Tolerating Broken Robots

Size: px
Start display at page:

Download "Tolerating Broken Robots"

Transcription

1 Tolerating Broken Robots Lachlan Murray, Jon Timmis and Andy Tyrrell Intelligent Systems Group Department of Electronics University of York, UK March 24, 2011 People say my broken friend is useless, but I say his mind is free. There s a lot of things my mangled robot friend could be... - Bender Bending Rodríguez (3003)

2 Outline SYMBRION IR anomaly detection Future...

3 Symbiotic Evolutionary Robot Organisms Swarm Robotic System Re-configurable Robotic System Collective Robotic System Heterogeneous, re-configurable, collective robotic system Robots share energy and computational resources Terminology: Individual/Module - a single independent robot Swarm - multiple cooperating individuals Organism - multiple connected individuals Collective - combination of individuals, swarms and organisms

4 SYMBRION Robots I Three specialised modules: I I I Active Wheel Scout Robot Backbone Robot I Common docking interface I 15 working robots by May! (sort of)

5 SYMBRION Grand Challenges GC1-100 Robots 100 Days Long-term survival in a changing environment GC2 - The Emergence of Multi-cellularity Online, on-board evolution

6 IR Anomaly Detection Immune-inspired anomaly detection using the modified Dendritic Cell Algorithm (mdca) Detecting anomalies in the infrared sensors of simulated SYMBRION-style robots Optimised using NSGA-II Compared with Support Vector Machines (SVM) Performance measured in terms of: Classification accuracy Execution time Long-term survival

7 Simulator Extended version of Stage (in collab. UWE) SYMBRION-style robots Realistic IR data Energy sharing capabilities

8 Behavioural Controller Wait Wander Reverse Un-dock Approach Recharge Provide Recover Avoid Align Dock Default behaviour: random wandering with obstacle avoidance Every state transitions to Recover when motors stall

9 Energy Sharing Strategy Wait Wander Reverse Un-dock Approach Recharge Provide Recover Avoid Align Dock If moderately in need of energy - recharge If desperately in need of energy - wait If another robot is in need - provide

10 Recovery Strategy If sensor is returning anomalous data take value of neighbouring sensor If neighbour is returning anomalous data panic assume a small value If either of the front two sensor are returning anomalous data prevent controller from entering Approach state Important - this strategy is not perfect

11 (Video not included)

12 Anomalies Anomalies in IR sensor data may originate from: Faulty sensors Interference from other robots Other environmental factors Here we consider transient faults in the sensors Three types of fault are simulated: 1. Stuck-at-value - sensor always returns the same value, regardless of the state of the environment being measured 2. Sensor noise - the value returned by the sensor is a random amount away from the ideal value 3. Sensor bias - the value returned by the sensor is consistently a fixed amount away from the value it should be

13 mdca Mai s algorithm Immune-inspired anomaly detection algorithm Simplified version of the DCA - for resource constrained systems Inspired by danger theory Behaviour determined by five parameters

14 mdca Mai s algorithm Immune-inspired anomaly detection algorithm Simplified version of the DCA - for resource constrained systems σ i (t) = waa i (t)+w b B i (t)+w cc i (t) w a+w b +w c φ i (t) = { 0 σ i (t) < τ 1 σ i (t) τ Inspired by danger theory c i (t) = 1 ω φi (k) Behaviour determined by five parameters

15 mdca Mai s algorithm Immune-inspired anomaly detection algorithm Simplified version of the DCA - for resource constrained systems σ i (t) = waa i (t)+w b B i (t)+w cc i (t) w a+w b +w c φ i (t) = { 0 σ i (t) < τ 1 σ i (t) τ Inspired by danger theory c i (t) = 1 ω φi (k) Behaviour determined by five parameters

16 mdca Mai s algorithm Immune-inspired anomaly detection algorithm Simplified version of the DCA - for resource constrained systems σ i (t) = waa i (t)+w b B i (t)+w cc i (t) w a+w b +w c φ i (t) = { 0 σ i (t) < τ 1 σ i (t) τ Inspired by danger theory c i (t) = 1 ω φi (k) Behaviour determined by five parameters

17 mdca Mai s algorithm Immune-inspired anomaly detection algorithm Simplified version of the DCA - for resource constrained systems σ i (t) = waa i (t)+w b B i (t)+w cc i (t) w a+w b +w c φ i (t) = { 0 σ i (t) < τ 1 σ i (t) τ Inspired by danger theory c i (t) = 1 ω φi (k) Behaviour determined by five parameters

18 mdca Mai s algorithm Immune-inspired anomaly detection algorithm Simplified version of the DCA - for resource constrained systems σ i (t) = waa i (t)+w b B i (t)+w cc i (t) w a+w b +w c φ i (t) = { 0 σ i (t) < τ 1 σ i (t) τ Inspired by danger theory c i (t) = 1 ω φi (k) Behaviour determined by five parameters

19 Feature Extraction σ i (t) = waa i (t)+w b B i (t)+w cc i (t) w a+w b +w c A i (t), B i (t) and C i (t) are features extracted from sensor data Example features include: Distance between neighbouring sensor values Mean of recent sensor values Standard deviation Skewness Six more parameters: fa, f b, f c - types of feature ta, t b, t c - associated time windows

20 Support Vector Machines Standard classification technique Given a set of labelled training data (features) In a training phase SVMs construct a hyperplane that maximises distance between classes New data instances are classified according to which side of the plane they are situated

21 Multi-objective Parameter Optimisation Objectives of a resource constrained anomaly detector: Maximise true positives whilst minimising false positives Maximise the speed of detection without loss of accuracy Minimise the computational cost of the system Three experiments: 1. mdca with fixed features 2. mdca with evolvable features 3. SVM with evolvable features NSGA-II used for all three

