Escobar. Based on the authors
More information[CROSS COUNTRY SCORING]
2018 The Race Director Guide [CROSS COUNTRY SCORING] This document describes the setup and scoring processes employed when scoring a cross country race with Race Director. Contents Intro... 3 Division
More informationSIDT 2017 XXII SEMINARIO SCIENTIFICO DELLA SOCIETÀ ITALIANA DOCENTI DI TRASPORTI
SIDT 2017 XXII SEMINARIO SCIENTIFICO DELLA SOCIETÀ ITALIANA DOCENTI DI TRASPORTI POLITECNICO DI BARI TOWARDS SUSTAINABLE CITIES: INNOVATIVE SOLUTIONS FOR MOBILITY AND LOGISTICS SEPTEMBER 14-15, 2017 SESSION
More informationAn approach for optimising railway traffic flow on high speed lines with differing signalling systems
Computers in Railways XIII 27 An approach for optimising railway traffic flow on high speed lines with differing signalling systems N. Zhao, C. Roberts & S. Hillmansen Birmingham Centre for Railway Research
More informationA Chiller Control Algorithm for Multiple Variablespeed Centrifugal Compressors
Purdue University Purdue e-pubs International Compressor Engineering Conference School of Mechanical Engineering 2014 A Chiller Control Algorithm for Multiple Variablespeed Centrifugal Compressors Piero
More informationLegendre et al Appendices and Supplements, p. 1
Legendre et al. 2010 Appendices and Supplements, p. 1 Appendices and Supplement to: Legendre, P., M. De Cáceres, and D. Borcard. 2010. Community surveys through space and time: testing the space-time interaction
More informationANG, a family of multi-mode,
ANG, a family of multi-mode, low cost walking aid J-P. Merlet COPRIN Project team INRIA 1/9. p.1/6 Motivation 2/9. p.2/6 Motivation rapid increase of the number of elderly 2/9. p.2/6 Motivation rapid increase
More informationComputing s Energy Problem:
Computing s Energy Problem: (and what we can do about it) Mark Horowitz Stanford University horowitz@ee.stanford.edu 1 of 46 Everything Has A Computer Inside 2of 46 The Reason is Simple: Moore s Law Made
More informationREGIONAL MULTIMODAL TRANSIT ACCESS
www.rtachicago.org REGIONAL MULTIMODAL TRANSIT ACCESS GIS in Transit Conference September 3, 2015 Will Gillespie Principal Analyst, Strategic Planning THE RTA Regional financial & budgetary oversight,
More informationPaul Burkhardt. May 19, 2016
GraphEx Symposium 2016 U.S. National Security Agency Research Directorate May 19, 2016 Graphs are everywhere! Why graph? Graph-theoretic approaches are appealing to many fields simple data abstraction
More informationGUIDE TO RUNNING A BIKE SHARE. h o w t o p l a n a n d o p e r a t e a s u c c e s s f u l b i k e s h a r e p r o g r a m
GUIDE TO RUNNING A BIKE SHARE h o w t o p l a n a n d o p e r a t e a s u c c e s s f u l b i k e s h a r e p r o g r a m 20150209 The bicycle is the most loved form of transportation. No other machine
More informationDistributed Systems [Fall 2013]
Distributed Systems [Fall 2013] Lec 7: Time and Synchronization Slide acks: Dave Andersen, Randy Bryant (http://www.cs.cmu.edu/~dga/15-440/f11/lectures/09-time+synch.pdf) 1 Any Questions for HW 2? Deadline
More informationThe Incremental Evolution of Gaits for Hexapod Robots
The Incremental Evolution of Gaits for Hexapod Robots Abstract Gait control programs for hexapod robots are learned by incremental evolution. The first increment is used to learn the activations required
More informationMarch Madness Basketball Tournament
March Madness Basketball Tournament Math Project COMMON Core Aligned Decimals, Fractions, Percents, Probability, Rates, Algebra, Word Problems, and more! To Use: -Print out all the worksheets. -Introduce
More informationSelecting Flow Metering Sites And Evaluating Data
Insert Slide Title Here Selecting Flow Metering Sites And Evaluating Data Patrick L. Stevens, PE Luis Mijares PStevens2@Idexcorp.com LMijares@Idexcorp.com ADS Environmental Services Strategy for Selecting
More informationBiostatistics & SAS programming
Biostatistics & SAS programming Kevin Zhang March 6, 2017 ANOVA 1 Two groups only Independent groups T test Comparison One subject belongs to only one groups and observed only once Thus the observations
More information