Sample Complexity of Learning Independent of Set Theory

Size: px
Start display at page:

Download "Sample Complexity of Learning Independent of Set Theory"

Transcription

1 Sample Complexity of Learning Independent of Set Theory Shai Ben-David University of Waterloo, Canada Based on joint work with Pavel Hrubes, Shay Moran, Amir Shpilka and Amir Yehudayoff Simons workshop, June 2018

2 The 3 components of this talk Sta$s$cal learning theory. Combinatorics. Set Theory.

3 Outline of the talk 1. Introduce a general learning task Expecta)on Maximiza)on (EMX) and show that many common sta$s$cal learning problems are subcases. 2. Introduce a combinatorial no$on of sample compression schemes and show that it characterizes EMX learnability. 3. Show that for certain basic classes, the existence of such compressions is independent of the set theory axioms ZFC. 4. Conclude that EMX learnability is independent of set Theory. 5. Conclude that EMX learnability cannot be captured by any no$on of finite character dimension (VC dimension like). 6. Discuss implica$ons.

4 A General Learning Problem -- Expectation Maximization (EMX) Let H be a collec$on of real valued func$ons over some domain X. Input: a sample S, i.i.d. generated by some unknown probability distribu$on P over X. Output: Some h in H with as high as possible expecta$on w.r.t. P.

5 Vapnik s generalized loss A sta$s$cal learning framework is defined by triplet: a domain set X, a set of models M, and a loss func$on l: X x M è R. Ø For a probability distribu$on P over X and h in M, define L P (h) to be the expecta$on over x ~P of l(h, x). Ø The learner is given a sample S generated i.i.d. by P and aims to find h in M the minimizes L P (h). It is easy to see that the EMX problem encapsulates this problem.

6 Examples of EMX problems Binary classifica$on predic$on ( proper ). Mul$-class predic$on. K-center clustering problems. Linear regression. Learning when the labels are known. Sta$s$cal loss minimiza$on.

7 Common Tools of Trade PAC Learnability Learnability by Empirical Risk Minimiza$on Weak Learnability Uniform Convergence Existence of Sample Compression Schemes Finiteness of Combinatorial Dimensions

8 Binary classimication (-- the clean case) The Fundamental Theorem of StaPsPcal Learning : Given a class H, all the above are equivalent (the relevant dimension is the VC-dimension of the class). The equivalence holds quan$ta$vely.

9 The Theorem breaks down for multi-class classimication. Theorem [Daniely, B-D, Sabato, Shalev-Shwartz 2010] There are mul$-class H s for which not all ERM algorithms are equal -- Some ERM may learn a class that other ERM algorithms fail to learn. Corollary: PAC learnability may hold when uniform convergence fails.

10 Non-equivalence for EMX When H is the class of all subsets of any infinite domain Claim 1: Uniform convergence fails. Claim 2: EMX learnability holds. (Why?...)

11 Our goal Characterize/figure-out: Which classes are EMX learnable? Can one have a purely combinatorial characteriza$on?

12 Sample Compression Schemes [Littlestone and Warmuth 89] Let H be a class of binary valued func)ons over some domain set X, A k-size Sample Compression Scheme for H is a func$on G: (X \$mes {0,1}) k to {0,1} X so that For any h in H and every finite subset S=((x 1,h(x 1 )), (x m, h(x m ))), there is S h, a subset of S of size at most k, so that for all i m, G(S h )(x i )=h(x i ).

13 Recall For binary classifica$on learning problems, PAC learnability is equivalent to the existence of such compressions schemes.

14 The case of Subset Probability Maximization H is a set of (characteris$c func$ons of) subsets of some domain set X. EMX for H is the task of finding a set h in H whose probability weight is within epsilon to the maximum-weight h in H.

15 Monotone Compression Schemes Let H be a class of real valued func)ons over some domain set X, A k-size Monotone Compression Scheme for H is a func$on G: X k to H so that For any finite domain subset S and every h in H, there is a subset S h of size at most k so that for all x in S, G(S h )(x) h(x).

16 Example class Consider H X Fin the collecpon of all finite subsets of a set X. Monotone compression boils down to the following game: Alice: Gets a finite set S, picks a subset S and sends it to Bob. Bob: Outputs a finite subset of X, η(s ). Their goal: To find a strategy by which, for every S, η(s ) is a superset of S. Can Alice send Bob subsets of bounded size?

17 Example continued The answer depends on the cardinality of X. It is trivial is X is finite ( X is a bound on the size of all subsets ) Easy if X is countable (think of the natural numbers ) It gets more interes)ng when X is uncountable.

18 The sub-problem we focus on Let H be a family of subsets of some X. We say that H is union bounded if For every h 1, h 2 in H there is h 3 in H that contains both. Examples: Union closed classes, Axis aligned rectangles in any dimension, convex polygons.

19 The components of the main result For EMX over unions-bounded classes, Theorem: The following are equivalent 1. PAC learnability 2. Weak Learnability 3. Finite size Monotone Compression

20 Recall that for binary classimication all of the following are equivalent PAC Learnability Learnability by Empirical Risk Minimiza$on Weak Learnability Uniform Convergence Existence of Sample Compression Schemes Finiteness of the VC Dimensions

21 A Quantitative version Let d H =d H (1/3, 1/3) be the sample size needed for performing EMX over a class H with accuracy epsilon=1/3 and confidence delta=1/3. Learnability è Compression: Every class H has monotone compression to size k(m)= O(d H log(m)). Compression è Learnability: If a class has k-size monotone compression then d H =O(klog(k)).

