Attacking and defending neural networks. HU Xiaolin ( 胡晓林 ) Department of Computer Science and Technology Tsinghua University, Beijing, China

Size: px
Start display at page:

Download "Attacking and defending neural networks. HU Xiaolin ( 胡晓林 ) Department of Computer Science and Technology Tsinghua University, Beijing, China"

Transcription

1 Attacking and defending neural networks HU Xiaolin ( 胡晓林 ) Department of Computer Science and Technology Tsinghua University, Beijing, China

2 Outline Background Attacking methods Defending methods 2

3 AI is booming US and China have huge AI plans United Kingdom plans $1.3 billion AI push France spends $1.8 billion on AI to compete with US and China EU wants to invest 18 billion pounds in AI development 3

4 Neural networks are powerful Many powerful models Resnet (He et al., 2016) Inception V1 (Szegedy et al., 2014) Inception V2 (Szegedy et al., 2016) Inception V3 (Szegedy et al., 2016) DenseNet (Huang et al., 2017) Most of them achieve higher accuracy than humans on the ImageNet classification task 4

5 but Unreliable! legitimate Examples Adversarial Examples Crafted specifically to fool ResNet 5

6 Black-box Attacks (Transferability) Cross-model transferability ResNet Inc V3 dog dog Cross-data transferability 6

7 Physically realizable attack Sharif et al., CCS 2016 Kaylee Defer Nancy Travis 7

8 Physical attack Athalye et al., ICLR

9 Definitions Given a classifier ff xx : xx XX yy YY, which maps the input sample xx to the label yy. Add a small noise δδ to xx and we have xx = xx + δδ. If ff xx yy, then xx is called an adversarial example, and xx is called the normal sample or legitimate sample. δδ is called adversarial perturbation, whose magnitude is often limited such as δδ pp εε, where pp could be 0, 1, 2 or, εε is a small constant If ff xx = yy, where yy is a specified label, then this attack is called targeted attack If ff xx is not specified to any class, then this attack is called untargeted attack 9

10 Definitions Given a classifier ff xx : xx XX yy YY, which maps the input sample xx to the label yy. Add a small noise δδ to xx and we have xx = xx + δδ. Attack a certain model: perturb xx to xx and make the model misclassify xx, i.e., ff xx yy White-box attack: the model is known Black-box attack: the model is unknown Defend a certain model: make the model map xx to yy 10

11 Outline Background Attacking methods Defending methods Interpretability Summary 11

12 Principles for attacking a classification model Recall: Learning a neural network amounts to minimizing a loss function LL (often cross entropy + regularizer) w.r.t. the weights and biases min ww,bb LL(xx, yy; ww, bb) where (xx, yy) denotes the input and desired output pair Attacking: maximizing the above loss function w.r.t. the input max LL(xx, yy; ww, bb) xx subject to xx xx < δδ pp 12

13 Attacking methods One-step FGSM (linearity of decision boundary) xx = xx + εε sign( xx LL(xx, yy)) Poor white-box attack ability, good black-box attack ability Iterative FGSM (I-FGSM) xx 0 = xx, xx tt+1 = clip(xx tt + α sign xx LL xx tt, yy ) Optimization-based methods min dd xx, xx LL(xx, yy) Poor white-box attack ability, good black-box attack ability 13

14 Optimization with Momentum Constrained optimization of adversarial attacks: argmax xx LL xx, yy ss. tt. xx xx εε Optimization algorithm with momentum Accelerate gradient descent; Escape from poor local minima and maxima; Stabilize update directions of stochastic gradient descent; Momentum can be used for adversarial attacks The white-box attack ability and black-box attack ability are both strong (transferability) Dong, Liao, Pang, Su, Zhu, Hu, Li, Boosting Adversarial Attacks with Momentum, CVPR

15 Momentum Iterative FGSM xx 0 = xx, xx tt+1 = clip(xx tt + α sign xx LL xx tt, yy ) xx 0 = xx, gg 0 = 0 gg tt+1 = μμ gg tt + xxll xx tt, yy xx LL xx tt, yy 1 = clip(xx tt + αα sign gg tt+1 ) xx tt+1 Momentum μμ is the decay factor; gg tt accumulates the gradient w.r.t. input space of the first tt iterations; The current gradient is normalized. 15

16 Non-targeted Results εε = 16, μμ = 1.0, 10 iterations 16

17 Attacking an Ensemble of Models If an adversarial example remains adversarial for multiple models, it is more likely to be misclassified by other black-box models. Ensemble in logits ll xx = KK ii=1 ww ii ll ii (xx) The loss is defined as JJ xx, yy = 1 yy log(softmax(ll(xx))) Comparisons: Ensemble in predictions: pp xx = KK ii=1 ww ii pp ii (xx) Ensemble in loss: JJ xx, yy = KK ii=1 ww ii JJ ii (xx, yy) 17

