Regression-based Statistical Bounds on Software Execution Time

Size: px
Start display at page:

Download "Regression-based Statistical Bounds on Software Execution Time"

Transcription

1 ecos 17, Montreal Regression-based Statistical Bounds on Software Execution Time Ayoub Nouri P. Poplavko, L. Angelis, A. Zerzelidis, S. Bensalem, P. Katsaros University of Grenoble-Alpes - France August 24, 2017

2 WCET: Statistical vs. Common Approaches WCET is useful for... Schedulability of real-time systems Performance Evaluation, e.g, Network Calculus Model-based design, i.e. building faithful Hw/Sw models 2 / 29

3 WCET: Statistical vs. Common Approaches WCET is useful for... Schedulability of real-time systems Performance Evaluation, e.g, Network Calculus Model-based design, i.e. building faithful Hw/Sw models Statistical approaches probabilistic upper-bounds, correctness probability 1 α Common approaches 100% certain upper-bounds, α = 0 Prob. Pr (TT > tt ii ) TT is a R.V modeling Exec. Time αα Exec. time measurements tt 1 tt ii tt αα>0 tt αα=0 Execution time tt nn 2 / 29

4 Advantages and Shortcomings Both approaches are accepted in the context of safety-critical systems with hard real-time constraints achieve α (EVT) and α = 0 (common approaches) 3 / 29

5 Advantages and Shortcomings Both approaches are accepted in the context of safety-critical systems with hard real-time constraints achieve α (EVT) and α = 0 (common approaches) Extreme Value Theory (EVT): i.i.d exec. times random Hw, e.g., cache Common approaches: costly faithful model of Hw 3 / 29

6 Advantages and Shortcomings Both approaches are accepted in the context of safety-critical systems with hard real-time constraints achieve α (EVT) and α = 0 (common approaches) Extreme Value Theory (EVT): i.i.d exec. times random Hw, e.g., cache Common approaches: costly faithful model of Hw Our approach: Maximal Execution Time (MET) statistical measurement-based non-safety-critical systems with soft real-time constraints e.g., video streaming systems 3 / 29

7 Measurement-Based Timing Analysis (MBTA) Program Exec. time measures tt ii Analysis WCET/MET ττ Input data samples 4 / 29

8 Measurement-Based Timing Analysis (MBTA) Program Exec. time measures tt ii Analysis WCET/MET ττ Input data samples Probabilistic MBTA Given α, compute the probabilistic upper-bound τ as τ = min t i s.t. P r{t < t i } (1 α) Prob. Pr (TT > tt ii ) Pr (TT < tt ii ) 1 αα TT is a R.V modeling Exec. Time αα Execution time 4 / 29

9 A simplified view... Density Density 1 αα Execution time Figure: Histogram and density Figure: Quantile function tt ii Execution time each exec. time observation t i is modeled as a random variable T i Assume it was a probability distribution... suppose T i are i.i.d and T i f then given α, the solution τ s.t. P r{t < t i } = (1 α) is τ = Q f (1 α) where, Q f is the Quantile function of f 5 / 29

10 Exec. Time is not Purely Random Density Execution time T i are not i.i.d no f such that T i f Figure: Execution time in practice Executions time dependencies input data, e.g., images X 1 : captures the size of input images, X 2 : captures their type mono/multi-chromatic,... Hw effect, e.g., cache 6 / 29

11 Main Contributions Maximal Regression Model a model that captures dependencies enables to derive execution times upper-bound from }{{} T = h(x 1, X 2,... ) + }{{}}{{} ɛ Execution Time Oracle for execution time Random error based on Linear Regression and Confidence Intervals 7 / 29

12 Main Contributions Maximal Regression Model a model that captures dependencies enables to derive execution times upper-bound from }{{} T = h(x 1, X 2,... ) + }{{}}{{} ɛ Execution Time Oracle for execution time Random error based on Linear Regression and Confidence Intervals Construction Methods a method to identify the set of most pertinent X i a method to control the quality of input data (measurements) 7 / 29

13 Main Contributions Maximal Regression Model a model that captures dependencies enables to derive execution times upper-bound from }{{} T = h(x 1, X 2,... ) + }{{}}{{} ɛ Execution Time Oracle for execution time Random error based on Linear Regression and Confidence Intervals Construction Methods a method to identify the set of most pertinent X i a method to control the quality of input data (measurements) An automated design-flow to compute MET integrates and implements the previous contributions used to analyze a JPEG Decoder 7 / 29

14 Outline 1 The Maximal Regression Model 2 Model Construction Techniques 3 A JPEG Decoder Study 4 Conclusions 8 / 29

15 Linear Regression (LR) Why is LR an appropriate model? Execution time can be always decomposed as a linear combination of code block contributions 9 / 29

16 Linear Regression (LR) Why is LR an appropriate model? Execution time can be always decomposed as a linear combination of code block contributions Ideal model Y = β 0 + β 1 X 1 + β 2 X β p X p + ɛ Y is the explained variable, i.e., execution time, X i are the explanatory vars. (predictors), i.e. block counters, β i are parameters/coefficients, i.e. block costs, ɛ is the regression error 9 / 29

17 Linear Regression (LR) Approximated model Y = β 0 + β 1 X 1 + β 2 X β p X p + ɛ 10 / 29

18 Linear Regression (LR) Approximated model Y = β 0 + β 1 X 1 + β 2 X β p X p + ɛ Ŷ = b 0 + b 1 X 1 + b 2 X b p X p Ŷ approximates the explained variable, i.e., execution time, X i are the explanatory variables (predictors), b i are estimations of β i to be computed (measurements), ɛ can be neglected (on average) 10 / 29

19 LR in the nutshell Greatest Common Divisor (GCD) I n p u t : i n t n, k ( n >= k ) Output : GCD( n, k ) r e p e a t { p := n ; n := k ; k := p mod n ; } u n t i l k=0; Execution time Y is proportional to the cost of a loop iteration (β 1 ) Nb of iterations (X 1 ) depends on the input parameters (n, k) r e t u r n n ; 11 / 29