22 Monotone compression of H X Fin Theorem: For every k, H X Fin has monotone compression of size at most k if and only if X < Aleph k Corollary: The class of finite subsets of the real unit interval is EMX learnable if and only if 2 N < Aleph ω

23 Proof: 1. If X < Aleph k then H X Fin has monotone compression of size at most k : By induc$on on k, using a well ordering of X of type ω k 2. If H X Fin has monotone compression of size at most k then for every subset Y of X of smaller cardinality, H Y Fin has monotone compression of size at most k-1.

24 Corollaries 1) If [0,1] = Aleph_k then H [0,1] Fin is EMX- learnable and Ck/ log(k) < d H (1/3, 1/3) < Ck log(k) (for some constant C). 2) If [0,1] > Aleph ω then H [0,1] Fin is not EMX- learnable.

25 A set theoretic aspect The cardinality of the reals can be changed (using forcing) without changing the set of real. Therefore the EMX learnability of the class H [0,1] Fin can be switched without changing either the domain set or the class.

26 A Dimension for EMX Is there a combinatorial dimension that characterizes EMX learnability the way VC dimension characterizes binary classifica$on PAC learning?

27 Formalizing the notion of dimension We say that a property of a domain set X and a class H is Finite Character if it can be expressed by a first order formula all of whose quan)fiers are over X and H (and may be real/natural numbers as well). Note that VCdim(H)>d is of Finite Character. So are the Fat Shatering dimension, the Graph dimension and the Natarajan dimension.

28 A model theoretic observation Claim: If M 0 is a submodel of M 1, both models of ZFC and they share the same sets X and H (and same set of real numbers), then The truth value of any finite character property of (X, H) is the same in both models.

29 No Dimension for EMX learnability Corollary: There is no finite character no$on of dimension that characterizes EMX learnability.

30 Discussion Where can this independence of set theory come from? The no$ons of sample complexity are defined in terms of learning func$ons. (rather than learning algorithms). Func$ons over infinite domains, in contrast with programs, are infinitary objects.

On a learning problem that is independent of the set theory ZFC axioms

On a learning problem that is independent of the set theory ZFC axioms On a learning problem that is independent of the set theory ZFC axioms Shai Ben-David, Pavel Hrubeš, Shay Moran, Amir Shpilka, and Amir Yehudayoff Abstract. We consider the following statistical estimation

More information

Littlestone s Dimension and Online Learnability

Littlestone s Dimension and Online Learnability Littlestone s Dimension and Online Learnability Shai Shalev-Shwartz Toyota Technological Institute at Chicago The Hebrew University Talk at UCSD workshop, February, 2009 Joint work with Shai Ben-David

More information

The sample complexity of agnostic learning with deterministic labels

The sample complexity of agnostic learning with deterministic labels The sample complexity of agnostic learning with deterministic labels Shai Ben-David Cheriton School of Computer Science University of Waterloo Waterloo, ON, N2L 3G CANADA shai@uwaterloo.ca Ruth Urner College

More information

Introduction to Machine Learning (67577) Lecture 3

Introduction to Machine Learning (67577) Lecture 3 Introduction to Machine Learning (67577) Lecture 3 Shai Shalev-Shwartz School of CS and Engineering, The Hebrew University of Jerusalem General Learning Model and Bias-Complexity tradeoff Shai Shalev-Shwartz

More information

Lecture Learning infinite hypothesis class via VC-dimension and Rademacher complexity;

Lecture Learning infinite hypothesis class via VC-dimension and Rademacher complexity; CSCI699: Topics in Learning and Game Theory Lecture 2 Lecturer: Ilias Diakonikolas Scribes: Li Han Today we will cover the following 2 topics: 1. Learning infinite hypothesis class via VC-dimension and

More information

Multiclass Learnability and the ERM principle

Multiclass Learnability and the ERM principle Multiclass Learnability and the ERM principle Amit Daniely Sivan Sabato Shai Ben-David Shai Shalev-Shwartz November 5, 04 arxiv:308.893v [cs.lg] 4 Nov 04 Abstract We study the sample complexity of multiclass

More information

Understanding Machine Learning A theory Perspective

Understanding Machine Learning A theory Perspective Understanding Machine Learning A theory Perspective Shai Ben-David University of Waterloo MLSS at MPI Tubingen, 2017 Disclaimer Warning. This talk is NOT about how cool machine learning is. I am sure you

More information

Multiclass Learnability and the ERM Principle

Multiclass Learnability and the ERM Principle Journal of Machine Learning Research 6 205 2377-2404 Submitted 8/3; Revised /5; Published 2/5 Multiclass Learnability and the ERM Principle Amit Daniely amitdaniely@mailhujiacil Dept of Mathematics, The

More information

Generalization bounds

Generalization bounds Advanced Course in Machine Learning pring 200 Generalization bounds Handouts are jointly prepared by hie Mannor and hai halev-hwartz he problem of characterizing learnability is the most basic question

More information

COMS 4771 Introduction to Machine Learning. Nakul Verma

COMS 4771 Introduction to Machine Learning. Nakul Verma COMS 4771 Introduction to Machine Learning Nakul Verma Announcements HW2 due now! Project proposal due on tomorrow Midterm next lecture! HW3 posted Last time Linear Regression Parametric vs Nonparametric

More information

VC Dimension Review. The purpose of this document is to review VC dimension and PAC learning for infinite hypothesis spaces.