18 Non-targeted Results (2) A total of 7 models: in each row, ensemble of 6 models and test on one model as indicated in the 1 st column 18

19 NIPS 2017 Competition Three tracks: Non-targeted Adversarial Attack 1 st Targeted Adversarial Attack 1 st Defense Against Adversarial Attack 1 st Evaluation: Given a dataset (5000images ImageNet Compatible) Score aaaaaaaaaaaa = ddeeeeeeeeeeee NN kk=1 [defense(attack(xx kk )) yy kk ] Score tttttttttttt = ddeeeeeeeeeeee NN kk=1 [defense target xx kk = yy kk ] Score dddddddddddddd = aaaaaaaaaaaa NN kk=1 [defense attack xx kk = yy kk ] Requirement: 4 εε 16, based on LL norm. Running time 500ss for 100 images 19

20 Outline Background Attacking methods Defending methods Interpretability Summary 22

21 Defending methods Obfuscated gradient methods Input transformations (e.g., JPEG compression) (Guo et al. 2018) Thermometer encoding (Buckman et al., 2018) Local intrinsic dimensionality (LID) (Ma et al., 2018) Stochastic activation pruning (Dhillon et al.,2018) Etc. Not powerful enough (Athalye, et al., ICML 2018 Best paper) Cannot defend black-box attack well 23

22 Defending black-box attacks Adversarial training (Kurakin, Goodfellow, Bengio, 2017) Add adversarial examples into the training set Computationally expensive Ensemble adversarial training (Tramèr et al., 2018) Use more than one model for adversarial training Even more computationally expensive Worked the best till our method HGD (Liao et al., 2018) 24

23 Motivation for denoising Misclassification: xx + dddd yy After denoising: xx + dddd xx yy Adv img Est. clean img 25

24 Neural networks for denoising Denoising Autoencoder Denoising U-Net Pixel Guided Denoiser (PGD) L1 dd xx L1 xx xx xx xx + xx xx Liao, Liang, Dong, Pang, Hu, Zhu, Defense against Adversarial Attacks Using High-Level Representation Guided Denoiser, CVPR

25 Architecture of the DU-Net 27

26 Experiment setting Attack methods FGSM attack IFGSM attack Training set 30K source images Test set for whitebox attack 10K source images Test set for blackbox attack 10K source images 28

27 Results DAE is poor for reconstructing large images DU-NET removes more noise for white-box attack, but the accuracy was lower than DAE in white box attack Why? 29

28 Error amplification Network Image Clean Im Adv Im Let s construct a loss in higher layers! 30

29 Proposed schemes Train Test Naive CNN Adv Im Pixel Guided Denoiser (PGD) Adv Im D Den Im L1 Clean Im L1 Adv Im D Den Im CNN High-level Representation Guided Denoiser (HGD) Adv Im D Den Im Feat1 CNN Feat2 CNN Clean Im Adv Im D Den Im CNN 31

30 Variants of HGD Feature guided denoiser FGD Logits guided denoiser LGD Class label guided denoiser CGD 32

31 Robustness of HGD 33

32 Transferability of HGD L1 Feat1 Feat2 D CNN1 CNN1 D CNN2 Adv Im Den Im Clean Im Adv Im Den Im The two CNNs for training and testing could be different CNN1 The target model CNN2 is Resnet 34

33 NIPS 2017 Competition Three tracks: Non-targeted Adversarial Attack 1 st Targeted Adversarial Attack 1 st Defense Against Adversarial Attack 1 st Evaluation: Given a dataset (5000images ImageNet Compatible) Score aaaaaaaaaaaa = ddeeeeeeeeeeee NN kk=1 [defense(attack(xx kk )) yy kk ] Score tttttttttttt = ddeeeeeeeeeeee NN kk=1 [defense target xx kk = yy kk ] Score dddddddddddddd = aaaaaaaaaaaa NN kk=1 [defense attack xx kk = yy kk ] Requirement: 4 εε 16, based on LL norm. Running time 500ss for 100 images 35

34 Summary and discussion Current deep learning models are not robust If DL is more widely used, the risk of adversarial attack will be higher One attacking method Momentum IFGSM is presented One defending methods High-level representation guided denoiser (HGD) is presented Defending techniques are not effective if they are known to the attacker HGD can be also fooled Human is much more robust to adversarial examples Brain-inspired computing is promising 37

