Top Tagging with Lorentz Boost Networks and Simulation of Electromagnetic Showers with a Wasserstein GAN

Size: px
Start display at page:

Download "Top Tagging with Lorentz Boost Networks and Simulation of Electromagnetic Showers with a Wasserstein GAN"

Transcription

1 Top Tagging with Lorentz Boost Networks and Simulation of Electromagnetic Showers with a Wasserstein GAN Y. Rath, M. Erdmann, B. Fischer, L. Geiger, E. Geiser, J.Glombitza, D. Noll, T. Quast, M. Rieger, D. Schmidt, M. Wirtz III. Physikalisches Institut A, RWTH Aachen University

2 2 Introduction Presenting two applications of deep learning related to jet physics Lorentz Boost Networks (LBN) as a network architecture exploiting physical structure Application to top-tagging Fast simulation of electromagnetic showers with a Wasserstein GAN using label conditioning

3 Top Tagging with Lorentz Boost Networks

4 LBN - Motivation 4 Success of deep learning in computer science through dedicated architectures exploiting problem structure [Zeiler & Fergus 2013], adapted by Yann LeCun Udacity Course 730, Deep Learning Similar developments in physics, e.g. LoLa, Recursive NN (and others!) LBN: Utilize knowledge of particle combinations, rest frames through Lorentz boosts

5 LBN Structure 5 Input: Particle four vectors Combine into particles and rest frames (trainable), perform Lorentz boost Physical interpretation of network weights

6 6 LBN Structure (II) Build generic features from boosted particles, e.g. masses, pair-wise angles Fully connected layers to combine information

7 7 Feature Construction Example of feature construction: cos(θ*)

8 8 Top Tagging with LBN Example application: Top-tagging Distinguish boosted top quarks from QCD jets Use public dataset based on arxiv: enables comparison of diferent algorithms

9 9 LBN Setup Input ordering: Based on clustering history Cluster sub-jets with anti-kt, R=0.25 Order constituents based on clustering history Order sub-jets based on clustering order with anti-kt, R=0.8 LBN architecture: 50 combined particles and rest frames 4 feedforward layers at the end

10 10 Performance LBN competitive with other sophisticated approaches Only minor diferences, hitting physical limit of dataset? Additional method comparisons will follow

11 Simulation of Electromagnetic Showers

12 12 Introduction - WGAN Calorimeter simulations computationally very expensive Ongoing research into simulation with Generative Adversarial Networks (GANs) See e.g. LAGAN, CaloGAN Promise speed-up of several orders of magnitude (~10³ to 10⁵) compared to full simulation

13 13 Experimental Setup Electromagnetic compartment of a CMS High Granularity Calorimeter (HGCAL) prototype Seven sensitive layers with >100 hexagonal pixels each Training data: Geant4 based simulation of electron showers with GeV, diferent impact positions

14 14 Generative Adversarial Networks Concept: NN generates samples from noise following 'real' data distribution Adversary tries to distinguish generated from real events Feedback to the generator Training requires careful balance of both networks Variant: Wasserstein GAN

15 15 WGAN Concept Wasserstein metric as distance measure Earth mover distance : Work required to move one distribution to the other (mass x distance) Can be realized by a neural network with limited gradients, the 'critic' Meaningful gradients everywhere

16 WGAN Setup 16 Translate hexagonal pixels to cartesian coordinate system Add starting conditions (energy, impact position) to inputs Generator Latent input + starting conditions (10 + 3) x 1 x 1 Critic Shower (12x15x7) linear 192x1x1 22xxFully connected 192x1x1 2 x Reshape Fully connected 192x1x1 Reshape 3x4x16 3x4x16 7x Reshape DeConv 33xxDeConv 3 x Convolution DeConv Convolution Conv2D 3x4x16 24x32x64 24x32x64 24x32x64 20x24x1 20x24x1 20x24x1... concat 3 x Conv2D Starting conditions 3x1x1 2 x Fully con. 192x1x1 Reshape 3 x Conv2D 14x17x128 Locally connected 12x15x7 14x17x128 Locally connected 12x15x7 Shower (12x15x7) Critic (1x1x1) 3x4x16

17 17 Constrainer Networks Simulated events should depend on starting conditions (here: initial particle energy, impact position) Use label conditioning Introduced by auxiliary classifer GANs (AC-GANs) Starting conditions given to both the generator and the critic Train constrainer networks to reconstruct energy/impact position of real samples Additional loss term for the generator