VC Dimension Review. The purpose of this document is to review VC dimension and PAC learning for infinite hypothesis spaces. VC Dimension Review The purpose of this document is to review VC dimension and PAC learning for infinite hypothesis spaces. Previously, in discussing PAC learning, we were trying to answer questions about

More information

Computational Learning Theory

Computational Learning Theory CS 446 Machine Learning Fall 2016 OCT 11, 2016 Computational Learning Theory Professor: Dan Roth Scribe: Ben Zhou, C. Cervantes 1 PAC Learning We want to develop a theory to relate the probability of successful

More information

Statistical Learning Learning From Examples

Statistical Learning Learning From Examples Statistical Learning Learning From Examples We want to estimate the working temperature range of an iphone. We could study the physics and chemistry that affect the performance of the phone too hard We

More information

On the tradeoff between computational complexity and sample complexity in learning

On the tradeoff between computational complexity and sample complexity in learning On the tradeoff between computational complexity and sample complexity in learning Shai Shalev-Shwartz School of Computer Science and Engineering The Hebrew University of Jerusalem Joint work with Sham

More information

Introduction to Machine Learning

Introduction to Machine Learning Introduction to Machine Learning Vapnik Chervonenkis Theory Barnabás Póczos Empirical Risk and True Risk 2 Empirical Risk Shorthand: True risk of f (deterministic): Bayes risk: Let us use the empirical

More information

Learnability, Stability, Regularization and Strong Convexity

Learnability, Stability, Regularization and Strong Convexity Learnability, Stability, Regularization and Strong Convexity Nati Srebro Shai Shalev-Shwartz HUJI Ohad Shamir Weizmann Karthik Sridharan Cornell Ambuj Tewari Michigan Toyota Technological Institute Chicago

More information

An Introduction to Statistical Theory of Learning. Nakul Verma Janelia, HHMI

An Introduction to Statistical Theory of Learning. Nakul Verma Janelia, HHMI An Introduction to Statistical Theory of Learning Nakul Verma Janelia, HHMI Towards formalizing learning What does it mean to learn a concept? Gain knowledge or experience of the concept. The basic process

More information

Agnostic Online learnability

Agnostic Online learnability Technical Report TTIC-TR-2008-2 October 2008 Agnostic Online learnability Shai Shalev-Shwartz Toyota Technological Institute Chicago shai@tti-c.org ABSTRACT We study a fundamental question. What classes

More information

The information-theoretic value of unlabeled data in semi-supervised learning

The information-theoretic value of unlabeled data in semi-supervised learning The information-theoretic value of unlabeled data in semi-supervised learning Alexander Golovnev Dávid Pál Balázs Szörényi January 5, 09 Abstract We quantify the separation between the numbers of labeled

More information

IFT Lecture 7 Elements of statistical learning theory

IFT Lecture 7 Elements of statistical learning theory IFT 6085 - Lecture 7 Elements of statistical learning theory This version of the notes has not yet been thoroughly checked. Please report any bugs to the scribes or instructor. Scribe(s): Brady Neal and

More information

A Result of Vapnik with Applications

A Result of Vapnik with Applications A Result of Vapnik with Applications Martin Anthony Department of Statistical and Mathematical Sciences London School of Economics Houghton Street London WC2A 2AE, U.K. John Shawe-Taylor Department of

More information

Machine Learning Theory (CS 6783)

Machine Learning Theory (CS 6783) Machine Learning Theory (CS 6783) Tu-Th 1:25 to 2:40 PM Kimball, B-11 Instructor : Karthik Sridharan ABOUT THE COURSE No exams! 5 assignments that count towards your grades (55%) One term project (40%)

More information

Machine Learning. Regularization and Feature Selection. Fabio Vandin November 13, 2017

Machine Learning. Regularization and Feature Selection. Fabio Vandin November 13, 2017 Machine Learning Regularization and Feature Selection Fabio Vandin November 13, 2017 1 Learning Model A: learning algorithm for a machine learning task S: m i.i.d. pairs z i = (x i, y i ), i = 1,..., m,

More information

COMP9444: Neural Networks. Vapnik Chervonenkis Dimension, PAC Learning and Structural Risk Minimization

COMP9444: Neural Networks. Vapnik Chervonenkis Dimension, PAC Learning and Structural Risk Minimization : Neural Networks Vapnik Chervonenkis Dimension, PAC Learning and Structural Risk Minimization 11s2 VC-dimension and PAC-learning 1 How good a classifier does a learner produce? Training error is the precentage

More information

The Vapnik-Chervonenkis Dimension

The Vapnik-Chervonenkis Dimension The Vapnik-Chervonenkis Dimension Prof. Dan A. Simovici UMB 1 / 91 Outline 1 Growth Functions 2 Basic Definitions for Vapnik-Chervonenkis Dimension 3 The Sauer-Shelah Theorem 4 The Link between VCD and