20 LR in the nutshell Greatest Common Divisor (GCD) I n p u t : i n t n, k ( n >= k ) Output : GCD( n, k ) r e p e a t { p := n ; n := k ; k := p mod n ; } u n t i l k=0; Execution time Y is proportional to the cost of a loop iteration (β 1 ) Nb of iterations (X 1 ) depends on the input parameters (n, k) r e t u r n n ; Y = β 0 + β 1 X 1 + ɛ 11 / 29

21 LR in the nutshell Artificial Program I n p u t : i n t i1, i2, i 3 Output : out out := i 1 i 2 ; i f ( out <0) { out := out ; } w h i l e ( out > i 3 ) { out := out 0. 9 ; } r e t u r n out ; Execution time Y is proportional to the cost of the if body (β 1 ) and the while loop iteration (β 2 ) Nb of times if branch (X 1 ) is taken and nb of iterations (X 2 ) depend on the input parameters (i1, i2, i3) Y = β 0 + β 1 X 1 + β 2 X 2 + ɛ 12 / 29

22 The Maximal Regression Model Approximated LR model Ŷ = b 0 + b 1 X 1 + b 2 X b p X p Generally used to compute average execution time 13 / 29

23 The Maximal Regression Model Approximated LR model Ŷ = b 0 + b 1 X 1 + b 2 X b p X p Generally used to compute average execution time Execution Time + YY YY YY bb ii ββ ii bb ii + Figure: Confidence Interval on Ŷ Figure: Confidence Interval on β i 13 / 29

24 The Maximal Regression Model Approximated LR model Ŷ = b 0 + b 1 X 1 + b 2 X b p X p Generally used to compute average execution time Execution Time + YY YY YY bb ii ββ ii bb ii + Figure: Confidence Interval on Ŷ Figure: Confidence Interval on β i Our model Ŷ + = b b+ 1 X 1 + b + 2 X b p X p + ɛ + ɛ + is obtained by the same principle using a χ 2 distribution 13 / 29

25 Probabilistic bound The associated probability Ŷ + = b b+ 1 X 1 + b + 2 X b p X p + ɛ + P r{y < Ŷ + (p + 2)α } 1 2 assuming (1 α) confidence intervals for each b i and ɛ Challenges find predictors X i the most influential on the execution time avoid redundancy and inter-correlation consider a representative set of measurements w.r.t input data 14 / 29

26 Outline 1 The Maximal Regression Model 2 Model Construction Techniques 3 A JPEG Decoder Study 4 Conclusions 15 / 29

27 Identifying Predictors: Step-wise Regression Potential predictors P P ={ a predictor per block of code e.g. if, switch / case, loops } Identifying the most relevant predictors, why? parsimony: smallest/relevant/sufficient subset p P rule of thumb N > 5p, i.e. smaller model - less measurements keeping too many variables induces over-fitting SNR(X i, p) < α p p \ { X i } P P { X i } X i := argmin SNR(X i, p) P; p = { X 0 };α X i := argmax SNR(X i, P) SNR(X i, P) > α p p { X i } P P \ { X i } End 16 / 29

28 What About Input Data?... identify algorithmic complexity factors, and come-up with an input data-set, where every combination of these factors is represented fairly Cook s Distance A distance measure D(t i ) that quantifies the influence of t i on the regression model recall that execution time depends on input data can be seen as a distance measure on input data Measurements shouldn t be dominated by outliers, i.e. D(t i ) > θ either add more similar samples, or remove them from the measurement set (keep for testing) 17 / 29

29 Computing MET Pragmatic MET Assumption Ŷ + = b b+ 1 X 1 + b + 2 X b p X p + ɛ + p MET: τ = b (bx) + i (bx) + i = i=1 + ɛ +, where { b + X + if b + > 0 b + X if b + < 0 know the min/max bounds of each X i, i.e. X + i, X i often, intuitive human interpretation can be given for X i then, an expert can give an min/max bounds on X i e.g. X 1 images size : max images size can be determined from camera resolution 18 / 29

30 Outline 1 The Maximal Regression Model 2 Model Construction Techniques 3 A JPEG Decoder Study 4 Conclusions 19 / 29

31 Experiments Setting JPEG Files Sequence of compressed MCUs (16x16 or 8x8 pixels) Each MCU contains pixel blocks In color format 4:1:1, it contains 6 blocks In monochromatic, it contains 1 block Pixel blocks are represented by a matrix of DCT coef. JPEG Decoder FPGA SPARC V8 with 7-stage pipeline Double precision FPU 4 KB instruction cache 4 KB data cache 256 L2 cache and an SDRAM reset Data caches for each new program run (i.e. each new image) input: 99 JPEG images, different sizes and color formats, output: 99 measurements (X i and execution time Y ) 0 MCU: Minimum Coded Units 0 DCT: Discrete Cosine Transform 20 / 29

32 Measurements identified 95 predictors P = {X 1,..., X 95 } X = X 0 X 1 X t , Y = t 99 Some observations... the maximal measured execution time was Mcycles corresponds to a particularly large image the average execution time was 1000 Mcycles Pre-processing split the set of measurements into 70 measurements for training, and 29 measurements for testing 21 / 29

33 Training set vs. Test set Training set Test set X train = X 0 X 1 X t , Y train = used to compute coefficients b + i and ɛ + t 70 X 0 X 1 X t 71 X test = , Y test = used to test whether the obtained model can correctly (generalize to/handle) different unseen inputs t / 29

34 Results Basic Model (p = 1, α = 0.05) Y = β 0 + ɛ error ɛ + = 6650 Mcycles pragmatic MET 8000 Mcycles under-estimates the maximal measured time (23600 Mcycles) 23 / 29

35 Results Basic Model (p = 1, α = 0.05) Y = β 0 + ɛ error ɛ + = 6650 Mcycles pragmatic MET 8000 Mcycles under-estimates the maximal measured time (23600 Mcycles) Our Method (p = 6, α = 0.05) Y = β i=1 β ix i + ɛ 23 / 29

36 Identified predictors per MCU block X 3 Cost p Human interpretation per Byte X 1 byte count in the main body of JPEG per pixel block X 2 pixel block count for those blocks that had correct prediction in term of 0-th DCT coefficient number of elements in color format (e.g. 5 for color, 0 for monochromatic) X 4 number of padded image dimensions, i.e. those not exactly proportional to the MCU size X 5 total number of MCUs in monochromatic images - not costly in term of need bytes for encoding (contrib. of costly blocks captured by X 1 ) - such dimensions implies less processing/- data copying for partial MCU blocks - complimentary to X 3 24 / 29