22 mdca with Fixed Features Fixed features: fa - distance between neighbouring sensor values fb - standard deviation of recent history fc - binary distance between current and recent mean t a fixed to 1, t b and t c evolved as one parameter: t bc All other mdca parameters evolved Objective 1 - minimise the distance between ideal and actual output when data is anomalous Objective 2 - minimise the distance between ideal and actual output when data is normal Parameter t bc ω w a w b w c τ Range (1, 1000) (-100, 100)

23 mdca with Fixed Features - results Bad FPR (relatively) Good TPR (Video not included) Measure TPR FPR PPV NPV ACC Value

24 mdca with Fixed Features - example output mdca output Ideal output Raw sensor value mdca output Ideal output Raw sensor value Time x 10 4

25 (Video not included)

26 mdca with Evolvable Features All feature types and window sizes evolved All other mdca parameters evolved Objective 1 - minimise the distance between ideal and actual output when data is anomalous Objective 2 - minimise the distance between ideal and actual output when data is normal Parameter f a f b f c t a t b t c ω w a w b w c τ Range (0, 9) (1, 500) (-100, 100)

27 mdca with Evolvable Features - results Good FPR Reasonable TPR (Video not included) Measure TPR FPR PPV NPV ACC Value

28 mdca with Evolvable Features - example output mdca output Ideal output Raw sensor value Time x 10 4 mdca output Ideal output Raw sensor value

29 SVM with Evolvable Features All feature types and window sizes evolved Objective 1 - Minimise one minus the true positive rate Objective 2 - Minimise the false positive rate Parameter f a f b f c t a t b t c Range (0, 9) (1, 500)

30 SVM with Evolvable Features - results Reasonable FPR Good TPR (Video not included) Measure TPR FPR PPV NPV ACC Value

31 SVM with Evolvable Features - example output mdca output Ideal output Raw sensor value SVM output Ideal output Time x 10 4 Raw sensor value

32 Comparisons Experiment mdca-i mdca-ii SVM Mean Sim. Time (s)

33 Comparisons Summary mdca-i/mdca-ii mdca-i/svm mdca-ii/svm TPR mdca-i - SVM FPR mdca-ii SVM mdca-ii PPV mdca-ii SVM mdca-ii NPV mdca-i - SVM ACC mdca-ii SVM - Speed mdca-i mdca-i mdca-ii No single approach outperforms the others for all measures In terms of classification SVM is the most balanced but execution and optimisation time was slow

34 Long-term survival 50 robots operating over a 10 hour period Robots assigned random starting positions and energy levels 12 anomalies injected into each robot Six different approaches investigated: No anomalies - control No anomaly detection - baseline Ideal anomaly detection - gold standard? mdca with fixed features mdca with evolvable features SVM with evolvable features

35 Long-term survival - mean stored energy 1.8 x Energy No anom aly detection No anom alies m DCA I Idealanom aly detection m DCA II SVM Time SVM performed best but not significantly better than the rest Why is the ideal detector not better? Deficiencies in the recovery strategy mean that in certain situations the best response to an anomaly is to ignore it

36 Long-term survival - surviving number of robots Num berofrobots No anom aly detection No anom alies m DCA I Idealanom aly detection m DCA II SVM Time SVM performed best but not significantly better than the rest

37 Summary The parameters of three mdca and SVM based anomaly detection systems were optimised using NSGA-II Comparisons were made in terms of accuracy, speed of execution and long-term survival No system outperformed all others in terms of accuracy - though the SVM-based system was the most balanced Both mdca systems significantly outperformed the SVM in terms of speed, and informally in terms of evolvability The importance of a good recovery strategy was highlighted by the long-term survival experiments

38 (Video not included)

39 Thanks!

40 True Positive Rate (TPR) True Negative Rate (TNR) False Positive Rate (FPR) Positive Predictive Value (PPV) Negative Predictive Value (NPV) Accuracy (ACC) TP (TP+FN) TN (TN+FP) 1 TNR TP (TP+FP) TN (TN+FN) TP+TN (TP+FP+TN+FN)

41 Standard Deviation σ = 1 (N 1) (xi x) 2 Average distance distave = x µ Binary average distance bdistave = { 0 distave τ 1 distave > τ Distance from neighbours distn = µ a (µ b + µ c )/2 Skew skew = 1 N ((xi µ)/σ) 3 Mean µ = 1 N xi StdDev distance diststddev = σ a σ b Range range = max(x ) min(x ) Pair distance pairdist = 1 N xi+1 x i

Performance Measures. Sören Sonnenburg. Fraunhofer FIRST.IDA, Kekuléstr. 7, Berlin, Germany

Performance Measures. Sören Sonnenburg. Fraunhofer FIRST.IDA, Kekuléstr. 7, Berlin, Germany Sören Sonnenburg Fraunhofer FIRST.IDA, Kekuléstr. 7, 2489 Berlin, Germany Roadmap: Contingency Table Scores from the Contingency Table Curves from the Contingency Table Discussion Sören Sonnenburg Contingency

More information

Evaluation & Credibility Issues

Evaluation & Credibility Issues Evaluation & Credibility Issues What measure should we use? accuracy might not be enough. How reliable are the predicted results? How much should we believe in what was learned? Error on the training data

More information

Plan. Lecture: What is Chemoinformatics and Drug Design? Description of Support Vector Machine (SVM) and its used in Chemoinformatics.