More information

Generalization, Overfitting, and Model Selection

Generalization, Overfitting, and Model Selection Generalization, Overfitting, and Model Selection Sample Complexity Results for Supervised Classification Maria-Florina (Nina) Balcan 10/03/2016 Two Core Aspects of Machine Learning Algorithm Design. How

More information

Computational Learning Theory

Computational Learning Theory Computational Learning Theory Pardis Noorzad Department of Computer Engineering and IT Amirkabir University of Technology Ordibehesht 1390 Introduction For the analysis of data structures and algorithms

More information

Computational and Statistical Learning theory

Computational and Statistical Learning theory Computational and Statistical Learning theory Problem set 2 Due: January 31st Email solutions to : karthik at ttic dot edu Notation : Input space : X Label space : Y = {±1} Sample : (x 1, y 1,..., (x n,

More information

Generalization theory

Generalization theory Generalization theory Chapter 4 T.P. Runarsson (tpr@hi.is) and S. Sigurdsson (sven@hi.is) Introduction Suppose you are given the empirical observations, (x 1, y 1 ),..., (x l, y l ) (X Y) l. Consider the

More information

VC-dimension in model theory and other subjects

VC-dimension in model theory and other subjects VC-dimension in model theory and other subjects Artem Chernikov (Paris 7 / MSRI, Berkeley) UCLA, 2 May 2014 VC-dimension Let F be a family of subsets of a set X. VC-dimension Let F be a family of subsets

More information

Consistency of Nearest Neighbor Methods

Consistency of Nearest Neighbor Methods E0 370 Statistical Learning Theory Lecture 16 Oct 25, 2011 Consistency of Nearest Neighbor Methods Lecturer: Shivani Agarwal Scribe: Arun Rajkumar 1 Introduction In this lecture we return to the study

More information

On Learnability, Complexity and Stability

On Learnability, Complexity and Stability On Learnability, Complexity and Stability Silvia Villa, Lorenzo Rosasco and Tomaso Poggio 1 Introduction A key question in statistical learning is which hypotheses (function) spaces are learnable. Roughly

More information

Introduction to Machine Learning (67577) Lecture 5

Introduction to Machine Learning (67577) Lecture 5 Introduction to Machine Learning (67577) Lecture 5 Shai Shalev-Shwartz School of CS and Engineering, The Hebrew University of Jerusalem Nonuniform learning, MDL, SRM, Decision Trees, Nearest Neighbor Shai

More information

Introduction to Machine Learning (67577) Lecture 7

Introduction to Machine Learning (67577) Lecture 7 Introduction to Machine Learning (67577) Lecture 7 Shai Shalev-Shwartz School of CS and Engineering, The Hebrew University of Jerusalem Solving Convex Problems using SGD and RLM Shai Shalev-Shwartz (Hebrew

More information

Quadratic Upper Bound for Recursive Teaching Dimension of Finite VC Classes

Quadratic Upper Bound for Recursive Teaching Dimension of Finite VC Classes Proceedings of Machine Learning Research vol 65:1 10, 2017 Quadratic Upper Bound for Recursive Teaching Dimension of Finite VC Classes Lunjia Hu Ruihan Wu Tianhong Li Institute for Interdisciplinary Information

More information

Machine Learning in the Data Revolution Era

Machine Learning in the Data Revolution Era Machine Learning in the Data Revolution Era Shai Shalev-Shwartz School of Computer Science and Engineering The Hebrew University of Jerusalem Machine Learning Seminar Series, Google & University of Waterloo,

More information

Learners that Use Little Information

Learners that Use Little Information Journal of Machine Learning Research? (208)?-? Submitted 0/27; Published?/? Learners that Use Little Information Raef Bassily Department of Computer Science and Engineering The Ohio State University Columbus,

More information

Machine Learning. VC Dimension and Model Complexity. Eric Xing , Fall 2015

Machine Learning. VC Dimension and Model Complexity. Eric Xing , Fall 2015 Machine Learning 10-701, Fall 2015 VC Dimension and Model Complexity Eric Xing Lecture 16, November 3, 2015 Reading: Chap. 7 T.M book, and outline material Eric Xing @ CMU, 2006-2015 1 Last time: PAC and

More information

Introduction to Statistical Learning Theory

Introduction to Statistical Learning Theory Introduction to Statistical Learning Theory Definition Reminder: We are given m samples {(x i, y i )} m i=1 Dm and a hypothesis space H and we wish to return h H minimizing L D (h) = E[l(h(x), y)]. Problem

More information

From Batch to Transductive Online Learning

From Batch to Transductive Online Learning From Batch to Transductive Online Learning Sham Kakade Toyota Technological Institute Chicago, IL 60637 sham@tti-c.org Adam Tauman Kalai Toyota Technological Institute Chicago, IL 60637 kalai@tti-c.org

More information

Introduction to Machine Learning CMU-10701

Introduction to Machine Learning CMU-10701 Introduction to Machine Learning CMU10701 11. Learning Theory Barnabás Póczos Learning Theory We have explored many ways of learning from data But How good is our classifier, really? How much data do we

More information

Multiclass Learnability and the ERM principle

Multiclass Learnability and the ERM principle JMLR: Workshop and Conference Proceedings 19 2011 207 232 24th Annual Conference on Learning Theory Multiclass Learnability and the ERM principle Amit Daniely amit.daniely@mail.huji.ac.il Dept. of Mathematics,

More information

Does Unlabeled Data Help?