37 Results Basic Model (p = 1, α = 0.5) Y = β 0 + ɛ error ɛ + = 6650 Mcycles pragmatic MET 8000 Mcycles under-estimates the maximal measured time (23600 Mcycles) Our Method (p = 6, α = 0.05) Y = β i=1 β ix i + ɛ outliers detection error ɛ + = 240 pragmatic MET under-estim. on test-set 25 / 29

38 Results Basic Model (p = 1, α = 0.5) Y = β 0 + ɛ error ɛ + = 6650 Mcycles pragmatic MET 8000 Mcycles under-estimates the maximal measured time (23600 Mcycles) Our Method (p = 6, α = 0.05) Y = β i=1 β ix i + ɛ outliers detection error ɛ + = 240 pragmatic MET under-estim. on test-set + outliers detection error ɛ + = 52 pragmatic MET over-estim for all tests 25 / 29

39 Results Our Method (p = 8) Y = β i=1 β ix i + ɛ error ɛ + = 35 pragmatic MET (α = 0.05) pragmatic MET (α = ) 20 Residual in training set (Mcycles) 3000 Actual and Predicted Execution Times Actual alpha= alpha= Execution Times (Mcycles) test set 26 / 29

40 Outline 1 The Maximal Regression Model 2 Model Construction Techniques 3 A JPEG Decoder Study 4 Conclusions 27 / 29

41 Summary Problem computing a probabilistic upper-bound on embedded Sw execution time Approach for non safety-critical systems having soft real-time constraints statistical, Linear Regression measurement-based Contributions Maximal Regression Model (LR + CI) construction tech. (step-wise regression, Cook s distance, etc.) integrated design flow validated on a real-life system 28 / 29

42 Future Work Pragmatic MET is pessimistic likely to incur extra over-estimation including unfeasible execution paths e.g. switch-case branching: associates a separate predictor with every case/assumes that they all take maximal values simultaneously combine it with implicit path enumeration technique (IPET) keep only realistic/feasible execution paths Further directions model Hw effect using specially defined predictors investigate possible connection to Extreme Value Theory 29 / 29

Regression-based Statistical Bounds on Software Execution Time

Regression-based Statistical Bounds on Software Execution Time Regression-based Statistical Bounds on Software Execution Time Peter Poplavko 1, Ayoub Nouri 1, Lefteris Angelis 2,3, Alexandros Zerzelidis 2, Saddek Bensalem 1, and Panagiotis Katsaros 2,3 1 Univ. Grenoble

More information

Regression-based Statistical Bounds on Software Execution Time

Regression-based Statistical Bounds on Software Execution Time Regression-based Statistical Bounds on Software Execution Time Peter Poplavko, Lefteris Angelis, Ayoub Nouri, Alexandros Zerzelidis, Saddek Bensalem, Panagiotis Verimag Research Report n o TR-2016-7 November

More information

Lecture 18: Simple Linear Regression

Lecture 18: Simple Linear Regression Lecture 18: Simple Linear Regression BIOS 553 Department of Biostatistics University of Michigan Fall 2004 The Correlation Coefficient: r The correlation coefficient (r) is a number that measures the strength

More information

CSCI-567: Machine Learning (Spring 2019)

CSCI-567: Machine Learning (Spring 2019) CSCI-567: Machine Learning (Spring 2019) Prof. Victor Adamchik U of Southern California Mar. 19, 2019 March 19, 2019 1 / 43 Administration March 19, 2019 2 / 43 Administration TA3 is due this week March

More information

ECE521 Lecture 7/8. Logistic Regression

ECE521 Lecture 7/8. Logistic Regression ECE521 Lecture 7/8 Logistic Regression Outline Logistic regression (Continue) A single neuron Learning neural networks Multi-class classification 2 Logistic regression The output of a logistic regression

More information

Fine Grain Quality Management

Fine Grain Quality Management Fine Grain Quality Management Jacques Combaz Jean-Claude Fernandez Mohamad Jaber Joseph Sifakis Loïc Strus Verimag Lab. Université Joseph Fourier Grenoble, France DCS seminar, 10 June 2008, Col de Porte

More information

Complexity Theory VU , SS The Polynomial Hierarchy. Reinhard Pichler

Complexity Theory VU , SS The Polynomial Hierarchy. Reinhard Pichler Complexity Theory Complexity Theory VU 181.142, SS 2018 6. The Polynomial Hierarchy Reinhard Pichler Institut für Informationssysteme Arbeitsbereich DBAI Technische Universität Wien 15 May, 2018 Reinhard

More information

Outline. Complexity Theory EXACT TSP. The Class DP. Definition. Problem EXACT TSP. Complexity of EXACT TSP. Proposition VU 181.

Outline. Complexity Theory EXACT TSP. The Class DP. Definition. Problem EXACT TSP. Complexity of EXACT TSP. Proposition VU 181. Complexity Theory Complexity Theory Outline Complexity Theory VU 181.142, SS 2018 6. The Polynomial Hierarchy Reinhard Pichler Institut für Informationssysteme Arbeitsbereich DBAI Technische Universität

More information

Objective: Reduction of data redundancy. Coding redundancy Interpixel redundancy Psychovisual redundancy Fall LIST 2

Objective: Reduction of data redundancy. Coding redundancy Interpixel redundancy Psychovisual redundancy Fall LIST 2 Image Compression Objective: Reduction of data redundancy Coding redundancy Interpixel redundancy Psychovisual redundancy 20-Fall LIST 2 Method: Coding Redundancy Variable-Length Coding Interpixel Redundancy

More information

ANOVA Situation The F Statistic Multiple Comparisons. 1-Way ANOVA MATH 143. Department of Mathematics and Statistics Calvin College

ANOVA Situation The F Statistic Multiple Comparisons. 1-Way ANOVA MATH 143. Department of Mathematics and Statistics Calvin College 1-Way ANOVA MATH 143 Department of Mathematics and Statistics Calvin College An example ANOVA situation Example (Treating Blisters) Subjects: 25 patients with blisters Treatments: Treatment A, Treatment