35 Open source Momentum IFGSM Attacks Attack HGD 38

36 Q & A

Logistic Regression. Hongning Wang

Logistic 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 information

Introduction to Machine Learning NPFL 054

Introduction to Machine Learning NPFL 054 Introduction to Machine Learning NPFL 054 http://ufal.mff.cuni.cz/course/npfl054 Barbora Hladká hladka@ufal.mff.cuni.cz Martin Holub holub@ufal.mff.cuni.cz Charles University, Faculty of Mathematics and

More information

Special Topics: Data Science

Special Topics: Data Science Special Topics: Data Science L Linear Methods for Prediction Dr. Vidhyasaharan Sethu School of Electrical Engineering & Telecommunications University of New South Wales Sydney, Australia V. Sethu 1 Topics

More information

Lecture 5. Optimisation. Regularisation

Lecture 5. Optimisation. Regularisation Lecture 5. Optimisation. Regularisation COMP90051 Statistical Machine Learning Semester 2, 2017 Lecturer: Andrey Kan Copyright: University of Melbourne Iterative optimisation Loss functions Coordinate

More information

Bayesian Methods: Naïve Bayes

Bayesian 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 information

CS249: ADVANCED DATA MINING

CS249: ADVANCED DATA MINING CS249: ADVANCED DATA MINING Linear Regression, Logistic Regression, and GLMs Instructor: Yizhou Sun yzsun@cs.ucla.edu April 24, 2017 About WWW2017 Conference 2 Turing Award Winner Sir Tim Berners-Lee 3

More information

Communication Amid Uncertainty

Communication Amid Uncertainty Communication Amid Uncertainty Madhu Sudan Harvard University Based on joint works with Brendan Juba, Oded Goldreich, Adam Kalai, Sanjeev Khanna, Elad Haramaty, Jacob Leshno, Clement Canonne, Venkatesan

More information

knn & Naïve Bayes Hongning Wang

knn & 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 information

Lecture 39: Training Neural Networks (Cont d)

Lecture 39: Training Neural Networks (Cont d) Lecture 39: Training Neural Networks (Cont d) CS 4670/5670 Sean Bell Strawberry Goblet Throne (Side Note for PA5) AlexNet: 1 vs 2 parts Caffe represents caffe like the above image, but computes as if it

More information

Deformable Convolutional Networks

Deformable Convolutional Networks Deformable Convolutional Networks -- MSRA COCO Detection & Segmentation Challenge 2017 Entry Jifeng Dai With Haozhi Qi*, Zheng Zhang, Bin Xiao, Han Hu, Bowen Cheng*, Yichen Wei Visual Computing Group Microsoft

More information

Practical Approach to Evacuation Planning Via Network Flow and Deep Learning

Practical Approach to Evacuation Planning Via Network Flow and Deep Learning Practical Approach to Evacuation Planning Via Network Flow and Deep Learning Akira Tanaka Nozomi Hata Nariaki Tateiwa Katsuki Fujisawa Graduate School of Mathematics, Kyushu University Institute of Mathematics

More information

Support Vector Machines: Optimization of Decision Making. Christopher Katinas March 10, 2016

Support Vector Machines: Optimization of Decision Making. Christopher Katinas March 10, 2016 Support Vector Machines: Optimization of Decision Making Christopher Katinas March 10, 2016 Overview Background of Support Vector Machines Segregation Functions/Problem Statement Methodology Training/Testing

More information

Lecture 10. Support Vector Machines (cont.)

Lecture 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 information

Imperfectly Shared Randomness in Communication

Imperfectly 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 information

Communication Amid Uncertainty

Communication Amid Uncertainty Communication Amid Uncertainty Madhu Sudan Harvard University Based on joint works with Brendan Juba, Oded Goldreich, Adam Kalai, Sanjeev Khanna, Elad Haramaty, Jacob Leshno, Clement Canonne, Venkatesan

More information

Fun 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. 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 information

CS145: INTRODUCTION TO DATA MINING

CS145: INTRODUCTION TO DATA MINING CS145: INTRODUCTION TO DATA MINING 3: Vector Data: Logistic Regression Instructor: Yizhou Sun yzsun@cs.ucla.edu October 9, 2017 Methods to Learn Vector Data Set Data Sequence Data Text Data Classification

More information

Convolutional Neural Networks

Convolutional Neural Networks CS 1674: Intro to Computer Vision Convolutional Neural Networks Prof. Adriana Kovashka University of Pittsburgh March 13, 15, 20, 2018 Plan for the next few lectures Why (convolutional) neural networks?