Does Unlabeled Data Help? Does Unlabeled Data Help? Worst-case Analysis of the Sample Complexity of Semi-supervised Learning. Ben-David, Lu and Pal; COLT, 2008. Presentation by Ashish Rastogi Courant Machine Learning Seminar. Outline

More information

Machine Learning Theory (CS 6783)

Machine Learning Theory (CS 6783) Machine Learning Theory (CS 6783) Tu-Th 1:25 to 2:40 PM Hollister, 306 Instructor : Karthik Sridharan ABOUT THE COURSE No exams! 5 assignments that count towards your grades (55%) One term project (40%)

More information

Computational and Statistical Learning Theory

Computational and Statistical Learning Theory Computational and Statistical Learning Theory TTIC 31120 Prof. Nati Srebro Lecture 7: Computational Complexity of Learning Agnostic Learning Hardness of Learning via Crypto Assumption: No poly-time algorithm

More information

CS446: Machine Learning Spring Problem Set 4

CS446: Machine Learning Spring Problem Set 4 CS446: Machine Learning Spring 2017 Problem Set 4 Handed Out: February 27 th, 2017 Due: March 11 th, 2017 Feel free to talk to other members of the class in doing the homework. I am more concerned that

More information

PAC Model and Generalization Bounds

PAC Model and Generalization Bounds PAC Model and Generalization Bounds Overview Probably Approximately Correct (PAC) model Basic generalization bounds finite hypothesis class infinite hypothesis class Simple case More next week 2 Motivating

More information

Computational Learning Theory: Shattering and VC Dimensions. Machine Learning. Spring The slides are mainly from Vivek Srikumar

Computational Learning Theory: Shattering and VC Dimensions. Machine Learning. Spring The slides are mainly from Vivek Srikumar Computational Learning Theory: Shattering and VC Dimensions Machine Learning Spring 2018 The slides are mainly from Vivek Srikumar 1 This lecture: Computational Learning Theory The Theory of Generalization

More information

Narrowing confidence interval width of PAC learning risk function by algorithmic inference

Narrowing confidence interval width of PAC learning risk function by algorithmic inference Narrowing confidence interval width of PAC learning risk function by algorithmic inference Bruno Apolloni, Dario Malchiodi Dip. di Scienze dell Informazione, Università degli Studi di Milano Via Comelico

More information

On the Sample Complexity of Noise-Tolerant Learning

On the Sample Complexity of Noise-Tolerant Learning On the Sample Complexity of Noise-Tolerant Learning Javed A. Aslam Department of Computer Science Dartmouth College Hanover, NH 03755 Scott E. Decatur Laboratory for Computer Science Massachusetts Institute

More information

Learnability of the Superset Label Learning Problem

Learnability of the Superset Label Learning Problem Learnability of the Superset Label Learning Problem Li-Ping Liu LIULI@EECSOREGONSTATEEDU Thomas G Dietterich TGD@EECSOREGONSTATEEDU School of Electrical Engineering and Computer Science, Oregon State University,

More information

Computational Learning Theory. CS534 - Machine Learning

Computational Learning Theory. CS534 - Machine Learning Computational Learning Theory CS534 Machine Learning Introduction Computational learning theory Provides a theoretical analysis of learning Shows when a learning algorithm can be expected to succeed Shows

More information

Some Review and Hypothesis Tes4ng. Friday, March 15, 13

Some Review and Hypothesis Tes4ng. Friday, March 15, 13 Some Review and Hypothesis Tes4ng Outline Discussing the homework ques4ons from Joey and Phoebe Review of Sta4s4cal Inference Proper4es of OLS under the normality assump4on Confidence Intervals, T test,

More information

Computational Learning Theory: Probably Approximately Correct (PAC) Learning. Machine Learning. Spring The slides are mainly from Vivek Srikumar

Computational Learning Theory: Probably Approximately Correct (PAC) Learning. Machine Learning. Spring The slides are mainly from Vivek Srikumar Computational Learning Theory: Probably Approximately Correct (PAC) Learning Machine Learning Spring 2018 The slides are mainly from Vivek Srikumar 1 This lecture: Computational Learning Theory The Theory

More information

Computational and Statistical Learning Theory

Computational and Statistical Learning Theory Computational and Statistical Learning Theory TTIC 31120 Prof. Nati Srebro Lecture 4: MDL and PAC-Bayes Uniform vs Non-Uniform Bias No Free Lunch: we need some inductive bias Limiting attention to hypothesis

More information

March 3, The large and small in model theory: What are the amalgamation spectra of. infinitary classes? John T. Baldwin

March 3, The large and small in model theory: What are the amalgamation spectra of. infinitary classes? John T. Baldwin large and large and March 3, 2015 Characterizing cardinals by L ω1,ω large and L ω1,ω satisfies downward Lowenheim Skolem to ℵ 0 for sentences. It does not satisfy upward Lowenheim Skolem. Definition sentence

More information

Machine Learning. Lecture 9: Learning Theory. Feng Li.