Plan. Lecture: What is Chemoinformatics and Drug Design? Description of Support Vector Machine (SVM) and its used in Chemoinformatics. Plan Lecture: What is Chemoinformatics and Drug Design? Description of Support Vector Machine (SVM) and its used in Chemoinformatics. Exercise: Example and exercise with herg potassium channel: Use of

More information

Machine Learning Concepts in Chemoinformatics

Machine Learning Concepts in Chemoinformatics Machine Learning Concepts in Chemoinformatics Martin Vogt B-IT Life Science Informatics Rheinische Friedrich-Wilhelms-Universität Bonn BigChem Winter School 2017 25. October Data Mining in Chemoinformatics

More information

Model Accuracy Measures

Model Accuracy Measures Model Accuracy Measures Master in Bioinformatics UPF 2017-2018 Eduardo Eyras Computational Genomics Pompeu Fabra University - ICREA Barcelona, Spain Variables What we can measure (attributes) Hypotheses

More information

Anomaly Detection for the CERN Large Hadron Collider injection magnets

Anomaly Detection for the CERN Large Hadron Collider injection magnets Anomaly Detection for the CERN Large Hadron Collider injection magnets Armin Halilovic KU Leuven - Department of Computer Science In cooperation with CERN 2018-07-27 0 Outline 1 Context 2 Data 3 Preprocessing

More information

Performance Evaluation

Performance Evaluation Performance Evaluation Confusion Matrix: Detected Positive Negative Actual Positive A: True Positive B: False Negative Negative C: False Positive D: True Negative Recall or Sensitivity or True Positive

More information

Linear Classifiers as Pattern Detectors

Linear Classifiers as Pattern Detectors Intelligent Systems: Reasoning and Recognition James L. Crowley ENSIMAG 2 / MoSIG M1 Second Semester 2014/2015 Lesson 16 8 April 2015 Contents Linear Classifiers as Pattern Detectors Notation...2 Linear

More information

Time Series Classification

Time Series Classification Distance Measures Classifiers DTW vs. ED Further Work Questions August 31, 2017 Distance Measures Classifiers DTW vs. ED Further Work Questions Outline 1 2 Distance Measures 3 Classifiers 4 DTW vs. ED

More information

Linear Classifiers as Pattern Detectors

Linear Classifiers as Pattern Detectors Intelligent Systems: Reasoning and Recognition James L. Crowley ENSIMAG 2 / MoSIG M1 Second Semester 2013/2014 Lesson 18 23 April 2014 Contents Linear Classifiers as Pattern Detectors Notation...2 Linear

More information

Big Data Analytics: Evaluating Classification Performance April, 2016 R. Bohn. Some overheads from Galit Shmueli and Peter Bruce 2010

Big Data Analytics: Evaluating Classification Performance April, 2016 R. Bohn. Some overheads from Galit Shmueli and Peter Bruce 2010 Big Data Analytics: Evaluating Classification Performance April, 2016 R. Bohn 1 Some overheads from Galit Shmueli and Peter Bruce 2010 Most accurate Best! Actual value Which is more accurate?? 2 Why Evaluate

More information

Engineering Part IIB: Module 4F10 Statistical Pattern Processing Lecture 5: Single Layer Perceptrons & Estimating Linear Classifiers

Engineering Part IIB: Module 4F10 Statistical Pattern Processing Lecture 5: Single Layer Perceptrons & Estimating Linear Classifiers Engineering Part IIB: Module 4F0 Statistical Pattern Processing Lecture 5: Single Layer Perceptrons & Estimating Linear Classifiers Phil Woodland: pcw@eng.cam.ac.uk Michaelmas 202 Engineering Part IIB:

More information

Reducing False Alarm Rate in Anomaly Detection with Layered Filtering

Reducing False Alarm Rate in Anomaly Detection with Layered Filtering Reducing False Alarm Rate in Anomaly Detection with Layered Filtering Rafa l Pokrywka 1,2 1 Institute of Computer Science AGH University of Science and Technology al. Mickiewicza 30, 30-059 Kraków, Poland

More information

CS 188: Artificial Intelligence. Outline

CS 188: Artificial Intelligence. Outline CS 188: Artificial Intelligence Lecture 21: Perceptrons Pieter Abbeel UC Berkeley Many slides adapted from Dan Klein. Outline Generative vs. Discriminative Binary Linear Classifiers Perceptron Multi-class

More information

Errors, and What to Do. CS 188: Artificial Intelligence Fall What to Do About Errors. Later On. Some (Simplified) Biology

Errors, and What to Do. CS 188: Artificial Intelligence Fall What to Do About Errors. Later On. Some (Simplified) Biology CS 188: Artificial Intelligence Fall 2011 Lecture 22: Perceptrons and More! 11/15/2011 Dan Klein UC Berkeley Errors, and What to Do Examples of errors Dear GlobalSCAPE Customer, GlobalSCAPE has partnered

More information

CS 188: Artificial Intelligence Fall 2011

CS 188: Artificial Intelligence Fall 2011 CS 188: Artificial Intelligence Fall 2011 Lecture 22: Perceptrons and More! 11/15/2011 Dan Klein UC Berkeley Errors, and What to Do Examples of errors Dear GlobalSCAPE Customer, GlobalSCAPE has partnered

More information

Pattern Recognition and Machine Learning. Learning and Evaluation for Pattern Recognition

Pattern Recognition and Machine Learning. Learning and Evaluation for Pattern Recognition Pattern Recognition and Machine Learning Jaes L. Crowley ENSIMAG 3 - MMIS Fall Seester 2017 Lesson 1 4 October 2017 Outline Learning and Evaluation for Pattern Recognition Notation...2 1. The Pattern Recognition

More information

CS 343: Artificial Intelligence

CS 343: Artificial Intelligence CS 343: Artificial Intelligence Perceptrons Prof. Scott Niekum The University of Texas at Austin [These slides based on those of Dan Klein and Pieter Abbeel for CS188 Intro to AI at UC Berkeley. All CS188

More information

CC283 Intelligent Problem Solving 28/10/2013

CC283 Intelligent Problem Solving 28/10/2013 Machine Learning What is the research agenda? How to measure success? How to learn? Machine Learning Overview Unsupervised Learning Supervised Learning Training Testing Unseen data Data Observed x 1 x