More information

Universal Style Transfer via Feature Transforms

Universal 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 information

Visualizing and Understanding Stochastic Depth Networks

Visualizing and Understanding Stochastic Depth Networks Visualizing and Understanding Stochastic Depth Networks Russell Kaplan, Raphael Palefsky-Smith, Liu Jiang Stanford University 450 Serra Mall, Stanford, CA 94305 {rjkaplan, rpalefsk, liujiang}@stanford.edu

More information

JPEG-Compatibility Steganalysis Using Block-Histogram of Recompression Artifacts

JPEG-Compatibility Steganalysis Using Block-Histogram of Recompression Artifacts JPEG-Compatibility Steganalysis Using Block-Histogram of Recompression Artifacts Jan Kodovský, Jessica Fridrich May 16, 2012 / IH Conference 1 / 19 What is JPEG-compatibility steganalysis? Detects embedding

More information

Decision 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 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 information

Physical Design of CMOS Integrated Circuits

Physical Design of CMOS Integrated Circuits Physical Design of CMOS Integrated Circuits Dae Hyun Kim EECS Washington State University References John P. Uyemura, Introduction to VLSI Circuits and Systems, 2002. Chapter 5 Goal Understand how to physically

More information

Coupling distributed and symbolic execution for natural language queries. Lili Mou, Zhengdong Lu, Hang Li, Zhi Jin

Coupling distributed and symbolic execution for natural language queries. Lili Mou, Zhengdong Lu, Hang Li, Zhi Jin Coupling distributed and symbolic execution for natural language queries Lili Mou, Zhengdong Lu, Hang Li, Zhi Jin doublepower.mou@gmail.com Outline Introduction to neural enquirers Coupled approach of

More information

Operational Risk Management: Preventive vs. Corrective Control

Operational Risk Management: Preventive vs. Corrective Control Operational Risk Management: Preventive vs. Corrective Control Yuqian Xu (UIUC) July 2018 Joint Work with Lingjiong Zhu and Michael Pinedo 1 Research Questions How to manage operational risk? How does

More information

Basketball field goal percentage prediction model research and application based on BP neural network

Basketball field goal percentage prediction model research and application based on BP neural network ISSN : 0974-7435 Volume 10 Issue 4 BTAIJ, 10(4), 2014 [819-823] Basketball field goal percentage prediction model research and application based on BP neural network Jijun Guo Department of Physical Education,

More information

Mixture 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 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 information

Course 495: Advanced Statistical Machine Learning/Pattern Recognition

Course 495: Advanced Statistical Machine Learning/Pattern Recognition Course 495: Advanced Statistical Machine Learning/Pattern Recognition Lectures: Stefanos Zafeiriou Goal (Lectures): To present modern statistical machine learning/pattern recognition algorithms. The course

More information

Machine Learning Application in Aviation Safety

Machine Learning Application in Aviation Safety Machine Learning Application in Aviation Safety Surface Safety Metric MOR Classification Presented to: By: Date: ART Firdu Bati, PhD, FAA September, 2018 Agenda Surface Safety Metric (SSM) development

More information

Jasmin Smajic 1, Christian Hafner 2, Jürg Leuthold 2, March 16, 2015 Introduction to Finite Element Method (FEM) Part 1 (2-D FEM)

Jasmin Smajic 1, Christian Hafner 2, Jürg Leuthold 2, March 16, 2015 Introduction to Finite Element Method (FEM) Part 1 (2-D FEM) Jasmin Smajic 1, Christian Hafner 2, Jürg Leuthold 2, March 16, 2015 Introduction to Finite Element Method (FEM) Part 1 (2-D FEM) 1 HSR - University of Applied Sciences of Eastern Switzerland Institute

More information

CT4510: Computer Graphics. Transformation BOCHANG MOON

CT4510: Computer Graphics. Transformation BOCHANG MOON CT4510: Computer Graphics Transformation BOCHANG MOON 2D Translation Transformations such as rotation and scale can be represented using a matrix M ee. gg., MM = SSSS xx = mm 11 xx + mm 12 yy yy = mm 21

More information

EE582 Physical Design Automation of VLSI Circuits and Systems

EE582 Physical Design Automation of VLSI Circuits and Systems EE Prof. Dae Hyun Kim School of Electrical Engineering and Computer Science Washington State University Routing Grid Routing Grid Routing Grid Routing Grid Routing Grid Routing Lee s algorithm (Maze routing)

More information

CAAD CTF 2018 Rules June 21, 2018 Version 1.1

CAAD CTF 2018 Rules June 21, 2018 Version 1.1 CAAD CTF 2018 Rules June 21, 2018 Version 1.1 The organizer will invite 5 teams to participate CAAD CTF 2018. We will have it in Las Vegas on Aug. 10 th, 2018. The rules details are below: 1. Each team

More information

Performance 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 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 information

Pre-Kindergarten 2017 Summer Packet. Robert F Woodall Elementary

Pre-Kindergarten 2017 Summer Packet. Robert F Woodall Elementary Pre-Kindergarten 2017 Summer Packet Robert F Woodall Elementary In the fall, on your child s testing day, please bring this packet back for a special reward that will be awarded to your child for completion