Machine Learning. Lecture 9: Learning Theory. Feng Li. Machine Learning Lecture 9: Learning Theory Feng Li fli@sdu.edu.cn https://funglee.github.io School of Computer Science and Technology Shandong University Fall 2018 Why Learning Theory How can we tell

More information

Learning symmetric non-monotone submodular functions

Learning symmetric non-monotone submodular functions Learning symmetric non-monotone submodular functions Maria-Florina Balcan Georgia Institute of Technology ninamf@cc.gatech.edu Nicholas J. A. Harvey University of British Columbia nickhar@cs.ubc.ca Satoru

More information

Short Introduction to Admissible Recursion Theory

Short Introduction to Admissible Recursion Theory Short Introduction to Admissible Recursion Theory Rachael Alvir November 2016 1 Axioms of KP and Admissible Sets An admissible set is a transitive set A satisfying the axioms of Kripke-Platek Set Theory

More information

PAC Learning Introduction to Machine Learning. Matt Gormley Lecture 14 March 5, 2018

PAC Learning Introduction to Machine Learning. Matt Gormley Lecture 14 March 5, 2018 10-601 Introduction to Machine Learning Machine Learning Department School of Computer Science Carnegie Mellon University PAC Learning Matt Gormley Lecture 14 March 5, 2018 1 ML Big Picture Learning Paradigms:

More information

Machine Learning. Computational Learning Theory. Eric Xing , Fall Lecture 9, October 5, 2016

Machine Learning. Computational Learning Theory. Eric Xing , Fall Lecture 9, October 5, 2016 Machine Learning 10-701, Fall 2016 Computational Learning Theory Eric Xing Lecture 9, October 5, 2016 Reading: Chap. 7 T.M book Eric Xing @ CMU, 2006-2016 1 Generalizability of Learning In machine learning

More information

Neural Networks: Introduction

Neural Networks: Introduction Neural Networks: Introduction Machine Learning Fall 2017 Based on slides and material from Geoffrey Hinton, Richard Socher, Dan Roth, Yoav Goldberg, Shai Shalev-Shwartz and Shai Ben-David, and others 1

More information

DO FIVE OUT OF SIX ON EACH SET PROBLEM SET

DO FIVE OUT OF SIX ON EACH SET PROBLEM SET DO FIVE OUT OF SIX ON EACH SET PROBLEM SET 1. THE AXIOM OF FOUNDATION Early on in the book (page 6) it is indicated that throughout the formal development set is going to mean pure set, or set whose elements,

More information

Hence C has VC-dimension d iff. π C (d) = 2 d. (4) C has VC-density l if there exists K R such that, for all

Hence C has VC-dimension d iff. π C (d) = 2 d. (4) C has VC-density l if there exists K R such that, for all 1. Computational Learning Abstract. I discuss the basics of computational learning theory, including concept classes, VC-dimension, and VC-density. Next, I touch on the VC-Theorem, ɛ-nets, and the (p,

More information

PAC Learning. prof. dr Arno Siebes. Algorithmic Data Analysis Group Department of Information and Computing Sciences Universiteit Utrecht

PAC Learning. prof. dr Arno Siebes. Algorithmic Data Analysis Group Department of Information and Computing Sciences Universiteit Utrecht PAC Learning prof. dr Arno Siebes Algorithmic Data Analysis Group Department of Information and Computing Sciences Universiteit Utrecht Recall: PAC Learning (Version 1) A hypothesis class H is PAC learnable

More information

CS340 Machine learning Lecture 5 Learning theory cont'd. Some slides are borrowed from Stuart Russell and Thorsten Joachims

CS340 Machine learning Lecture 5 Learning theory cont'd. Some slides are borrowed from Stuart Russell and Thorsten Joachims CS340 Machine learning Lecture 5 Learning theory cont'd Some slides are borrowed from Stuart Russell and Thorsten Joachims Inductive learning Simplest form: learn a function from examples f is the target

More information

Machine Learning. Computational Learning Theory. Le Song. CSE6740/CS7641/ISYE6740, Fall 2012

Machine Learning. Computational Learning Theory. Le Song. CSE6740/CS7641/ISYE6740, Fall 2012 Machine Learning CSE6740/CS7641/ISYE6740, Fall 2012 Computational Learning Theory Le Song Lecture 11, September 20, 2012 Based on Slides from Eric Xing, CMU Reading: Chap. 7 T.M book 1 Complexity of Learning

More information

arxiv: v3 [cs.lg] 23 Apr 2018

arxiv: v3 [cs.lg] 23 Apr 2018 On Communication Complexity of Classification Problems Daniel M. Kane Roi Livni Shay Moran Amir Yehudayoff April 24, 2018 arxiv:1711.05893v3 [cs.lg] 23 Apr 2018 Abstract This work studies distributed learning

More information

COMP 562: Introduction to Machine Learning

COMP 562: Introduction to Machine Learning COMP 562: Introduction to Machine Learning Lecture 20 : Support Vector Machines, Kernels Mahmoud Mostapha 1 Department of Computer Science University of North Carolina at Chapel Hill mahmoudm@cs.unc.edu