More information

Anomaly Detection. Jing Gao. SUNY Buffalo

Anomaly Detection. Jing Gao. SUNY Buffalo Anomaly Detection Jing Gao SUNY Buffalo 1 Anomaly Detection Anomalies the set of objects are considerably dissimilar from the remainder of the data occur relatively infrequently when they do occur, their

More information

MIRA, SVM, k-nn. Lirong Xia

MIRA, SVM, k-nn. Lirong Xia MIRA, SVM, k-nn Lirong Xia Linear Classifiers (perceptrons) Inputs are feature values Each feature has a weight Sum is the activation activation w If the activation is: Positive: output +1 Negative, output

More information

Evolutionary Multiobjective. Optimization Methods for the Shape Design of Industrial Electromagnetic Devices. P. Di Barba, University of Pavia, Italy

Evolutionary Multiobjective. Optimization Methods for the Shape Design of Industrial Electromagnetic Devices. P. Di Barba, University of Pavia, Italy Evolutionary Multiobjective Optimization Methods for the Shape Design of Industrial Electromagnetic Devices P. Di Barba, University of Pavia, Italy INTRODUCTION Evolutionary Multiobjective Optimization

More information

Multiple regression: Categorical dependent variables

Multiple regression: Categorical dependent variables Multiple : Categorical Johan A. Elkink School of Politics & International Relations University College Dublin 28 November 2016 1 2 3 4 Outline 1 2 3 4 models models have a variable consisting of two categories.

More information

An Overview of Outlier Detection Techniques and Applications

An Overview of Outlier Detection Techniques and Applications Machine Learning Rhein-Neckar Meetup An Overview of Outlier Detection Techniques and Applications Ying Gu connygy@gmail.com 28.02.2016 Anomaly/Outlier Detection What are anomalies/outliers? The set of

More information

Online Advertising is Big Business

Online Advertising is Big Business Online Advertising Online Advertising is Big Business Multiple billion dollar industry $43B in 2013 in USA, 17% increase over 2012 [PWC, Internet Advertising Bureau, April 2013] Higher revenue in USA

More information

Chapter 1 Introduction

Chapter 1 Introduction Chapter 1 Introduction 1.1 Introduction to Chapter This chapter starts by describing the problems addressed by the project. The aims and objectives of the research are outlined and novel ideas discovered

More information

Swarm-bots. Marco Dorigo FNRS Research Director IRIDIA Université Libre de Bruxelles

Swarm-bots. Marco Dorigo FNRS Research Director IRIDIA Université Libre de Bruxelles Swarm-bots Marco Dorigo FNRS Research Director IRIDIA Université Libre de Bruxelles Swarm-bots The swarm-bot is an experiment in swarm robotics Swarm robotics is the application of swarm intelligence principles

More information

Support Vector Machine. Industrial AI Lab.

Support Vector Machine. Industrial AI Lab. Support Vector Machine Industrial AI Lab. Classification (Linear) Autonomously figure out which category (or class) an unknown item should be categorized into Number of categories / classes Binary: 2 different

More information

We provide two sections from the book (in preparation) Intelligent and Autonomous Road Vehicles, by Ozguner, Acarman and Redmill.

We provide two sections from the book (in preparation) Intelligent and Autonomous Road Vehicles, by Ozguner, Acarman and Redmill. We provide two sections from the book (in preparation) Intelligent and Autonomous Road Vehicles, by Ozguner, Acarman and Redmill. 2.3.2. Steering control using point mass model: Open loop commands We consider

More information

Anticipating Visual Representations from Unlabeled Data. Carl Vondrick, Hamed Pirsiavash, Antonio Torralba

Anticipating Visual Representations from Unlabeled Data. Carl Vondrick, Hamed Pirsiavash, Antonio Torralba Anticipating Visual Representations from Unlabeled Data Carl Vondrick, Hamed Pirsiavash, Antonio Torralba Overview Problem Key Insight Methods Experiments Problem: Predict future actions and objects Image

More information

Linear vs Non-linear classifier. CS789: Machine Learning and Neural Network. Introduction

Linear vs Non-linear classifier. CS789: Machine Learning and Neural Network. Introduction Linear vs Non-linear classifier CS789: Machine Learning and Neural Network Support Vector Machine Jakramate Bootkrajang Department of Computer Science Chiang Mai University Linear classifier is in the

More information

Moving Average Rules to Find. Confusion Matrix. CC283 Intelligent Problem Solving 05/11/2010. Edward Tsang (all rights reserved) 1

Moving Average Rules to Find. Confusion Matrix. CC283 Intelligent Problem Solving 05/11/2010. Edward Tsang (all rights reserved) 1 Machine Learning Overview Supervised Learning Training esting Te Unseen data Data Observed x 1 x 2... x n 1.6 7.1... 2.7 1.4 6.8... 3.1 2.1 5.4... 2.8... Machine Learning Patterns y = f(x) Target y Buy

More information

Real-time image-based parking occupancy detection using deep learning. Debaditya Acharya, Weilin Yan & Kourosh Khoshelham The University of Melbourne

Real-time image-based parking occupancy detection using deep learning. Debaditya Acharya, Weilin Yan & Kourosh Khoshelham The University of Melbourne Real-time image-based parking occupancy detection using deep learning Debaditya Acharya, Weilin Yan & Kourosh Khoshelham The University of Melbourne Slide 1/20 Prologue People spend on Does average that

More information

CS 188: Artificial Intelligence Fall 2008

CS 188: Artificial Intelligence Fall 2008 CS 188: Artificial Intelligence Fall 2008 Lecture 23: Perceptrons 11/20/2008 Dan Klein UC Berkeley 1 General Naïve Bayes A general naive Bayes model: C E 1 E 2 E n We only specify how each feature depends

More information

General Naïve Bayes. CS 188: Artificial Intelligence Fall Example: Overfitting. Example: OCR. Example: Spam Filtering. Example: Spam Filtering