18 18 Shower Generation Visual inspection: Showers scale with electron energy, move according to impact position To study full set of events, look at average occupancy

19 19 Validation Compare observable distributions between GEANT4 and WGAN Overall good agreement 70 GeV electrons (grey) not part of training set Assess interpolation capabilities

20 20 Validation (II) Beyond observable distributions: Correlations also well modelled Only diference: Low energy densities underrepresented Sparsity related variables challenge for GANs ~10% contribution to the signal

21 21 Summary LBN physics-motivated NN architecture Integrates particle combinations and Lorentz boosting Example application of top-tagging Simulation of electromagnetic showers with GAN Wasserstein distance to improve training stability Constrainer networks to incorporate initial conditions

Learning Particle Physics by Example:

Learning Particle Physics by Example: Learning Particle Physics by Example: Accelerating Science with Generative Adversarial Networks arxiv:1701.05927, arxiv:1705.02355 @lukede0 @lukedeo lukedeo@manifold.ai https://ldo.io Luke de Oliveira

More information

GANs, GANs everywhere

GANs, GANs everywhere GANs, GANs everywhere particularly, in High Energy Physics Maxim Borisyak Yandex, NRU Higher School of Economics Generative Generative models Given samples of a random variable X find X such as: P X P

More information

GAN Applications in High Energy Particle Physics

GAN Applications in High Energy Particle Physics 75.2355 7.5927 GAN Applications in High Energy Particle Physics Benjamin Nachman Lawrence Berkeley National Laboratory with collaborators Michela Paganini and Luke de Oliveira Outline: DNN with HEP images

More information

Boosted Top Tagging with Deep Neural Networks

Boosted Top Tagging with Deep Neural Networks Boosted Top Tagging with Deep Neural Networks Jannicke Pearkes University of British Columbia, Engineering Physics Wojtek Fedorko, Alison Lister, Colin Gay Inter-Experimental Machine Learning Workshop

More information

Deep generative models for fast shower simulation in ATLAS

Deep generative models for fast shower simulation in ATLAS Deep generative models for fast shower simulation in ATLAS Dalila Salamani University of Geneva dalila.salamani@cern.ch On behalf of the ATLAS collaboration 14th e-science IEEE International Conference

More information

Machine Learning for Computer Vision 8. Neural Networks and Deep Learning. Vladimir Golkov Technical University of Munich Computer Vision Group

Machine Learning for Computer Vision 8. Neural Networks and Deep Learning. Vladimir Golkov Technical University of Munich Computer Vision Group Machine Learning for Computer Vision 8. Neural Networks and Deep Learning Vladimir Golkov Technical University of Munich Computer Vision Group INTRODUCTION Nonlinear Coordinate Transformation http://cs.stanford.edu/people/karpathy/convnetjs/

More information

Tasks ADAS. Self Driving. Non-machine Learning. Traditional MLP. Machine-Learning based method. Supervised CNN. Methods. Deep-Learning based

Tasks ADAS. Self Driving. Non-machine Learning. Traditional MLP. Machine-Learning based method. Supervised CNN. Methods. Deep-Learning based UNDERSTANDING CNN ADAS Tasks Self Driving Localizati on Perception Planning/ Control Driver state Vehicle Diagnosis Smart factory Methods Traditional Deep-Learning based Non-machine Learning Machine-Learning

More information

bb and TopTagging in ATLAS

bb and TopTagging in ATLAS X bb and TopTagging in ATLAS Mike Nelson, University of Oxford michael.nelson@physics.ox.ac.uk Focus of the discussion I want to try and achieve two things: Introduce the basic tools employed in ATLAS

More information

Quark/Gluon Discrimination with Jet-Images and Deep Learning

Quark/Gluon Discrimination with Jet-Images and Deep Learning Quark/Gluon Discrimination with Jet-Images and Deep Learning BOOST 2017 Patrick T. Komiske Center for Theoretical Physics, Massachusetts Institute of Technology Based on arxiv:1612.01551 PTK, Eric M. Metodiev,

More information

Lecture 14: Deep Generative Learning

Lecture 14: Deep Generative Learning Generative Modeling CSED703R: Deep Learning for Visual Recognition (2017F) Lecture 14: Deep Generative Learning Density estimation Reconstructing probability density function using samples Bohyung Han