More information

Constructing Prediction Intervals for Random Forests

Constructing Prediction Intervals for Random Forests Senior Thesis in Mathematics Constructing Prediction Intervals for Random Forests Author: Benjamin Lu Advisor: Dr. Jo Hardin Submitted to Pomona College in Partial Fulfillment of the Degree of Bachelor

More information

Association studies and regression

Association studies and regression Association studies and regression CM226: Machine Learning for Bioinformatics. Fall 2016 Sriram Sankararaman Acknowledgments: Fei Sha, Ameet Talwalkar Association studies and regression 1 / 104 Administration

More information

Introduction to Machine Learning. PCA and Spectral Clustering. Introduction to Machine Learning, Slides: Eran Halperin

Introduction to Machine Learning. PCA and Spectral Clustering. Introduction to Machine Learning, Slides: Eran Halperin 1 Introduction to Machine Learning PCA and Spectral Clustering Introduction to Machine Learning, 2013-14 Slides: Eran Halperin Singular Value Decomposition (SVD) The singular value decomposition (SVD)

More information

Image Data Compression

Image Data Compression Image Data Compression Image data compression is important for - image archiving e.g. satellite data - image transmission e.g. web data - multimedia applications e.g. desk-top editing Image data compression

More information

Seq2Seq Losses (CTC)

Seq2Seq Losses (CTC) Seq2Seq Losses (CTC) Jerry Ding & Ryan Brigden 11-785 Recitation 6 February 23, 2018 Outline Tasks suited for recurrent networks Losses when the output is a sequence Kinds of errors Losses to use CTC Loss

More information

MLCC 2018 Variable Selection and Sparsity. Lorenzo Rosasco UNIGE-MIT-IIT

MLCC 2018 Variable Selection and Sparsity. Lorenzo Rosasco UNIGE-MIT-IIT MLCC 2018 Variable Selection and Sparsity Lorenzo Rosasco UNIGE-MIT-IIT Outline Variable Selection Subset Selection Greedy Methods: (Orthogonal) Matching Pursuit Convex Relaxation: LASSO & Elastic Net

More information

Linear Regression (9/11/13)

Linear Regression (9/11/13) STA561: Probabilistic machine learning Linear Regression (9/11/13) Lecturer: Barbara Engelhardt Scribes: Zachary Abzug, Mike Gloudemans, Zhuosheng Gu, Zhao Song 1 Why use linear regression? Figure 1: Scatter

More information

Empirical Likelihood Methods for Sample Survey Data: An Overview

Empirical Likelihood Methods for Sample Survey Data: An Overview AUSTRIAN JOURNAL OF STATISTICS Volume 35 (2006), Number 2&3, 191 196 Empirical Likelihood Methods for Sample Survey Data: An Overview J. N. K. Rao Carleton University, Ottawa, Canada Abstract: The use

More information

Embedded Systems 23 BF - ES

Embedded Systems 23 BF - ES Embedded Systems 23-1 - Measurement vs. Analysis REVIEW Probability Best Case Execution Time Unsafe: Execution Time Measurement Worst Case Execution Time Upper bound Execution Time typically huge variations

More information

Simple Linear Regression for the Climate Data

Simple Linear Regression for the Climate Data Prediction Prediction Interval Temperature 0.2 0.0 0.2 0.4 0.6 0.8 320 340 360 380 CO 2 Simple Linear Regression for the Climate Data What do we do with the data? y i = Temperature of i th Year x i =CO

More information

HOMEWORK #4: LOGISTIC REGRESSION

HOMEWORK #4: LOGISTIC REGRESSION HOMEWORK #4: LOGISTIC REGRESSION Probabilistic Learning: Theory and Algorithms CS 274A, Winter 2019 Due: 11am Monday, February 25th, 2019 Submit scan of plots/written responses to Gradebook; submit your

More information

Ch 2: Simple Linear Regression

Ch 2: Simple Linear Regression Ch 2: Simple Linear Regression 1. Simple Linear Regression Model A simple regression model with a single regressor x is y = β 0 + β 1 x + ɛ, where we assume that the error ɛ is independent random component

More information

HOMEWORK #4: LOGISTIC REGRESSION

HOMEWORK #4: LOGISTIC REGRESSION HOMEWORK #4: LOGISTIC REGRESSION Probabilistic Learning: Theory and Algorithms CS 274A, Winter 2018 Due: Friday, February 23rd, 2018, 11:55 PM Submit code and report via EEE Dropbox You should submit a

More information

MATH 644: Regression Analysis Methods

MATH 644: Regression Analysis Methods MATH 644: Regression Analysis Methods FINAL EXAM Fall, 2012 INSTRUCTIONS TO STUDENTS: 1. This test contains SIX questions. It comprises ELEVEN printed pages. 2. Answer ALL questions for a total of 100

More information

INFERENCE FOR REGRESSION

INFERENCE FOR REGRESSION CHAPTER 3 INFERENCE FOR REGRESSION OVERVIEW In Chapter 5 of the textbook, we first encountered regression. The assumptions that describe the regression model we use in this chapter are the following. We

More information

MLR Model Checking. Author: Nicholas G Reich, Jeff Goldsmith. This material is part of the statsteachr project

MLR Model Checking. Author: Nicholas G Reich, Jeff Goldsmith. This material is part of the statsteachr project MLR Model Checking Author: Nicholas G Reich, Jeff Goldsmith This material is part of the statsteachr project Made available under the Creative Commons Attribution-ShareAlike 3.0 Unported License: http://creativecommons.org/licenses/by-sa/3.0/deed.en

More information

Source Coding. Master Universitario en Ingeniería de Telecomunicación. I. Santamaría Universidad de Cantabria

Source Coding. Master Universitario en Ingeniería de Telecomunicación. I. Santamaría Universidad de Cantabria Source Coding Master Universitario en Ingeniería de Telecomunicación I. Santamaría Universidad de Cantabria Contents Introduction Asymptotic Equipartition Property Optimal Codes (Huffman Coding) Universal

More information

Overview. Analog capturing device (camera, microphone) PCM encoded or raw signal ( wav, bmp, ) A/D CONVERTER. Compressed bit stream (mp3, jpg, )