General Naïve Bayes. CS 188: Artificial Intelligence Fall Example: Overfitting. Example: OCR. Example: Spam Filtering. Example: Spam Filtering CS 188: Artificial Intelligence Fall 2008 General Naïve Bayes A general naive Bayes model: C Lecture 23: Perceptrons 11/20/2008 E 1 E 2 E n Dan Klein UC Berkeley We only specify how each feature depends

More information

CS798: Selected topics in Machine Learning

CS798: Selected topics in Machine Learning CS798: Selected topics in Machine Learning Support Vector Machine Jakramate Bootkrajang Department of Computer Science Chiang Mai University Jakramate Bootkrajang CS798: Selected topics in Machine Learning

More information

Sparse Kernel Machines - SVM

Sparse Kernel Machines - SVM Sparse Kernel Machines - SVM Henrik I. Christensen Robotics & Intelligent Machines @ GT Georgia Institute of Technology, Atlanta, GA 30332-0280 hic@cc.gatech.edu Henrik I. Christensen (RIM@GT) Support

More information

CS 5522: Artificial Intelligence II

CS 5522: Artificial Intelligence II CS 5522: Artificial Intelligence II Perceptrons Instructor: Alan Ritter Ohio State University [These slides were adapted from CS188 Intro to AI at UC Berkeley. All materials available at http://ai.berkeley.edu.]

More information

International Journal of Scientific & Engineering Research, Volume 7, Issue 2, February ISSN

International Journal of Scientific & Engineering Research, Volume 7, Issue 2, February ISSN International Journal of Scientific & Engineering Research, Volume 7, Issue 2, February-2016 9 Automated Methodology for Context Based Semantic Anomaly Identification in Big Data Hema.R 1, Vidhya.V 2,

More information

Applying Machine Learning for Gravitational-wave Burst Data Analysis

Applying Machine Learning for Gravitational-wave Burst Data Analysis Applying Machine Learning for Gravitational-wave Burst Data Analysis Junwei Cao LIGO Scientific Collaboration Research Group Research Institute of Information Technology Tsinghua University June 29, 2016

More information

Support Vector Machine. Industrial AI Lab. Prof. Seungchul Lee

Support Vector Machine. Industrial AI Lab. Prof. Seungchul Lee Support Vector Machine Industrial AI Lab. Prof. Seungchul Lee Classification (Linear) Autonomously figure out which category (or class) an unknown item should be categorized into Number of categories /

More information

Lecture Slides for INTRODUCTION TO. Machine Learning. ETHEM ALPAYDIN The MIT Press,

Lecture Slides for INTRODUCTION TO. Machine Learning. ETHEM ALPAYDIN The MIT Press, Lecture Slides for INTRODUCTION TO Machine Learning ETHEM ALPAYDIN The MIT Press, 2004 alpaydin@boun.edu.tr http://www.cmpe.boun.edu.tr/~ethem/i2ml CHAPTER 14: Assessing and Comparing Classification Algorithms

More information

Learning from Corrupted Binary Labels via Class-Probability Estimation

Learning from Corrupted Binary Labels via Class-Probability Estimation Learning from Corrupted Binary Labels via Class-Probability Estimation Aditya Krishna Menon Brendan van Rooyen Cheng Soon Ong Robert C. Williamson xxx National ICT Australia and The Australian National

More information

CSC314 / CSC763 Introduction to Machine Learning

CSC314 / CSC763 Introduction to Machine Learning CSC314 / CSC763 Introduction to Machine Learning COMSATS Institute of Information Technology Dr. Adeel Nawab More on Evaluating Hypotheses/Learning Algorithms Lecture Outline: Review of Confidence Intervals

More information

Swarm-bots and Swarmanoid: Two experiments in embodied swarm intelligence

Swarm-bots and Swarmanoid: Two experiments in embodied swarm intelligence Swarm-bots and Swarmanoid: Two experiments in embodied swarm intelligence Marco Dorigo FNRS Research Director IRIDIA Université Libre de Bruxelles IAT - 17.9.2009 - Milano, Italy What is swarm intelligence?

More information

CS4495/6495 Introduction to Computer Vision. 8C-L3 Support Vector Machines

CS4495/6495 Introduction to Computer Vision. 8C-L3 Support Vector Machines CS4495/6495 Introduction to Computer Vision 8C-L3 Support Vector Machines Discriminative classifiers Discriminative classifiers find a division (surface) in feature space that separates the classes Several

More information

Performance Evaluation

Performance Evaluation Performance Evaluation David S. Rosenberg Bloomberg ML EDU October 26, 2017 David S. Rosenberg (Bloomberg ML EDU) October 26, 2017 1 / 36 Baseline Models David S. Rosenberg (Bloomberg ML EDU) October 26,

More information

Tackling the Poor Assumptions of Naive Bayes Text Classifiers

Tackling the Poor Assumptions of Naive Bayes Text Classifiers Tackling the Poor Assumptions of Naive Bayes Text Classifiers Jason Rennie MIT Computer Science and Artificial Intelligence Laboratory jrennie@ai.mit.edu Joint work with Lawrence Shih, Jaime Teevan and

More information

Hypothesis Evaluation

Hypothesis Evaluation Hypothesis Evaluation Machine Learning Hamid Beigy Sharif University of Technology Fall 1395 Hamid Beigy (Sharif University of Technology) Hypothesis Evaluation Fall 1395 1 / 31 Table of contents 1 Introduction

More information

Pattern Recognition 2018 Support Vector Machines

Pattern Recognition 2018 Support Vector Machines Pattern Recognition 2018 Support Vector Machines Ad Feelders Universiteit Utrecht Ad Feelders ( Universiteit Utrecht ) Pattern Recognition 1 / 48 Support Vector Machines Ad Feelders ( Universiteit Utrecht

More information

Power Supply Quality Analysis Using S-Transform and SVM Classifier

Power Supply Quality Analysis Using S-Transform and SVM Classifier Journal of Power and Energy Engineering, 2014, 2, 438-447 Published Online April 2014 in SciRes. http://www.scirp.org/journal/jpee http://dx.doi.org/10.4236/jpee.2014.24059 Power Supply Quality Analysis

More information

OPSIAL Manual. v Xiaofeng Tan. All Rights Reserved

OPSIAL Manual. v Xiaofeng Tan. All Rights Reserved OPSIAL Manual v1.0 2016 Xiaofeng Tan. All Rights Reserved 1. Introduction... 3 1.1 Spectral Calculator & Fitter (SCF)... 3 1.2 Automated Analyzer (AA)... 3 2. Working Principles and Workflows of OPSIAL...