More information

Understanding How ConvNets See

Understanding How ConvNets See Understanding How ConvNets See Slides from Andrej Karpathy Springerberg et al, Striving for Simplicity: The All Convolutional Net (ICLR 2015 workshops) CSC321: Intro to Machine Learning and Neural Networks,

More information

Machine learning in ALICE

Machine learning in ALICE Activities in ALICE and heavy-ion physics (CERN) (05.04.2018) Outline This lecture consists of two parts 1) Overview on machine learning activities in ALICE Jets Particle identification Charmed baryons

More information

THE MULTIPLICIY JUMP! FINDING B S IN MULTI-TEV JETS W/O TRACKS

THE MULTIPLICIY JUMP! FINDING B S IN MULTI-TEV JETS W/O TRACKS 1 THE MULTIPLICIY JUMP! FINDING B S IN MULTI-TEV JETS W/O TRACKS TODD HUFFMAN, OXFORD UNIVERSITY THOMAS RUSSELL, CURRENTLY @ BLOOMBERG L.P. JEFF TSENG, OXFORD UNIVERSITY ARXIV:1701.06832 ALSO: 2016 J.PHYS.

More information

Pre-Processing and Re-Weighting Jet Images with Different Substructure Variables

Pre-Processing and Re-Weighting Jet Images with Different Substructure Variables Pre-Processing and Re-Weighting Jet Images with Different Substructure Variables Lynn Huynh University of California, Davis Department of Mechanical Engineering CERN Work Project Report CERN, ATLAS, Jet

More information

Generative Adversarial Networks. Presented by Yi Zhang

Generative Adversarial Networks. Presented by Yi Zhang Generative Adversarial Networks Presented by Yi Zhang Deep Generative Models N(O, I) Variational Auto-Encoders GANs Unreasonable Effectiveness of GANs GANs Discriminator tries to distinguish genuine data

More information

Particle Flow Algorithms

Particle Flow Algorithms Particle Flow Algorithms Daniel Jeans, KEK IAS Program on High Energy Physics HKUST Hong Kong January, 2018 introduction and motivation bias towards e+ e- collisions general features of detectors and reconstruction

More information

Singing Voice Separation using Generative Adversarial Networks

Singing Voice Separation using Generative Adversarial Networks Singing Voice Separation using Generative Adversarial Networks Hyeong-seok Choi, Kyogu Lee Music and Audio Research Group Graduate School of Convergence Science and Technology Seoul National University

More information

Wasserstein GAN. Juho Lee. Jan 23, 2017

Wasserstein GAN. Juho Lee. Jan 23, 2017 Wasserstein GAN Juho Lee Jan 23, 2017 Wasserstein GAN (WGAN) Arxiv submission Martin Arjovsky, Soumith Chintala, and Léon Bottou A new GAN model minimizing the Earth-Mover s distance (Wasserstein-1 distance)

More information

Distinguishing quark and gluon jets at the LHC

Distinguishing quark and gluon jets at the LHC Distinguishing quark and jets at the LHC Giorgia Rauco (on behalf of the ALAS and CMS Collaborations) Universität Zürich, Zürich, Switzerland Abstract: Studies focused on discriminating between jets originating

More information

Predicting Deeper into the Future of Semantic Segmentation Supplementary Material

Predicting Deeper into the Future of Semantic Segmentation Supplementary Material Predicting Deeper into the Future of Semantic Segmentation Supplementary Material Pauline Luc 1,2 Natalia Neverova 1 Camille Couprie 1 Jakob Verbeek 2 Yann LeCun 1,3 1 Facebook AI Research 2 Inria Grenoble,

More information

CAUSAL GAN: LEARNING CAUSAL IMPLICIT GENERATIVE MODELS WITH ADVERSARIAL TRAINING

CAUSAL GAN: LEARNING CAUSAL IMPLICIT GENERATIVE MODELS WITH ADVERSARIAL TRAINING CAUSAL GAN: LEARNING CAUSAL IMPLICIT GENERATIVE MODELS WITH ADVERSARIAL TRAINING (Murat Kocaoglu, Christopher Snyder, Alexandros G. Dimakis & Sriram Vishwanath, 2017) Summer Term 2018 Created for the Seminar

More information

Dark matter searches and prospects at the ATLAS experiment

Dark matter searches and prospects at the ATLAS experiment Dark matter searches and prospects at the ATLAS experiment Wendy Taylor (York University) for the ATLAS Collaboration TeVPA 2017 Columbus, Ohio, USA August 7-11, 2017 Dark Matter at ATLAS Use 13 TeV proton-proton