Overview. Analog capturing device (camera, microphone) PCM encoded or raw signal ( wav, bmp, ) A/D CONVERTER. Compressed bit stream (mp3, jpg, ) Overview Analog capturing device (camera, microphone) Sampling Fine Quantization A/D CONVERTER PCM encoded or raw signal ( wav, bmp, ) Transform Quantizer VLC encoding Compressed bit stream (mp3, jpg,

More information

Density Temp vs Ratio. temp

Density Temp vs Ratio. temp Temp Ratio Density 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Density 0.0 0.2 0.4 0.6 0.8 1.0 1. (a) 170 175 180 185 temp 1.0 1.5 2.0 2.5 3.0 ratio The histogram shows that the temperature measures have two peaks,

More information

IEOR E4703: Monte-Carlo Simulation

IEOR E4703: Monte-Carlo Simulation IEOR E4703: Monte-Carlo Simulation Output Analysis for Monte-Carlo Martin Haugh Department of Industrial Engineering and Operations Research Columbia University Email: martin.b.haugh@gmail.com Output Analysis

More information

Multimedia Networking ECE 599

Multimedia Networking ECE 599 Multimedia Networking ECE 599 Prof. Thinh Nguyen School of Electrical Engineering and Computer Science Based on lectures from B. Lee, B. Girod, and A. Mukherjee 1 Outline Digital Signal Representation

More information

Six Sigma Black Belt Study Guides

Six Sigma Black Belt Study Guides Six Sigma Black Belt Study Guides 1 www.pmtutor.org Powered by POeT Solvers Limited. Analyze Correlation and Regression Analysis 2 www.pmtutor.org Powered by POeT Solvers Limited. Variables and relationships

More information

Regression Estimation - Least Squares and Maximum Likelihood. Dr. Frank Wood

Regression Estimation - Least Squares and Maximum Likelihood. Dr. Frank Wood Regression Estimation - Least Squares and Maximum Likelihood Dr. Frank Wood Least Squares Max(min)imization Function to minimize w.r.t. β 0, β 1 Q = n (Y i (β 0 + β 1 X i )) 2 i=1 Minimize this by maximizing

More information

Regression Model Building

Regression Model Building Regression Model Building Setting: Possibly a large set of predictor variables (including interactions). Goal: Fit a parsimonious model that explains variation in Y with a small set of predictors Automated

More information

Statistics 3858 : Maximum Likelihood Estimators

Statistics 3858 : Maximum Likelihood Estimators Statistics 3858 : Maximum Likelihood Estimators 1 Method of Maximum Likelihood In this method we construct the so called likelihood function, that is L(θ) = L(θ; X 1, X 2,..., X n ) = f n (X 1, X 2,...,

More information

Unit 6: Branch Prediction

Unit 6: Branch Prediction CIS 501: Computer Architecture Unit 6: Branch Prediction Slides developed by Joe Devie/, Milo Mar4n & Amir Roth at Upenn with sources that included University of Wisconsin slides by Mark Hill, Guri Sohi,

More information

Goldreich-Levin Hardcore Predicate. Lecture 28: List Decoding Hadamard Code and Goldreich-L

Goldreich-Levin Hardcore Predicate. Lecture 28: List Decoding Hadamard Code and Goldreich-L Lecture 28: List Decoding Hadamard Code and Goldreich-Levin Hardcore Predicate Recall Let H : {0, 1} n {+1, 1} Let: L ε = {S : χ S agrees with H at (1/2 + ε) fraction of points} Given oracle access to

More information

Lecture: Workload Models (Advanced Topic)

Lecture: Workload Models (Advanced Topic) Lecture: Workload Models (Advanced Topic) Real-Time Systems, HT11 Martin Stigge 28. September 2011 Martin Stigge Workload Models 28. September 2011 1 System

More information

10701/15781 Machine Learning, Spring 2007: Homework 2

10701/15781 Machine Learning, Spring 2007: Homework 2 070/578 Machine Learning, Spring 2007: Homework 2 Due: Wednesday, February 2, beginning of the class Instructions There are 4 questions on this assignment The second question involves coding Do not attach

More information

Error Detection & Correction

Error Detection & Correction Error Detection & Correction Error detection & correction noisy channels techniques in networking error detection error detection capability retransmition error correction reconstruction checksums redundancy

More information

Loss Functions, Decision Theory, and Linear Models

Loss Functions, Decision Theory, and Linear Models Loss Functions, Decision Theory, and Linear Models CMSC 678 UMBC January 31 st, 2018 Some slides adapted from Hamed Pirsiavash Logistics Recap Piazza (ask & answer questions): https://piazza.com/umbc/spring2018/cmsc678

More information

Wavelet Scalable Video Codec Part 1: image compression by JPEG2000

Wavelet Scalable Video Codec Part 1: image compression by JPEG2000 1 Wavelet Scalable Video Codec Part 1: image compression by JPEG2000 Aline Roumy aline.roumy@inria.fr May 2011 2 Motivation for Video Compression Digital video studio standard ITU-R Rec. 601 Y luminance

More information

Image Compression. Fundamentals: Coding redundancy. The gray level histogram of an image can reveal a great deal of information about the image

Image Compression. Fundamentals: Coding redundancy. The gray level histogram of an image can reveal a great deal of information about the image Fundamentals: Coding redundancy The gray level histogram of an image can reveal a great deal of information about the image That probability (frequency) of occurrence of gray level r k is p(r k ), p n

More information

Lecture 8. Instructor: Haipeng Luo

Lecture 8. Instructor: Haipeng Luo Lecture 8 Instructor: Haipeng Luo Boosting and AdaBoost In this lecture we discuss the connection between boosting and online learning. Boosting is not only one of the most fundamental theories in machine

More information

Generalized Linear Models

Generalized Linear Models Generalized Linear Models Lecture 3. Hypothesis testing. Goodness of Fit. Model diagnostics GLM (Spring, 2018) Lecture 3 1 / 34 Models Let M(X r ) be a model with design matrix X r (with r columns) r n

More information

Source Coding and Function Computation: Optimal Rate in Zero-Error and Vanishing Zero-Error Regime