More information

Neural Nets Using Backpropagation. Chris Marriott Ryan Shirley CJ Baker Thomas Tannahill

Neural Nets Using Backpropagation. Chris Marriott Ryan Shirley CJ Baker Thomas Tannahill Neural Nets Using Backpropagation Chris Marriott Ryan Shirley CJ Baker Thomas Tannahill Agenda Review of Neural Nets and Backpropagation Backpropagation: The Math Advantages and Disadvantages of Gradient

More information

CS 1675: Intro to Machine Learning. Neural Networks. Prof. Adriana Kovashka University of Pittsburgh November 1, 2018

CS 1675: Intro to Machine Learning. Neural Networks. Prof. Adriana Kovashka University of Pittsburgh November 1, 2018 CS 1675: Intro to Machine Learning Neural Networks Prof. Adriana Kovashka University of Pittsburgh November 1, 2018 Plan for this lecture Neural network basics Definition and architecture Biological inspiration

More information

Introduction to Pattern Recognition

Introduction 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 information

POKEMON HACKS. Jodie Ashford Josh Baggott Chloe Barnes Jordan bird

POKEMON HACKS. Jodie Ashford Josh Baggott Chloe Barnes Jordan bird POKEMON HACKS Jodie Ashford Josh Baggott Chloe Barnes Jordan bird Why pokemon? 1997 The Pokemon Problem an episode of the anime caused 685 children to have seizures Professor Graham Harding from Aston

More information

ECO 745: Theory of International Economics. Jack Rossbach Fall Lecture 6

ECO 745: Theory of International Economics. Jack Rossbach Fall Lecture 6 ECO 745: Theory of International Economics Jack Rossbach Fall 2015 - Lecture 6 Review We ve covered several models of trade, but the empirics have been mixed Difficulties identifying goods with a technological

More information

THe rip currents are very fast moving narrow channels,

THe 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 information

A Novel Approach to Predicting the Results of NBA Matches

A Novel Approach to Predicting the Results of NBA Matches A Novel Approach to Predicting the Results of NBA Matches Omid Aryan Stanford University aryano@stanford.edu Ali Reza Sharafat Stanford University sharafat@stanford.edu Abstract The current paper presents

More information

Conservation of Energy. Chapter 7 of Essential University Physics, Richard Wolfson, 3 rd Edition

Conservation of Energy. Chapter 7 of Essential University Physics, Richard Wolfson, 3 rd Edition Conservation of Energy Chapter 7 of Essential University Physics, Richard Wolfson, 3 rd Edition 1 Different Types of Force, regarding the Work they do. gravity friction 2 Conservative Forces BB WW cccccccc

More information

Tie Breaking Procedure

Tie Breaking Procedure Ohio Youth Basketball Tie Breaking Procedure The higher seeded team when two teams have the same record after completion of pool play will be determined by the winner of their head to head competition.

More information

Midterm Exam 1, section 2. Thursday, September hour, 15 minutes

Midterm Exam 1, section 2. Thursday, September hour, 15 minutes San Francisco State University Michael Bar ECON 312 Fall 2018 Midterm Exam 1, section 2 Thursday, September 27 1 hour, 15 minutes Name: Instructions 1. This is closed book, closed notes exam. 2. You can

More information

Improving Context Modeling for Video Object Detection and Tracking

Improving Context Modeling for Video Object Detection and Tracking Learning and Vision Group (NUS), ILSVRC 2017 - VID tasks Improving Context Modeling for Video Object Detection and Tracking National University of Singapore: Yunchao Wei, Mengdan Zhang, Jianan Li, Yunpeng

More information

Predicting the Total Number of Points Scored in NFL Games

Predicting the Total Number of Points Scored in NFL Games Predicting the Total Number of Points Scored in NFL Games Max Flores (mflores7@stanford.edu), Ajay Sohmshetty (ajay14@stanford.edu) CS 229 Fall 2014 1 Introduction Predicting the outcome of National Football

More information

Neural Networks II. Chen Gao. Virginia Tech Spring 2019 ECE-5424G / CS-5824

Neural Networks II. Chen Gao. Virginia Tech Spring 2019 ECE-5424G / CS-5824 Neural Networks II Chen Gao ECE-5424G / CS-5824 Virginia Tech Spring 2019 Neural Networks Origins: Algorithms that try to mimic the brain. What is this? A single neuron in the brain Input Output Slide

More information

HIGH RESOLUTION DEPTH IMAGE RECOVERY ALGORITHM USING GRAYSCALE IMAGE.