More information

Classifaction of gg gh against qg qh

Classifaction of gg gh against qg qh Jet Variables, Machine Learning, Box vs Triangle Diagrams Joshua Lin + Ben Nachman joshua.z.lin@gmail.com LBNL, Berkeley 2018 January 8-12 Conference, LBNL Introduction 2 These days, we understand very

More information

Boosted hadronic object identification using jet substructure in ATLAS Run-2

Boosted hadronic object identification using jet substructure in ATLAS Run-2 Boosted hadronic object identification using jet substructure in ATLAS Run-2 Emma Winkels on behalf of the ATLAS collaboration HEPMAD18 Outline Jets and jet substructure Top and W tagging H bb tagging

More information

Introduction to Convolutional Neural Networks (CNNs)

Introduction to Convolutional Neural Networks (CNNs) Introduction to Convolutional Neural Networks (CNNs) nojunk@snu.ac.kr http://mipal.snu.ac.kr Department of Transdisciplinary Studies Seoul National University, Korea Jan. 2016 Many slides are from Fei-Fei

More information

Introduction to Deep Learning CMPT 733. Steven Bergner

Introduction to Deep Learning CMPT 733. Steven Bergner Introduction to Deep Learning CMPT 733 Steven Bergner Overview Renaissance of artificial neural networks Representation learning vs feature engineering Background Linear Algebra, Optimization Regularization

More information

Nishant Gurnani. GAN Reading Group. April 14th, / 107

Nishant Gurnani. GAN Reading Group. April 14th, / 107 Nishant Gurnani GAN Reading Group April 14th, 2017 1 / 107 Why are these Papers Important? 2 / 107 Why are these Papers Important? Recently a large number of GAN frameworks have been proposed - BGAN, LSGAN,

More information

W vs. QCD Jet Tagging at the Large Hadron Collider

W vs. QCD Jet Tagging at the Large Hadron Collider W vs. QCD Jet Tagging at the Large Hadron Collider Bryan Anenberg: anenberg@stanford.edu; CS229 December 13, 2013 Problem Statement High energy collisions of protons at the Large Hadron Collider (LHC)

More information

Image Processing, Computer Vision, and Deep Learning: new approaches to the analysis and physics interpretation of LHC events

Image Processing, Computer Vision, and Deep Learning: new approaches to the analysis and physics interpretation of LHC events Journal of Physics: Conference Series PAPER OPEN ACCESS Image Processing, Computer Vision, and Deep Learning: new approaches to the analysis and physics interpretation of LHC events o cite this article:

More information

arxiv: v1 [cs.lg] 20 Apr 2017

arxiv: v1 [cs.lg] 20 Apr 2017 Softmax GAN Min Lin Qihoo 360 Technology co. ltd Beijing, China, 0087 mavenlin@gmail.com arxiv:704.069v [cs.lg] 0 Apr 07 Abstract Softmax GAN is a novel variant of Generative Adversarial Network (GAN).

More information

Machine Learning Basics

Machine Learning Basics Security and Fairness of Deep Learning Machine Learning Basics Anupam Datta CMU Spring 2019 Image Classification Image Classification Image classification pipeline Input: A training set of N images, each

More information

Spectral Analysis of Jet Substructure with Neural Network: Boosted Higgs Case

Spectral Analysis of Jet Substructure with Neural Network: Boosted Higgs Case Spectral Analysis of Jet Sustructure with Neural Network: Boosted iggs Case Sung ak Lim Supplementary materials for BOOST 28, Paris, France Jul. 28 ased on S.. Lim, M. M. Nojiri, arxiv:87.332 / 26 parton

More information

Convolutional Neural Network Architecture

Convolutional Neural Network Architecture Convolutional Neural Network Architecture Zhisheng Zhong Feburary 2nd, 2018 Zhisheng Zhong Convolutional Neural Network Architecture Feburary 2nd, 2018 1 / 55 Outline 1 Introduction of Convolution Motivation

More information

End-to-End Event Classification of High-Energy Physics Data

End-to-End Event Classification of High-Energy Physics Data End-to-End Event Classification of High-Energy Physics Data M Andrews 1, M Paulini 1, S Gleyzer 2, B Poczos 3 1 Department of Physics, Carnegie Mellon University, Pittsburgh, USA 2 Department of Physics,