More information

Methods and Criteria for Model Selection. CS57300 Data Mining Fall Instructor: Bruno Ribeiro

Methods and Criteria for Model Selection. CS57300 Data Mining Fall Instructor: Bruno Ribeiro Methods and Criteria for Model Selection CS57300 Data Mining Fall 2016 Instructor: Bruno Ribeiro Goal } Introduce classifier evaluation criteria } Introduce Bias x Variance duality } Model Assessment }

More information

Activity Mining in Sensor Networks

Activity Mining in Sensor Networks MITSUBISHI ELECTRIC RESEARCH LABORATORIES http://www.merl.com Activity Mining in Sensor Networks Christopher R. Wren, David C. Minnen TR2004-135 December 2004 Abstract We present results from the exploration

More information

The Application of Extreme Learning Machine based on Gaussian Kernel in Image Classification

The Application of Extreme Learning Machine based on Gaussian Kernel in Image Classification he Application of Extreme Learning Machine based on Gaussian Kernel in Image Classification Weijie LI, Yi LIN Postgraduate student in College of Survey and Geo-Informatics, tongji university Email: 1633289@tongji.edu.cn

More information

8. Classifier Ensembles for Changing Environments

8. Classifier Ensembles for Changing Environments 1 8. Classifier Ensembles for Changing Environments 8.1. Streaming data and changing environments. 8.2. Approach 1: Change detection. An ensemble method 8.2. Approach 2: Constant updates. Classifier ensembles

More information

Sparse vectors recap. ANLP Lecture 22 Lexical Semantics with Dense Vectors. Before density, another approach to normalisation.

Sparse vectors recap. ANLP Lecture 22 Lexical Semantics with Dense Vectors. Before density, another approach to normalisation. ANLP Lecture 22 Lexical Semantics with Dense Vectors Henry S. Thompson Based on slides by Jurafsky & Martin, some via Dorota Glowacka 5 November 2018 Previous lectures: Sparse vectors recap How to represent

More information

ANLP Lecture 22 Lexical Semantics with Dense Vectors

ANLP Lecture 22 Lexical Semantics with Dense Vectors ANLP Lecture 22 Lexical Semantics with Dense Vectors Henry S. Thompson Based on slides by Jurafsky & Martin, some via Dorota Glowacka 5 November 2018 Henry S. Thompson ANLP Lecture 22 5 November 2018 Previous

More information

Spectral Methods for Subgraph Detection

Spectral Methods for Subgraph Detection Spectral Methods for Subgraph Detection Nadya T. Bliss & Benjamin A. Miller Embedded and High Performance Computing Patrick J. Wolfe Statistics and Information Laboratory Harvard University 12 July 2010

More information

Unit 8: Introduction to neural networks. Perceptrons

Unit 8: Introduction to neural networks. Perceptrons Unit 8: Introduction to neural networks. Perceptrons D. Balbontín Noval F. J. Martín Mateos J. L. Ruiz Reina A. Riscos Núñez Departamento de Ciencias de la Computación e Inteligencia Artificial Universidad

More information

Scribe to lecture Tuesday March

Scribe to lecture Tuesday March Scribe to lecture Tuesday March 16 2004 Scribe outlines: Message Confidence intervals Central limit theorem Em-algorithm Bayesian versus classical statistic Note: There is no scribe for the beginning of

More information

Jeff Howbert Introduction to Machine Learning Winter

Jeff Howbert Introduction to Machine Learning Winter Classification / Regression Support Vector Machines Jeff Howbert Introduction to Machine Learning Winter 2012 1 Topics SVM classifiers for linearly separable classes SVM classifiers for non-linearly separable

More information

Performance evaluation of binary classifiers

Performance evaluation of binary classifiers Performance evaluation of binary classifiers Kevin P. Murphy Last updated October 10, 2007 1 ROC curves We frequently design systems to detect events of interest, such as diseases in patients, faces in

More information

Recent Advances in Flares Prediction and Validation

Recent Advances in Flares Prediction and Validation Recent Advances in Flares Prediction and Validation R. Qahwaji 1, O. Ahmed 1, T. Colak 1, P. Higgins 2, P. Gallagher 2 and S. Bloomfield 2 1 University of Bradford, UK 2 Trinity College Dublin, Ireland

More information

Performance Evaluation and Hypothesis Testing

Performance Evaluation and Hypothesis Testing Performance Evaluation and Hypothesis Testing 1 Motivation Evaluating the performance of learning systems is important because: Learning systems are usually designed to predict the class of future unlabeled

More information

Blind Source Separation Using Artificial immune system

Blind Source Separation Using Artificial immune system American Journal of Engineering Research (AJER) e-issn : 2320-0847 p-issn : 2320-0936 Volume-03, Issue-02, pp-240-247 www.ajer.org Research Paper Open Access Blind Source Separation Using Artificial immune

More information

Anomaly detection in recordings from in-vehicle networks. Andreas Theissler. IT-Designers GmbH, Esslingen, Germany

Anomaly detection in recordings from in-vehicle networks. Andreas Theissler. IT-Designers GmbH, Esslingen, Germany Anomaly detection in recordings from in-vehicle networks Andreas Theissler IT-Designers GmbH, Esslingen, Germany andreas.theissler@it-designers.de published in proceedings of BIG DATA APPLICATIONS AND

More information

Linear Classifiers: Expressiveness

Linear Classifiers: Expressiveness Linear Classifiers: Expressiveness Machine Learning Spring 2018 The slides are mainly from Vivek Srikumar 1 Lecture outline Linear classifiers: Introduction What functions do linear classifiers express?