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 information

SNARKs with Preprocessing. Eran Tromer

SNARKs with Preprocessing. Eran Tromer SNARKs with Preprocessing Eran Tromer BIU Winter School on Verifiable Computation and Special Encryption 4-7 Jan 206 G: Someone generates and publishes a common reference string P: Prover picks NP statement

More information

Introduction to Pattern Recognition

Introduction 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 information

Predicting NBA Shots

Predicting NBA Shots Predicting NBA Shots Brett Meehan Stanford University https://github.com/brettmeehan/cs229 Final Project bmeehan2@stanford.edu Abstract This paper examines the application of various machine learning algorithms

More information

Combining Experimental and Non-Experimental Design in Causal Inference

Combining Experimental and Non-Experimental Design in Causal Inference Combining Experimental and Non-Experimental Design in Causal Inference Kari Lock Morgan Department of Statistics Penn State University Rao Prize Conference May 12 th, 2017 A Tribute to Don Design trumps

More information

NCSS Statistical Software

NCSS 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 information

Spatio-temporal analysis of team sports Joachim Gudmundsson

Spatio-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 information

Coaches, Parents, Players and Fans

Coaches, Parents, Players and Fans P.O. Box 865 * Lancaster, OH 43130 * 740-808-0380 * www.ohioyouthbasketball.com Coaches, Parents, Players and Fans Sunday s Championship Tournament in Boys Grades 5th thru 10/11th will be conducted in

More information

Substitution of Energy and Capital an Its Uncertainty for China

Substitution of Energy and Capital an Its Uncertainty for China Substitution of Energy and Capital an Its Uncertainty for China Zhaoning ZHENG Research Center of Contemporary Management Global Climate Change Institute School of Public policy and management Tsinghua

More information

DNS Study on Three Vortex Identification Methods

DNS Study on Three Vortex Identification Methods Γ DNS Study on Three Vortex Identification Methods Yinlin Dong Yong Yang Chaoqun Liu Technical Report 2016-07 http://www.uta.edu/math/preprint/ DNS Study on Three Vortex Identification Methods Yinlin Dong

More information

Evaluating and Classifying NBA Free Agents

Evaluating 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 information

Predicting Horse Racing Results with TensorFlow

Predicting 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 information

A 28nm SoC with a 1.2GHz 568nJ/ Prediction Sparse Deep-Neural-Network Engine with >0.1 Timing Error Rate Tolerance for IoT Applications

A 28nm SoC with a 1.2GHz 568nJ/ Prediction Sparse Deep-Neural-Network Engine with >0.1 Timing Error Rate Tolerance for IoT Applications A 28nm SoC with a 1.2GHz 568nJ/ Prediction Sparse Deep-Neural-Network Engine with >0.1 Timing Error Rate Tolerance for IoT Applications Paul Whatmough, S. K. Lee, H. Lee, S. Rama, D. Brooks, G.-Y. Wei

More information

The sycc Color Space

The sycc Color Space The sycc Color Space Douglas A. Kerr, P.E. Issue 2 July 24, 215 ABSTRACT The sycc color space is an alternative representation of the srgb color space, with a special wrinkle though which it can represent

More information

Parsimonious Linear Fingerprinting for Time Series

Parsimonious 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 information

Naïve Bayes. Robot Image Credit: Viktoriya Sukhanova 123RF.com

Naïve 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 information

Machine Learning Methods for Climbing Route Classification

Machine Learning Methods for Climbing Route Classification Machine Learning Methods for Climbing Route Classification Alejandro Dobles Mathematics adobles@stanford.edu Juan Carlos Sarmiento Management Science & Engineering jcs10@stanford.edu Abstract Peter Satterthwaite

More information

Chapter 10 Aggregate Demand I: Building the IS LM Model

Chapter 10 Aggregate Demand I: Building the IS LM Model Chapter 10 Aggregate Demand I: Building the IS LM Model Zhengyu Cai Ph.D. Institute of Development Southwestern University of Finance and Economics All rights reserved http://www.escience.cn/people/zhengyucai/index.html

More information

Analysis of Gini s Mean Difference for Randomized Block Design

Analysis of Gini s Mean Difference for Randomized Block Design American Journal of Mathematics and Statistics 2015, 5(3): 111-122 DOI: 10.5923/j.ajms.20150503.02 Analysis of Gini s Mean Difference for Randomized Block Design Elsayed A. H. Elamir Department of Statistics

More information

Biomechanics and Models of Locomotion

Biomechanics 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 information

Naïve Bayes. Robot Image Credit: Viktoriya Sukhanova 123RF.com

Naïve 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 information

Predicting Horse Racing Results with Machine Learning

Predicting Horse Racing Results with Machine Learning Predicting Horse Racing Results with Machine Learning LYU 1703 LIU YIDE 1155062194 Supervisor: Professor Michael R. Lyu Outline Recap of last semester Object of this semester Data Preparation Set to sequence