Source Coding and Function Computation: Optimal Rate in Zero-Error and Vanishing Zero-Error Regime Source Coding and Function Computation: Optimal Rate in Zero-Error and Vanishing Zero-Error Regime Solmaz Torabi Dept. of Electrical and Computer Engineering Drexel University st669@drexel.edu Advisor:

More information

HOMEWORK 4: SVMS AND KERNELS

HOMEWORK 4: SVMS AND KERNELS HOMEWORK 4: SVMS AND KERNELS CMU 060: MACHINE LEARNING (FALL 206) OUT: Sep. 26, 206 DUE: 5:30 pm, Oct. 05, 206 TAs: Simon Shaolei Du, Tianshu Ren, Hsiao-Yu Fish Tung Instructions Homework Submission: Submit

More information

ECE521 week 3: 23/26 January 2017

ECE521 week 3: 23/26 January 2017 ECE521 week 3: 23/26 January 2017 Outline Probabilistic interpretation of linear regression - Maximum likelihood estimation (MLE) - Maximum a posteriori (MAP) estimation Bias-variance trade-off Linear

More information

Quiz 2 Date: Monday, November 21, 2016

Quiz 2 Date: Monday, November 21, 2016 10-704 Information Processing and Learning Fall 2016 Quiz 2 Date: Monday, November 21, 2016 Name: Andrew ID: Department: Guidelines: 1. PLEASE DO NOT TURN THIS PAGE UNTIL INSTRUCTED. 2. Write your name,

More information

A Multi-Periodic Synchronous Data-Flow Language

A Multi-Periodic Synchronous Data-Flow Language Julien Forget 1 Frédéric Boniol 1 David Lesens 2 Claire Pagetti 1 firstname.lastname@onera.fr 1 ONERA - Toulouse, FRANCE 2 EADS Astrium Space Transportation - Les Mureaux, FRANCE November 19, 2008 1 /

More information

Image Compression - JPEG

Image Compression - JPEG Overview of JPEG CpSc 86: Multimedia Systems and Applications Image Compression - JPEG What is JPEG? "Joint Photographic Expert Group". Voted as international standard in 99. Works with colour and greyscale

More information

ST430 Exam 1 with Answers

ST430 Exam 1 with Answers ST430 Exam 1 with Answers Date: October 5, 2015 Name: Guideline: You may use one-page (front and back of a standard A4 paper) of notes. No laptop or textook are permitted but you may use a calculator.

More information

LINEAR CLASSIFICATION, PERCEPTRON, LOGISTIC REGRESSION, SVC, NAÏVE BAYES. Supervised Learning

LINEAR CLASSIFICATION, PERCEPTRON, LOGISTIC REGRESSION, SVC, NAÏVE BAYES. Supervised Learning LINEAR CLASSIFICATION, PERCEPTRON, LOGISTIC REGRESSION, SVC, NAÏVE BAYES Supervised Learning Linear vs non linear classifiers In K-NN we saw an example of a non-linear classifier: the decision boundary

More information

ECON Introductory Econometrics. Lecture 11: Binary dependent variables

ECON Introductory Econometrics. Lecture 11: Binary dependent variables ECON4150 - Introductory Econometrics Lecture 11: Binary dependent variables Monique de Haan (moniqued@econ.uio.no) Stock and Watson Chapter 11 Lecture Outline 2 The linear probability model Nonlinear probability

More information

STA 114: Statistics. Notes 21. Linear Regression

STA 114: Statistics. Notes 21. Linear Regression STA 114: Statistics Notes 1. Linear Regression Introduction A most widely used statistical analysis is regression, where one tries to explain a response variable Y by an explanatory variable X based on

More information

Linear Regression. Simple linear regression model determines the relationship between one dependent variable (y) and one independent variable (x).

Linear Regression. Simple linear regression model determines the relationship between one dependent variable (y) and one independent variable (x). Linear Regression Simple linear regression model determines the relationship between one dependent variable (y) and one independent variable (x). A dependent variable is a random variable whose variation

More information

Supplemental Material : A Unified Framework for Multi-Target Tracking and Collective Activity Recognition

Supplemental Material : A Unified Framework for Multi-Target Tracking and Collective Activity Recognition Supplemental Material : A Unified Framewor for Multi-Target Tracing and Collective Activity Recognition Wongun Choi and Silvio Savarese Electrical and Computer Engineering University of Michigan Ann Arbor

More information

Least Squares Classification

Least Squares Classification Least Squares Classification Stephen Boyd EE103 Stanford University November 4, 2017 Outline Classification Least squares classification Multi-class classifiers Classification 2 Classification data fitting

More information

Remedial Measures for Multiple Linear Regression Models

Remedial Measures for Multiple Linear Regression Models Remedial Measures for Multiple Linear Regression Models Yang Feng http://www.stat.columbia.edu/~yangfeng Yang Feng (Columbia University) Remedial Measures for Multiple Linear Regression Models 1 / 25 Outline

More information

Machine Learning and Computational Statistics, Spring 2017 Homework 2: Lasso Regression

Machine Learning and Computational Statistics, Spring 2017 Homework 2: Lasso Regression Machine Learning and Computational Statistics, Spring 2017 Homework 2: Lasso Regression Due: Monday, February 13, 2017, at 10pm (Submit via Gradescope) Instructions: Your answers to the questions below,

More information

Analysing data: regression and correlation S6 and S7

Analysing data: regression and correlation S6 and S7 Basic medical statistics for clinical and experimental research Analysing data: regression and correlation S6 and S7 K. Jozwiak k.jozwiak@nki.nl 2 / 49 Correlation So far we have looked at the association

More information

Integrating Cache Related Preemption Delay Analysis into EDF Scheduling

Integrating Cache Related Preemption Delay Analysis into EDF Scheduling Integrating Cache Related Preemption Delay Analysis into EDF Scheduling Will Lunniss 1 Sebastian Altmeyer 2 Claire Maiza 3 Robert I. Davis 1 1 Real-Time Systems Research Group, University of York, UK {wl510,

More information

Measuring the fit of the model - SSR

Measuring the fit of the model - SSR Measuring the fit of the model - SSR Once we ve determined our estimated regression line, we d like to know how well the model fits. How far/close are the observations to the fitted line? One way to do