More information

ECE3510 Lab #5 PID Control

ECE3510 Lab #5 PID Control ECE3510 Lab #5 ID Control Objectives The objective of this lab is to study basic design issues for proportionalintegral-derivative control laws. Emphasis is placed on transient responses and steady-state

More information

Midterm Exam, Spring 2005

Midterm Exam, Spring 2005 10-701 Midterm Exam, Spring 2005 1. Write your name and your email address below. Name: Email address: 2. There should be 15 numbered pages in this exam (including this cover sheet). 3. Write your name

More information

Advances in Military Technology Vol. 8, No. 2, December Fault Detection and Diagnosis Based on Extensions of PCA

Advances in Military Technology Vol. 8, No. 2, December Fault Detection and Diagnosis Based on Extensions of PCA AiMT Advances in Military Technology Vol. 8, No. 2, December 203 Fault Detection and Diagnosis Based on Extensions of PCA Y. Zhang *, C.M. Bingham and M. Gallimore School of Engineering, University of

More information

Regularization. CSCE 970 Lecture 3: Regularization. Stephen Scott and Vinod Variyam. Introduction. Outline

Regularization. CSCE 970 Lecture 3: Regularization. Stephen Scott and Vinod Variyam. Introduction. Outline Other Measures 1 / 52 sscott@cse.unl.edu learning can generally be distilled to an optimization problem Choose a classifier (function, hypothesis) from a set of functions that minimizes an objective function

More information

CS395T Computational Statistics with Application to Bioinformatics

CS395T Computational Statistics with Application to Bioinformatics CS395T Computational Statistics with Application to Bioinformatics Prof. William H. Press Spring Term, 2009 The University of Texas at Austin Unit 21: Support Vector Machines The University of Texas at

More information

Computational Statistics with Application to Bioinformatics. Unit 18: Support Vector Machines (SVMs)

Computational Statistics with Application to Bioinformatics. Unit 18: Support Vector Machines (SVMs) Computational Statistics with Application to Bioinformatics Prof. William H. Press Spring Term, 2008 The University of Texas at Austin Unit 18: Support Vector Machines (SVMs) The University of Texas at

More information

Logistic Regression. COMP 527 Danushka Bollegala

Logistic Regression. COMP 527 Danushka Bollegala Logistic Regression COMP 527 Danushka Bollegala Binary Classification Given an instance x we must classify it to either positive (1) or negative (0) class We can use {1,-1} instead of {1,0} but we will

More information

Probability and Statistics. Terms and concepts

Probability and Statistics. Terms and concepts Probability and Statistics Joyeeta Dutta Moscato June 30, 2014 Terms and concepts Sample vs population Central tendency: Mean, median, mode Variance, standard deviation Normal distribution Cumulative distribution

More information

Artificial Ecosystems for Creative Discovery

Artificial Ecosystems for Creative Discovery Artificial Ecosystems for Creative Discovery Jon McCormack Centre for Electronic Media Art Monash University Clayton 3800, Australia www.csse.monash.edu.au/~jonmc Jon.McCormack@infotech.monash.edu.au 1

More information

Unsupervised Anomaly Detection for High Dimensional Data

Unsupervised Anomaly Detection for High Dimensional Data Unsupervised Anomaly Detection for High Dimensional Data Department of Mathematics, Rowan University. July 19th, 2013 International Workshop in Sequential Methodologies (IWSM-2013) Outline of Talk Motivation

More information

A Support Vector Regression Model for Forecasting Rainfall

A Support Vector Regression Model for Forecasting Rainfall A Support Vector Regression for Forecasting Nasimul Hasan 1, Nayan Chandra Nath 1, Risul Islam Rasel 2 Department of Computer Science and Engineering, International Islamic University Chittagong, Bangladesh

More information

Proteomics and Variable Selection

Proteomics and Variable Selection Proteomics and Variable Selection p. 1/55 Proteomics and Variable Selection Alex Lewin With thanks to Paul Kirk for some graphs Department of Epidemiology and Biostatistics, School of Public Health, Imperial

More information

Chunking with Support Vector Machines

Chunking with Support Vector Machines NAACL2001 Chunking with Support Vector Machines Graduate School of Information Science, Nara Institute of Science and Technology, JAPAN Taku Kudo, Yuji Matsumoto {taku-ku,matsu}@is.aist-nara.ac.jp Chunking

More information

Deep Learning for Computer Vision

Deep Learning for Computer Vision Deep Learning for Computer Vision Lecture 4: Curse of Dimensionality, High Dimensional Feature Spaces, Linear Classifiers, Linear Regression, Python, and Jupyter Notebooks Peter Belhumeur Computer Science

More information

Evaluation criteria or Quantifying the information content of the BCI feedback

Evaluation criteria or Quantifying the information content of the BCI feedback Evaluation criteria or Quantifying the information content of the BCI feedback!"# Scheme of a BCI OFFLINE Measure of separability Classification and performance analysis Classification Time signal processing

More information

Incorporating detractors into SVM classification

Incorporating detractors into SVM classification Incorporating detractors into SVM classification AGH University of Science and Technology 1 2 3 4 5 (SVM) SVM - are a set of supervised learning methods used for classification and regression SVM maximal