More information

Jet Reconstruction and Energy Scale Determination in ATLAS

Jet Reconstruction and Energy Scale Determination in ATLAS Jet Reconstruction and Energy Scale Determination in ATLAS Ariel Schwartzman 3 rd Top Workshop: from the Tevatron to ATLAS Grenoble, 23-Oct-2008 1 Outline ATLAS Calorimeters Calorimeter signal reconstruction:

More information

Convolutional Neural Networks

Convolutional Neural Networks Convolutional Neural Networks Books» http://www.deeplearningbook.org/ Books http://neuralnetworksanddeeplearning.com/.org/ reviews» http://www.deeplearningbook.org/contents/linear_algebra.html» http://www.deeplearningbook.org/contents/prob.html»

More information

Deep Learning Basics Lecture 7: Factor Analysis. Princeton University COS 495 Instructor: Yingyu Liang

Deep Learning Basics Lecture 7: Factor Analysis. Princeton University COS 495 Instructor: Yingyu Liang Deep Learning Basics Lecture 7: Factor Analysis Princeton University COS 495 Instructor: Yingyu Liang Supervised v.s. Unsupervised Math formulation for supervised learning Given training data x i, y i

More information

Variational Autoencoders (VAEs)

Variational Autoencoders (VAEs) September 26 & October 3, 2017 Section 1 Preliminaries Kullback-Leibler divergence KL divergence (continuous case) p(x) andq(x) are two density distributions. Then the KL-divergence is defined as Z KL(p

More information

Deep learning / Ian Goodfellow, Yoshua Bengio and Aaron Courville. - Cambridge, MA ; London, Spis treści

Deep learning / Ian Goodfellow, Yoshua Bengio and Aaron Courville. - Cambridge, MA ; London, Spis treści Deep learning / Ian Goodfellow, Yoshua Bengio and Aaron Courville. - Cambridge, MA ; London, 2017 Spis treści Website Acknowledgments Notation xiii xv xix 1 Introduction 1 1.1 Who Should Read This Book?

More information

GENERATIVE ADVERSARIAL LEARNING

GENERATIVE ADVERSARIAL LEARNING GENERATIVE ADVERSARIAL LEARNING OF MARKOV CHAINS Jiaming Song, Shengjia Zhao & Stefano Ermon Computer Science Department Stanford University {tsong,zhaosj12,ermon}@cs.stanford.edu ABSTRACT We investigate

More information

Boosted Top Tagging with Neural Networks

Boosted Top Tagging with Neural Networks Boosted Top Tagging with Neural Networks Maxim Perelstein Cornell CMS Group Physics Retreat, March 5 2015 Based on work with Leo Almeida, Mihailo Backovic, Mathieu Cliche, Seung Lee [arxiv:1501.05968 +

More information

Resolved Top Tagger in CMS

Resolved Top Tagger in CMS Resolved Top Tagger in CMS An MVA-based tool used for tagging low to moderately boosted hadronic tops Stanislava Sevova, Northwestern University on behalf of CMS Collaboration 1 Overview Motivation Resolved

More information

Digital Calorimetry for Future Linear Colliders. Tony Price University of Birmingham University of Birmingham PPE Seminar 13 th November 2013

Digital Calorimetry for Future Linear Colliders. Tony Price University of Birmingham University of Birmingham PPE Seminar 13 th November 2013 Digital Calorimetry for Future Linear Colliders Tony Price University of Birmingham University of Birmingham PPE Seminar 13 th November 2013 Overview The ILC Digital Calorimetry The TPAC Sensor Electromagnetic

More information

arxiv: v1 [hep-ex] 18 Feb 2009

arxiv: v1 [hep-ex] 18 Feb 2009 GARLIC - GAmma Reconstruciton for the LInear Collider Marcel Reinhard and Jean-Claude Brient arxiv:92.342v1 [hep-ex] 18 Feb 29 LLR - Ecole polytechnique, IN2P3/CNRS Palaiseau - France In order to profit

More information

Convolution and Pooling as an Infinitely Strong Prior

Convolution and Pooling as an Infinitely Strong Prior Convolution and Pooling as an Infinitely Strong Prior Sargur Srihari srihari@buffalo.edu This is part of lecture slides on Deep Learning: http://www.cedar.buffalo.edu/~srihari/cse676 1 Topics in Convolutional