More information

Lectures on Simple Linear Regression Stat 431, Summer 2012

Lectures on Simple Linear Regression Stat 431, Summer 2012 Lectures on Simple Linear Regression Stat 43, Summer 0 Hyunseung Kang July 6-8, 0 Last Updated: July 8, 0 :59PM Introduction Previously, we have been investigating various properties of the population

More information

STAT 111 Recitation 7

STAT 111 Recitation 7 STAT 111 Recitation 7 Xin Lu Tan xtan@wharton.upenn.edu October 25, 2013 1 / 13 Miscellaneous Please turn in homework 6. Please pick up homework 7 and the graded homework 5. Please check your grade and

More information

The preemptive uniprocessor scheduling of mixed-criticality implicit-deadline sporadic task systems

The preemptive uniprocessor scheduling of mixed-criticality implicit-deadline sporadic task systems The preemptive uniprocessor scheduling of mixed-criticality implicit-deadline sporadic task systems Sanjoy Baruah 1 Vincenzo Bonifaci 2 3 Haohan Li 1 Alberto Marchetti-Spaccamela 4 Suzanne Van Der Ster

More information

Regression, Ridge Regression, Lasso

Regression, Ridge Regression, Lasso Regression, Ridge Regression, Lasso Fabio G. Cozman - fgcozman@usp.br October 2, 2018 A general definition Regression studies the relationship between a response variable Y and covariates X 1,..., X n.

More information

Simple Linear Regression

Simple Linear Regression Simple Linear Regression September 24, 2008 Reading HH 8, GIll 4 Simple Linear Regression p.1/20 Problem Data: Observe pairs (Y i,x i ),i = 1,...n Response or dependent variable Y Predictor or independent

More information

1 Introduction to Minitab

1 Introduction to Minitab 1 Introduction to Minitab Minitab is a statistical analysis software package. The software is freely available to all students and is downloadable through the Technology Tab at my.calpoly.edu. When you

More information

23. Inference for regression

23. Inference for regression 23. Inference for regression The Practice of Statistics in the Life Sciences Third Edition 2014 W. H. Freeman and Company Objectives (PSLS Chapter 23) Inference for regression The regression model Confidence

More information

Small Area Confidence Bounds on Small Cell Proportions in Survey Populations

Small Area Confidence Bounds on Small Cell Proportions in Survey Populations Small Area Confidence Bounds on Small Cell Proportions in Survey Populations Aaron Gilary, Jerry Maples, U.S. Census Bureau U.S. Census Bureau Eric V. Slud, U.S. Census Bureau Univ. Maryland College Park

More information

9. Linear Regression and Correlation

9. Linear Regression and Correlation 9. Linear Regression and Correlation Data: y a quantitative response variable x a quantitative explanatory variable (Chap. 8: Recall that both variables were categorical) For example, y = annual income,

More information

Lecture 2: August 31

Lecture 2: August 31 0-704: Information Processing and Learning Fall 206 Lecturer: Aarti Singh Lecture 2: August 3 Note: These notes are based on scribed notes from Spring5 offering of this course. LaTeX template courtesy

More information

STAT 512 sp 2018 Summary Sheet

STAT 512 sp 2018 Summary Sheet STAT 5 sp 08 Summary Sheet Karl B. Gregory Spring 08. Transformations of a random variable Let X be a rv with support X and let g be a function mapping X to Y with inverse mapping g (A = {x X : g(x A}

More information

Math 423/533: The Main Theoretical Topics

Math 423/533: The Main Theoretical Topics Math 423/533: The Main Theoretical Topics Notation sample size n, data index i number of predictors, p (p = 2 for simple linear regression) y i : response for individual i x i = (x i1,..., x ip ) (1 p)

More information

Su Liu 1, Alexandros Papakonstantinou 2, Hongjun Wang 1,DemingChen 2

Su Liu 1, Alexandros Papakonstantinou 2, Hongjun Wang 1,DemingChen 2 Real-Time Object Tracking System on FPGAs Su Liu 1, Alexandros Papakonstantinou 2, Hongjun Wang 1,DemingChen 2 1 School of Information Science and Engineering, Shandong University, Jinan, China 2 Electrical

More information

Math 2311 Written Homework 6 (Sections )

Math 2311 Written Homework 6 (Sections ) Math 2311 Written Homework 6 (Sections 5.4 5.6) Name: PeopleSoft ID: Instructions: Homework will NOT be accepted through email or in person. Homework must be submitted through CourseWare BEFORE the deadline.

More information

Runtime Model Predictive Verification on Embedded Platforms 1

Runtime Model Predictive Verification on Embedded Platforms 1 Runtime Model Predictive Verification on Embedded Platforms 1 Pei Zhang, Jianwen Li, Joseph Zambreno, Phillip H. Jones, Kristin Yvonne Rozier Presenter: Pei Zhang Iowa State University peizhang@iastate.edu

More information

Information Theory, Statistics, and Decision Trees

Information Theory, Statistics, and Decision Trees Information Theory, Statistics, and Decision Trees Léon Bottou COS 424 4/6/2010 Summary 1. Basic information theory. 2. Decision trees. 3. Information theory and statistics. Léon Bottou 2/31 COS 424 4/6/2010

More information

An Introduction to Sparse Approximation

An Introduction to Sparse Approximation An Introduction to Sparse Approximation Anna C. Gilbert Department of Mathematics University of Michigan Basic image/signal/data compression: transform coding Approximate signals sparsely Compress images,

More information

STA 414/2104: Lecture 8

STA 414/2104: Lecture 8 STA 414/2104: Lecture 8 6-7 March 2017: Continuous Latent Variable Models, Neural networks Delivered by Mark Ebden With thanks to Russ Salakhutdinov, Jimmy Ba and others Outline Continuous latent variable

More information

1-Way ANOVA MATH 143. Spring Department of Mathematics and Statistics Calvin College

1-Way ANOVA MATH 143. Spring Department of Mathematics and Statistics Calvin College 1-Way ANOVA MATH 143 Department of Mathematics and Statistics Calvin College Spring 2010 The basic ANOVA situation Two variables: 1 Categorical, 1 Quantitative Main Question: Do the (means of) the quantitative

More information

STA 414/2104: Lecture 8

STA 414/2104: Lecture 8 STA 414/2104: Lecture 8 6-7 March 2017: Continuous Latent Variable Models, Neural networks With thanks to Russ Salakhutdinov, Jimmy Ba and others Outline Continuous latent variable models Background PCA

More information

STA 302 H1F / 1001 HF Fall 2007 Test 1 October 24, 2007

STA 302 H1F / 1001 HF Fall 2007 Test 1 October 24, 2007 STA 302 H1F / 1001 HF Fall 2007 Test 1 October 24, 2007 LAST NAME: SOLUTIONS FIRST NAME: STUDENT NUMBER: ENROLLED IN: (circle one) STA 302 STA 1001 INSTRUCTIONS: Time: 90 minutes Aids allowed: calculator.