More information

12.1 A Polynomial Bound on the Sample Size m for PAC Learning

12.1 A Polynomial Bound on the Sample Size m for PAC Learning 67577 Intro. to Machine Learning Fall semester, 2008/9 Lecture 12: PAC III Lecturer: Amnon Shashua Scribe: Amnon Shashua 1 In this lecture will use the measure of VC dimension, which is a combinatorial

More information

Generalization and Overfitting

Generalization and Overfitting Generalization and Overfitting Model Selection Maria-Florina (Nina) Balcan February 24th, 2016 PAC/SLT models for Supervised Learning Data Source Distribution D on X Learning Algorithm Expert / Oracle

More information

Sample width for multi-category classifiers

Sample width for multi-category classifiers R u t c o r Research R e p o r t Sample width for multi-category classifiers Martin Anthony a Joel Ratsaby b RRR 29-2012, November 2012 RUTCOR Rutgers Center for Operations Research Rutgers University

More information

1.4 Cardinality. Tom Lewis. Fall Term Tom Lewis () 1.4 Cardinality Fall Term / 9

1.4 Cardinality. Tom Lewis. Fall Term Tom Lewis () 1.4 Cardinality Fall Term / 9 1.4 Cardinality Tom Lewis Fall Term 2006 Tom Lewis () 1.4 Cardinality Fall Term 2006 1 / 9 Outline 1 Functions 2 Cardinality 3 Cantor s theorem Tom Lewis () 1.4 Cardinality Fall Term 2006 2 / 9 Functions

More information

On Finding Planted Cliques and Solving Random Linear Equa9ons. Math Colloquium. Lev Reyzin UIC

On Finding Planted Cliques and Solving Random Linear Equa9ons. Math Colloquium. Lev Reyzin UIC On Finding Planted Cliques and Solving Random Linear Equa9ons Math Colloquium Lev Reyzin UIC 1 random graphs 2 Erdős- Rényi Random Graphs G(n,p) generates graph G on n ver9ces by including each possible

More information

Classical Theory of Cardinal Characteristics

Classical Theory of Cardinal Characteristics Classical Theory of Cardinal Characteristics Andreas Blass University of Michigan 22 August, 2018 Andreas Blass (University of Michigan) Classical Theory of Cardinal Characteristics 22 August, 2018 1 /

More information

The No-Regret Framework for Online Learning

The No-Regret Framework for Online Learning The No-Regret Framework for Online Learning A Tutorial Introduction Nahum Shimkin Technion Israel Institute of Technology Haifa, Israel Stochastic Processes in Engineering IIT Mumbai, March 2013 N. Shimkin,

More information

Lecture Support Vector Machine (SVM) Classifiers

Lecture Support Vector Machine (SVM) Classifiers Introduction to Machine Learning Lecturer: Amir Globerson Lecture 6 Fall Semester Scribe: Yishay Mansour 6.1 Support Vector Machine (SVM) Classifiers Classification is one of the most important tasks in

More information

Online Learning Summer School Copenhagen 2015 Lecture 1

Online Learning Summer School Copenhagen 2015 Lecture 1 Online Learning Summer School Copenhagen 2015 Lecture 1 Shai Shalev-Shwartz School of CS and Engineering, The Hebrew University of Jerusalem Online Learning Shai Shalev-Shwartz (Hebrew U) OLSS Lecture

More information

Agnostic Online Learning

Agnostic Online Learning Agnostic Online Learning Shai Ben-David and Dávid Pál David R. Cheriton School of Computer Science University of Waterloo Waterloo, ON, Canada {shai,dpal}@cs.uwaterloo.ca Shai Shalev-Shwartz Toyota Technological

More information

THE VAPNIK- CHERVONENKIS DIMENSION and LEARNABILITY

THE VAPNIK- CHERVONENKIS DIMENSION and LEARNABILITY THE VAPNIK- CHERVONENKIS DIMENSION and LEARNABILITY Dan A. Simovici UMB, Doctoral Summer School Iasi, Romania What is Machine Learning? The Vapnik-Chervonenkis Dimension Probabilistic Learning Potential

More information

Understanding Machine Learning Solution Manual

Understanding Machine Learning Solution Manual Understanding Machine Learning Solution Manual Written by Alon Gonen Edited by Dana Rubinstein Noveber 17, 2014 2 Gentle Start 1. Given S = ((x i, y i )), define the ultivariate polynoial p S (x) = i []:y

More information

Conditional Sparse Linear Regression

Conditional Sparse Linear Regression COMS 6998-4 Fall 07 November 3, 07 Introduction. Background Conditional Sparse Linear Regression esenter: Xingyu Niu Scribe: Jinyi Zhang Given a distribution D over R d R and data samples (y, z) D, linear

More information

CITS2211 Discrete Structures (2017) Cardinality and Countability

CITS2211 Discrete Structures (2017) Cardinality and Countability CITS2211 Discrete Structures (2017) Cardinality and Countability Highlights What is cardinality? Is it the same as size? Types of cardinality and infinite sets Reading Sections 45 and 81 84 of Mathematics

More information

Online Convex Optimization

Online Convex Optimization Advanced Course in Machine Learning Spring 2010 Online Convex Optimization Handouts are jointly prepared by Shie Mannor and Shai Shalev-Shwartz A convex repeated game is a two players game that is performed