More information

Colin Jessop. University of Notre Dame

Colin Jessop. University of Notre Dame Colin Jessop University of Notre Dame Test the evolution of QCD to higher energies ( and lower x) Search for new particles with Quarks, gluons and photons in Final state Jet production is dominant process

More information

Do you like to be successful? Able to see the big picture

Do you like to be successful? Able to see the big picture Do you like to be successful? Able to see the big picture 1 Are you able to recognise a scientific GEM 2 How to recognise good work? suggestions please item#1 1st of its kind item#2 solve problem item#3

More information

COMPLEX INPUT CONVOLUTIONAL NEURAL NETWORKS FOR WIDE ANGLE SAR ATR

COMPLEX INPUT CONVOLUTIONAL NEURAL NETWORKS FOR WIDE ANGLE SAR ATR COMPLEX INPUT CONVOLUTIONAL NEURAL NETWORKS FOR WIDE ANGLE SAR ATR Michael Wilmanski #*1, Chris Kreucher *2, & Alfred Hero #3 # University of Michigan 500 S State St, Ann Arbor, MI 48109 1 wilmansk@umich.edu,

More information

ECE 521. Lecture 11 (not on midterm material) 13 February K-means clustering, Dimensionality reduction

ECE 521. Lecture 11 (not on midterm material) 13 February K-means clustering, Dimensionality reduction ECE 521 Lecture 11 (not on midterm material) 13 February 2017 K-means clustering, Dimensionality reduction With thanks to Ruslan Salakhutdinov for an earlier version of the slides Overview K-means clustering

More information

LEARNING SPARSE STRUCTURED ENSEMBLES WITH STOCASTIC GTADIENT MCMC SAMPLING AND NETWORK PRUNING

LEARNING SPARSE STRUCTURED ENSEMBLES WITH STOCASTIC GTADIENT MCMC SAMPLING AND NETWORK PRUNING LEARNING SPARSE STRUCTURED ENSEMBLES WITH STOCASTIC GTADIENT MCMC SAMPLING AND NETWORK PRUNING Yichi Zhang Zhijian Ou Speech Processing and Machine Intelligence (SPMI) Lab Department of Electronic Engineering

More information

Modularity Matters: Learning Invariant Relational Reasoning Tasks

Modularity Matters: Learning Invariant Relational Reasoning Tasks Summer Review 3 Modularity Matters: Learning Invariant Relational Reasoning Tasks Jason Jo 1, Vikas Verma 2, Yoshua Bengio 1 1: MILA, Universit de Montral 2: Department of Computer Science Aalto University,

More information

Q-Learning in Continuous State Action Spaces

Q-Learning in Continuous State Action Spaces Q-Learning in Continuous State Action Spaces Alex Irpan alexirpan@berkeley.edu December 5, 2015 Contents 1 Introduction 1 2 Background 1 3 Q-Learning 2 4 Q-Learning In Continuous Spaces 4 5 Experimental

More information

CSE446: Neural Networks Spring Many slides are adapted from Carlos Guestrin and Luke Zettlemoyer

CSE446: Neural Networks Spring Many slides are adapted from Carlos Guestrin and Luke Zettlemoyer CSE446: Neural Networks Spring 2017 Many slides are adapted from Carlos Guestrin and Luke Zettlemoyer Human Neurons Switching time ~ 0.001 second Number of neurons 10 10 Connections per neuron 10 4-5 Scene

More information

Top tagging at CMS. Torben Dreyer on behalf of the CMS Collaboration. BOOST 2017, Buffalo

Top tagging at CMS. Torben Dreyer on behalf of the CMS Collaboration. BOOST 2017, Buffalo Torben Dreyer on behalf of the CMS Collaboration BOOST 2017, Buffalo Motivation LHC at s = 13 TeV More top quarks with high momentum (Standard Model and new physics) Top tagging more important than ever

More information

Deep Learning for Natural Language Processing. Sidharth Mudgal April 4, 2017

Deep Learning for Natural Language Processing. Sidharth Mudgal April 4, 2017 Deep Learning for Natural Language Processing Sidharth Mudgal April 4, 2017 Table of contents 1. Intro 2. Word Vectors 3. Word2Vec 4. Char Level Word Embeddings 5. Application: Entity Matching 6. Conclusion