More information

graphic standards manual Mountain States Health Alliance

graphic standards manual Mountain States Health Alliance manual Mountain States mountain states health alliance Bringing Loving Care to Health Care Table of Contents 4 5 6 7 9 -- Why do we need? Brandmark Specifications Brandmark Color Palette Corporate Typography

More information

ISyE 6414 Regression Analysis

ISyE 6414 Regression Analysis ISyE 6414 Regression Analysis Lecture 2: More Simple linear Regression: R-squared (coefficient of variation/determination) Correlation analysis: Pearson s correlation Spearman s rank correlation Variable

More information

NATIONAL FEDERATION RULES B. National Federation Rules Apply with the following TOP GUN EXCEPTIONS

NATIONAL FEDERATION RULES B. National Federation Rules Apply with the following TOP GUN EXCEPTIONS TOP GUN COACH PITCH RULES 8 & Girls Division Revised January 11, 2018 AGE CUT OFF A. Age 8 & under. Cut off date is January 1st. Player may not turn 9 before January 1 st. Please have Birth Certificates

More information

What is Restrained and Unrestrained Pipes and what is the Strength Criteria

What is Restrained and Unrestrained Pipes and what is the Strength Criteria What is Restrained and Unrestrained Pipes and what is the Strength Criteria Alex Matveev, September 11, 2018 About author: Alex Matveev is one of the authors of pipe stress analysis codes GOST 32388-2013

More information

Product Decomposition in Supply Chain Planning

Product 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 information

Open 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 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 information

The Evolution of Transport Planning

The Evolution of Transport Planning The Evolution of Transport Planning On Proportionality and Uniqueness in Equilibrium Assignment Michael Florian Calin D. Morosan Background Several bush-based algorithms for computing equilibrium assignments

More information

intended velocity ( u k arm movements

intended velocity ( u k arm movements Fig. A Complete Brain-Machine Interface B Human Subjects Closed-Loop Simulator ensemble action potentials (n k ) ensemble action potentials (n k ) primary motor cortex simulated primary motor cortex neuroprosthetic

More information

FORECASTING OF ROLLING MOTION OF SMALL FISHING VESSELS UNDER FISHING OPERATION APPLYING A NON-DETERMINISTIC METHOD

FORECASTING OF ROLLING MOTION OF SMALL FISHING VESSELS UNDER FISHING OPERATION APPLYING A NON-DETERMINISTIC METHOD 8 th International Conference on 633 FORECASTING OF ROLLING MOTION OF SMALL FISHING VESSELS UNDER FISHING OPERATION APPLYING A NON-DETERMINISTIC METHOD Nobuo Kimura, Kiyoshi Amagai Graduate School of Fisheries

More information

Folding Reticulated Shell Structure Wind Pressure Coefficient Prediction Research based on RBF Neural Network

Folding Reticulated Shell Structure Wind Pressure Coefficient Prediction Research based on RBF Neural Network International Conference on Information Sciences, Machinery, Materials and Energy (ICISMME 2015) Folding Reticulated Shell Structure Wind Pressure Coefficient Prediction Research based on RBF Neural Network

More information

Introduction to Genetics

Introduction to Genetics Name: Introduction to Genetics Keystone Assessment Anchor: BIO.B.2.1.1: Describe and/or predict observed patterns of inheritance (i.e. dominant, recessive, co-dominance, incomplete dominance, sex-linked,

More information

Phrase-based Image Captioning

Phrase-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 information

Addition and Subtraction of Rational Expressions

Addition and Subtraction of Rational Expressions RT.3 Addition and Subtraction of Rational Expressions Many real-world applications involve adding or subtracting algebraic fractions. Similarly as in the case of common fractions, to add or subtract algebraic

More information

Numerical Simulation of Wave Loads on Static Offshore Structures

Numerical Simulation of Wave Loads on Static Offshore Structures Numerical Simulation of Wave Loads on Static Offshore Structures Hrvoje Jasak, Inno Gatin, Vuko Vukčević Wikki Ltd, United Kingdom Faculty of Mechanical Engineering and Naval Architecture University of

More information

Functions of Random Variables & Expectation, Mean and Variance

Functions of Random Variables & Expectation, Mean and Variance Functions of Random Variables & Expectation, Mean and Variance Kuan-Yu Chen ( 陳冠宇 ) @ TR-409, NTUST Functions of Random Variables 1 Given a random variables XX, one may generate other random variables