More information

STAT 704 Sections IRLS and Bootstrap

STAT 704 Sections IRLS and Bootstrap STAT 704 Sections 11.4-11.5. IRLS and John Grego Department of Statistics, University of South Carolina Stat 704: Data Analysis I 1 / 14 LOWESS IRLS LOWESS LOWESS (LOcally WEighted Scatterplot Smoothing)

More information

Recent Developments in Compressed Sensing

Recent Developments in Compressed Sensing Recent Developments in Compressed Sensing M. Vidyasagar Distinguished Professor, IIT Hyderabad m.vidyasagar@iith.ac.in, www.iith.ac.in/ m vidyasagar/ ISL Seminar, Stanford University, 19 April 2018 Outline

More information

Stat 5102 Final Exam May 14, 2015

Stat 5102 Final Exam May 14, 2015 Stat 5102 Final Exam May 14, 2015 Name Student ID The exam is closed book and closed notes. You may use three 8 1 11 2 sheets of paper with formulas, etc. You may also use the handouts on brand name distributions

More information

ANOVA: Analysis of Variation

ANOVA: Analysis of Variation ANOVA: Analysis of Variation The basic ANOVA situation Two variables: 1 Categorical, 1 Quantitative Main Question: Do the (means of) the quantitative variables depend on which group (given by categorical

More information

2. the basis functions have different symmetries. 1 k = 0. x( t) 1 t 0 x(t) 0 t 1

2. the basis functions have different symmetries. 1 k = 0. x( t) 1 t 0 x(t) 0 t 1 In the next few lectures, we will look at a few examples of orthobasis expansions that are used in modern signal processing. Cosine transforms The cosine-i transform is an alternative to Fourier series;

More information

STAT 4385 Topic 03: Simple Linear Regression

STAT 4385 Topic 03: Simple Linear Regression STAT 4385 Topic 03: Simple Linear Regression Xiaogang Su, Ph.D. Department of Mathematical Science University of Texas at El Paso xsu@utep.edu Spring, 2017 Outline The Set-Up Exploratory Data Analysis

More information

The Perceptron algorithm

The Perceptron algorithm The Perceptron algorithm Tirgul 3 November 2016 Agnostic PAC Learnability A hypothesis class H is agnostic PAC learnable if there exists a function m H : 0,1 2 N and a learning algorithm with the following

More information

Sequence Modelling with Features: Linear-Chain Conditional Random Fields. COMP-599 Oct 6, 2015

Sequence Modelling with Features: Linear-Chain Conditional Random Fields. COMP-599 Oct 6, 2015 Sequence Modelling with Features: Linear-Chain Conditional Random Fields COMP-599 Oct 6, 2015 Announcement A2 is out. Due Oct 20 at 1pm. 2 Outline Hidden Markov models: shortcomings Generative vs. discriminative

More information

Single and multiple linear regression analysis

Single and multiple linear regression analysis Single and multiple linear regression analysis Marike Cockeran 2017 Introduction Outline of the session Simple linear regression analysis SPSS example of simple linear regression analysis Additional topics

More information

Applied Regression Modeling: A Business Approach Chapter 3: Multiple Linear Regression Sections

Applied Regression Modeling: A Business Approach Chapter 3: Multiple Linear Regression Sections Applied Regression Modeling: A Business Approach Chapter 3: Multiple Linear Regression Sections 3.4 3.6 by Iain Pardoe 3.4 Model assumptions 2 Regression model assumptions.............................................

More information

Part 6: Multivariate Normal and Linear Models

Part 6: Multivariate Normal and Linear Models Part 6: Multivariate Normal and Linear Models 1 Multiple measurements Up until now all of our statistical models have been univariate models models for a single measurement on each member of a sample of

More information

The simple linear regression model discussed in Chapter 13 was written as

The simple linear regression model discussed in Chapter 13 was written as 1519T_c14 03/27/2006 07:28 AM Page 614 Chapter Jose Luis Pelaez Inc/Blend Images/Getty Images, Inc./Getty Images, Inc. 14 Multiple Regression 14.1 Multiple Regression Analysis 14.2 Assumptions of the Multiple

More information

Statistical Data Analysis

Statistical Data Analysis DS-GA 0 Lecture notes 8 Fall 016 1 Descriptive statistics Statistical Data Analysis In this section we consider the problem of analyzing a set of data. We describe several techniques for visualizing the

More information

on a per-coecient basis in large images is computationally expensive. Further, the algorithm in [CR95] needs to be rerun, every time a new rate of com

on a per-coecient basis in large images is computationally expensive. Further, the algorithm in [CR95] needs to be rerun, every time a new rate of com Extending RD-OPT with Global Thresholding for JPEG Optimization Viresh Ratnakar University of Wisconsin-Madison Computer Sciences Department Madison, WI 53706 Phone: (608) 262-6627 Email: ratnakar@cs.wisc.edu

More information

IMAGE COMPRESSION-II. Week IX. 03/6/2003 Image Compression-II 1

IMAGE COMPRESSION-II. Week IX. 03/6/2003 Image Compression-II 1 IMAGE COMPRESSION-II Week IX 3/6/23 Image Compression-II 1 IMAGE COMPRESSION Data redundancy Self-information and Entropy Error-free and lossy compression Huffman coding Predictive coding Transform coding

More information

Homework 1: Solutions

Homework 1: Solutions Homework 1: Solutions Statistics 413 Fall 2017 Data Analysis: Note: All data analysis results are provided by Michael Rodgers 1. Baseball Data: (a) What are the most important features for predicting players

More information