More information

Lecture 5: Introduction to Complexity Theory

Lecture 5: Introduction to Complexity Theory Lecture 5: Introduction to Complexity Theory 1 Complexity Theory 1.1 Resource Consumption Complexity theory, or more precisely, Computational Complexity theory, deals with the resources required during

More information

Evaluation. Andrea Passerini Machine Learning. Evaluation

Evaluation. Andrea Passerini Machine Learning. Evaluation Andrea Passerini passerini@disi.unitn.it Machine Learning Basic concepts requires to define performance measures to be optimized Performance of learning algorithms cannot be evaluated on entire domain

More information

TDT4173 Machine Learning

TDT4173 Machine Learning TDT4173 Machine Learning Lecture 3 Bagging & Boosting + SVMs Norwegian University of Science and Technology Helge Langseth IT-VEST 310 helgel@idi.ntnu.no 1 TDT4173 Machine Learning Outline 1 Ensemble-methods

More information

Pointwise Exact Bootstrap Distributions of Cost Curves

Pointwise Exact Bootstrap Distributions of Cost Curves Pointwise Exact Bootstrap Distributions of Cost Curves Charles Dugas and David Gadoury University of Montréal 25th ICML Helsinki July 2008 Dugas, Gadoury (U Montréal) Cost curves July 8, 2008 1 / 24 Outline

More information

Support Vector Machines. Introduction to Data Mining, 2 nd Edition by Tan, Steinbach, Karpatne, Kumar

Support Vector Machines. Introduction to Data Mining, 2 nd Edition by Tan, Steinbach, Karpatne, Kumar Data Mining Support Vector Machines Introduction to Data Mining, 2 nd Edition by Tan, Steinbach, Karpatne, Kumar 02/03/2018 Introduction to Data Mining 1 Support Vector Machines Find a linear hyperplane

More information

15-388/688 - Practical Data Science: Nonlinear modeling, cross-validation, regularization, and evaluation

15-388/688 - Practical Data Science: Nonlinear modeling, cross-validation, regularization, and evaluation 15-388/688 - Practical Data Science: Nonlinear modeling, cross-validation, regularization, and evaluation J. Zico Kolter Carnegie Mellon University Fall 2016 1 Outline Example: return to peak demand prediction

More information

Evaluation of Two Level Classifier for Predicting Compressor Failures in Heavy Duty Vehicles. Slawomir Nowaczyk SAIS 2017 workshop, May

Evaluation of Two Level Classifier for Predicting Compressor Failures in Heavy Duty Vehicles. Slawomir Nowaczyk SAIS 2017 workshop, May Evaluation of Two Level Classifier for Predicting Compressor Failures in Heavy Duty Vehicles Yuantao Fan, Pablo De Moral & Slawomir Nowaczyk SAIS 2017 workshop, 15-16 May Objective & Motivation Predictive

More information

Stochastic gradient descent; Classification

Stochastic gradient descent; Classification Stochastic gradient descent; Classification Steve Renals Machine Learning Practical MLP Lecture 2 28 September 2016 MLP Lecture 2 Stochastic gradient descent; Classification 1 Single Layer Networks MLP

More information

Data Mining and Analysis: Fundamental Concepts and Algorithms

Data Mining and Analysis: Fundamental Concepts and Algorithms Data Mining and Analysis: Fundamental Concepts and Algorithms dataminingbook.info Mohammed J. Zaki 1 Wagner Meira Jr. 2 1 Department of Computer Science Rensselaer Polytechnic Institute, Troy, NY, USA

More information

Constrained Optimization and Support Vector Machines

Constrained Optimization and Support Vector Machines Constrained Optimization and Support Vector Machines Man-Wai MAK Dept. of Electronic and Information Engineering, The Hong Kong Polytechnic University enmwmak@polyu.edu.hk http://www.eie.polyu.edu.hk/

More information

QUANTIFYING RESILIENCE-BASED IMPORTANCE MEASURES USING BAYESIAN KERNEL METHODS

QUANTIFYING RESILIENCE-BASED IMPORTANCE MEASURES USING BAYESIAN KERNEL METHODS QUANTIFYING RESILIENCE-BASED IMPORTANCE MEASURES USING BAYESIAN KERNEL METHODS Hiba Baroud, Ph.D. Civil and Environmental Engineering Vanderbilt University Thursday, May 19, 2016 WHAT IS RESILIENCE? Photo:

More information

Support Vector Machines

Support Vector Machines Support Vector Machines Reading: Ben-Hur & Weston, A User s Guide to Support Vector Machines (linked from class web page) Notation Assume a binary classification problem. Instances are represented by vector

More information

Cheng Soon Ong & Christian Walder. Canberra February June 2018

Cheng Soon Ong & Christian Walder. Canberra February June 2018 Cheng Soon Ong & Christian Walder Research Group and College of Engineering and Computer Science Canberra February June 2018 Outlines Overview Introduction Linear Algebra Probability Linear Regression

More information

Bright Advance Corporation

Bright Advance Corporation USER INSTRUCTIONS TABLE OF CONTENTS INSTRUCTIONS FOR USE2 PREPARING TO USE THE SCALE2 DISPLAYS3 KEYBOARD FUNCTION4 OPERATION7 COUNTING14 DIFFERENT KEYBOARD TYPES21 INTERFACE31 POWER SOURCES40 1 INSTRUCTIONS

More information

Evaluation requires to define performance measures to be optimized

Evaluation requires to define performance measures to be optimized Evaluation Basic concepts Evaluation requires to define performance measures to be optimized Performance of learning algorithms cannot be evaluated on entire domain (generalization error) approximation

More information

Numerical Methods For Optimization Problems Arising In Energetic Districts

Numerical Methods For Optimization Problems Arising In Energetic Districts Numerical Methods For Optimization Problems Arising In Energetic Districts Elisa Riccietti, Stefania Bellavia and Stefano Sello Abstract This paper deals with the optimization of energy resources management

More information