More information

arxiv: v1 [cs.lg] 18 Feb 2017

arxiv: v1 [cs.lg] 18 Feb 2017 Quadratic Upper Bound for Recursive Teaching Dimension of Finite VC Classes Lunjia Hu, Ruihan Wu, Tianhong Li and Liwei Wang arxiv:1702.05677v1 [cs.lg] 18 Feb 2017 February 21, 2017 Abstract In this work

More information

CS340 Machine learning Lecture 4 Learning theory. Some slides are borrowed from Sebastian Thrun and Stuart Russell

CS340 Machine learning Lecture 4 Learning theory. Some slides are borrowed from Sebastian Thrun and Stuart Russell CS340 Machine learning Lecture 4 Learning theory Some slides are borrowed from Sebastian Thrun and Stuart Russell Announcement What: Workshop on applying for NSERC scholarships and for entry to graduate

More information

Web-Mining Agents Computational Learning Theory

Web-Mining Agents Computational Learning Theory Web-Mining Agents Computational Learning Theory Prof. Dr. Ralf Möller Dr. Özgür Özcep Universität zu Lübeck Institut für Informationssysteme Tanya Braun (Exercise Lab) Computational Learning Theory (Adapted)

More information

1 The Probably Approximately Correct (PAC) Model

1 The Probably Approximately Correct (PAC) Model COS 511: Theoretical Machine Learning Lecturer: Rob Schapire Lecture #3 Scribe: Yuhui Luo February 11, 2008 1 The Probably Approximately Correct (PAC) Model A target concept class C is PAC-learnable by

More information

A Tutorial on Computational Learning Theory Presented at Genetic Programming 1997 Stanford University, July 1997

A Tutorial on Computational Learning Theory Presented at Genetic Programming 1997 Stanford University, July 1997 A Tutorial on Computational Learning Theory Presented at Genetic Programming 1997 Stanford University, July 1997 Vasant Honavar Artificial Intelligence Research Laboratory Department of Computer Science

More information

Relating Data Compression and Learnability

Relating Data Compression and Learnability Relating Data Compression and Learnability Nick Littlestone, Manfred K. Warmuth Department of Computer and Information Sciences University of California at Santa Cruz June 10, 1986 Abstract We explore

More information

COS 402 Machine Learning and Artificial Intelligence Fall Lecture 3: Learning Theory

COS 402 Machine Learning and Artificial Intelligence Fall Lecture 3: Learning Theory COS 402 Machine Learning and Artificial Intelligence Fall 2016 Lecture 3: Learning Theory Sanjeev Arora Elad Hazan Admin Exercise 1 due next Tue, in class Enrolment Recap We have seen: AI by introspection

More information

References for online kernel methods

References for online kernel methods References for online kernel methods W. Liu, J. Principe, S. Haykin Kernel Adaptive Filtering: A Comprehensive Introduction. Wiley, 2010. W. Liu, P. Pokharel, J. Principe. The kernel least mean square

More information

Introduction to Machine Learning

Introduction to Machine Learning Introduction to Machine Learning 236756 Prof. Nir Ailon Lecture 4: Computational Complexity of Learning & Surrogate Losses Efficient PAC Learning Until now we were mostly worried about sample complexity

More information

Discriminative Models

Discriminative Models No.5 Discriminative Models Hui Jiang Department of Electrical Engineering and Computer Science Lassonde School of Engineering York University, Toronto, Canada Outline Generative vs. Discriminative models

More information

Discriminative Models

Discriminative Models No.5 Discriminative Models Hui Jiang Department of Electrical Engineering and Computer Science Lassonde School of Engineering York University, Toronto, Canada Outline Generative vs. Discriminative models

More information

Online Learning: Random Averages, Combinatorial Parameters, and Learnability

Online Learning: Random Averages, Combinatorial Parameters, and Learnability Online Learning: Random Averages, Combinatorial Parameters, and Learnability Alexander Rakhlin Department of Statistics University of Pennsylvania Karthik Sridharan Toyota Technological Institute at Chicago

More information

12. Structural Risk Minimization. ECE 830 & CS 761, Spring 2016

12. Structural Risk Minimization. ECE 830 & CS 761, Spring 2016 12. Structural Risk Minimization ECE 830 & CS 761, Spring 2016 1 / 23 General setup for statistical learning theory We observe training examples {x i, y i } n i=1 x i = features X y i = labels / responses

More information

Part of the slides are adapted from Ziko Kolter

Part of the slides are adapted from Ziko Kolter Part of the slides are adapted from Ziko Kolter OUTLINE 1 Supervised learning: classification........................................................ 2 2 Non-linear regression/classification, overfitting,

More information

Empirical Risk Minimization

Empirical Risk Minimization Empirical Risk Minimization Fabrice Rossi SAMM Université Paris 1 Panthéon Sorbonne 2018 Outline Introduction PAC learning ERM in practice 2 General setting Data X the input space and Y the output space

More information

Mathematics 220 Workshop Cardinality. Some harder problems on cardinality.

Mathematics 220 Workshop Cardinality. Some harder problems on cardinality. Some harder problems on cardinality. These are two series of problems with specific goals: the first goal is to prove that the cardinality of the set of irrational numbers is continuum, and the second

More information