More information

A Class of Regression Estimator with Cum-Dual Ratio Estimator as Intercept

A Class of Regression Estimator with Cum-Dual Ratio Estimator as Intercept International Journal of Probability and Statistics 015, 4(): 4-50 DOI: 10.593/j.ijps.015040.0 A Class of Regression Estimator with Cum-Dual Ratio Estimator as Intercept F. B. Adebola 1, N. A. Adegoke

More information

B. AA228/CS238 Component

B. AA228/CS238 Component Abstract Two supervised learning methods, one employing logistic classification and another employing an artificial neural network, are used to predict the outcome of baseball postseason series, given

More information

LOCOMOTION CONTROL CYCLES ADAPTED FOR DISABILITIES IN HEXAPOD ROBOTS

LOCOMOTION 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 information

ORF 201 Computer Methods in Problem Solving. Final Project: Dynamic Programming Optimal Sailing Strategies

ORF 201 Computer Methods in Problem Solving. Final Project: Dynamic Programming Optimal Sailing Strategies Princeton University Department of Operations Research and Financial Engineering ORF 201 Computer Methods in Problem Solving Final Project: Dynamic Programming Optimal Sailing Strategies Due 11:59 pm,

More information

CS 7641 A (Machine Learning) Sethuraman K, Parameswaran Raman, Vijay Ramakrishnan

CS 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 information

Name: Grade: LESSON ONE: Home Row

Name: Grade: LESSON ONE: Home Row LESSON ONE: Home Row asdfjkl; asdfjkl; asdfjkl; aa ss dd ff jj kk ll ;; aa ss dd ff jj kk ll ;; aa ss dd ff jj kk ll ;; aa ss dd ff jj kk ll ;; aa ss dd ff jj kk ll ;; aa ss dd ff jj kk ll ;; aa ss dd

More information

A Brief History of the Development of Artificial Neural Networks

A Brief History of the Development of Artificial Neural Networks A Brief History of the Development of Artificial Neural Networks Prof. Bernard Widrow Department of Electrical Engineering Stanford University Baidu July 18, 2018 Prof. Widrow @ Berkeley A Brief History

More information

Smart-Walk: An Intelligent Physiological Monitoring System for Smart Families

Smart-Walk: An Intelligent Physiological Monitoring System for Smart Families Smart-Walk: An Intelligent Physiological Monitoring System for Smart Families P. Sundaravadivel 1, S. P. Mohanty 2, E. Kougianos 3, V. P. Yanambaka 4, and M. K. Ganapathiraju 5 University of North Texas,

More information

Report for Experiment #11 Testing Newton s Second Law On the Moon

Report for Experiment #11 Testing Newton s Second Law On the Moon Report for Experiment #11 Testing Newton s Second Law On the Moon Neil Armstrong Lab partner: Buzz Aldrin TA: Michael Collins July 20th, 1969 Abstract In this experiment, we tested Newton s second law

More information

FIRE PROTECTION. In fact, hydraulic modeling allows for infinite what if scenarios including:

FIRE PROTECTION. In fact, hydraulic modeling allows for infinite what if scenarios including: By Phil Smith, Project Manager and Chen-Hsiang Su, PE, Senior Consultant, Lincolnshire, IL, JENSEN HUGHES A hydraulic model is a computer program configured to simulate flows for a hydraulic system. The

More information

My ABC Insect Discovery Book

My ABC Insect Discovery Book Act i vi t ypack I ns ectabcs I nt hi spac k: ABCASLSi gnfl as hcar ds I ns ec tabcac t i v i t y SongL y r i c s AndMor e! Ac i t i v i ess uppor ts i gnsandc onc ept st aughti n Rac hel&t het r eesc

More information

2 When Some or All Labels are Missing: The EM Algorithm

2 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 information

The Simple Linear Regression Model ECONOMETRICS (ECON 360) BEN VAN KAMMEN, PHD

The Simple Linear Regression Model ECONOMETRICS (ECON 360) BEN VAN KAMMEN, PHD The Simple Linear Regression Model ECONOMETRICS (ECON 360) BEN VAN KAMMEN, PHD Outline Definition. Deriving the Estimates. Properties of the Estimates. Units of Measurement and Functional Form. Expected

More information

This file is part of the following reference:

This file is part of the following reference: This file is part of the following reference: Hancock, Timothy Peter (2006) Multivariate consensus trees: tree-based clustering and profiling for mixed data types. PhD thesis, James Cook University. Access

More information

Real Time Early Warning Indicators for Costly Asset Price Boom/Bust Cycles: A Role for Global Liquidity

Real Time Early Warning Indicators for Costly Asset Price Boom/Bust Cycles: A Role for Global Liquidity Indicators Asset Price : A Role for Global Liquidity Lucia Alessi European Central Bank Carsten Detken Milan, 24 June 2010 The need for Models The set up of an System is one of the key tasks of the European

More information