More information

THE ATLAS TRIGGER SYSTEM UPGRADE AND PERFORMANCE IN RUN 2

THE ATLAS TRIGGER SYSTEM UPGRADE AND PERFORMANCE IN RUN 2 THE ATLAS TRIGGER SYSTEM UPGRADE AND PERFORMANCE IN RUN 2 S. Shaw a on behalf of the ATLAS Collaboration University of Manchester E-mail: a savanna.marie.shaw@cern.ch The ATLAS trigger has been used very

More information

Recent CMS results in the forward region with the CASTOR detector. Sebastian Baur for the CMS Collaboration

Recent CMS results in the forward region with the CASTOR detector. Sebastian Baur for the CMS Collaboration Recent CMS results in the forward region with the CASTOR detector Sebastian Baur for the CMS Collaboration The Forward Instrumentation of CMS 2 Overview CMS has an excellent calorimetric instrumentation

More information

Deep Feedforward Networks. Han Shao, Hou Pong Chan, and Hongyi Zhang

Deep Feedforward Networks. Han Shao, Hou Pong Chan, and Hongyi Zhang Deep Feedforward Networks Han Shao, Hou Pong Chan, and Hongyi Zhang Deep Feedforward Networks Goal: approximate some function f e.g., a classifier, maps input to a class y = f (x) x y Defines a mapping

More information

Lecture 7 Convolutional Neural Networks

Lecture 7 Convolutional Neural Networks Lecture 7 Convolutional Neural Networks CMSC 35246: Deep Learning Shubhendu Trivedi & Risi Kondor University of Chicago April 17, 2017 We saw before: ŷ x 1 x 2 x 3 x 4 A series of matrix multiplications:

More information

ECE521 Lectures 9 Fully Connected Neural Networks

ECE521 Lectures 9 Fully Connected Neural Networks ECE521 Lectures 9 Fully Connected Neural Networks Outline Multi-class classification Learning multi-layer neural networks 2 Measuring distance in probability space We learnt that the squared L2 distance

More information

Quark-gluon tagging in the forward region of ATLAS at the LHC.

Quark-gluon tagging in the forward region of ATLAS at the LHC. CS229. Fall 2016. Final Project Report. Quark-gluon tagging in the forward region of ATLAS at the LHC. Rob Mina (robmina), Randy White (whiteran) December 15, 2016 1 Introduction In modern hadron colliders,

More information

CALICE scintillator HCAL

CALICE scintillator HCAL CALICE scintillator HCAL Erika Garutti DESY (on behalf of the CALICE collaboration) OUTLINE: electromagnetic and hadronic shower analysis shower separation The test beam prototypes 10 GeV pion shower @

More information

Deep Generative Models. (Unsupervised Learning)

Deep Generative Models. (Unsupervised Learning) Deep Generative Models (Unsupervised Learning) CEng 783 Deep Learning Fall 2017 Emre Akbaş Reminders Next week: project progress demos in class Describe your problem/goal What you have done so far What

More information

Experiments on the Consciousness Prior

Experiments on the Consciousness Prior Yoshua Bengio and William Fedus UNIVERSITÉ DE MONTRÉAL, MILA Abstract Experiments are proposed to explore a novel prior for representation learning, which can be combined with other priors in order to

More information

Feedforward Neural Networks

Feedforward Neural Networks Feedforward Neural Networks Michael Collins 1 Introduction In the previous notes, we introduced an important class of models, log-linear models. In this note, we describe feedforward neural networks, which

More information

TUTORIAL PART 1 Unsupervised Learning

TUTORIAL PART 1 Unsupervised Learning TUTORIAL PART 1 Unsupervised Learning Marc'Aurelio Ranzato Department of Computer Science Univ. of Toronto ranzato@cs.toronto.edu Co-organizers: Honglak Lee, Yoshua Bengio, Geoff Hinton, Yann LeCun, Andrew

More information

Artificial Neural Networks D B M G. Data Base and Data Mining Group of Politecnico di Torino. Elena Baralis. Politecnico di Torino

Artificial Neural Networks D B M G. Data Base and Data Mining Group of Politecnico di Torino. Elena Baralis. Politecnico di Torino Artificial Neural Networks Data Base and Data Mining Group of Politecnico di Torino Elena Baralis Politecnico di Torino Artificial Neural Networks Inspired to the structure of the human brain Neurons as

More information

Deep Learning Architecture for Univariate Time Series Forecasting

Deep Learning Architecture for Univariate Time Series Forecasting CS229,Technical Report, 2014 Deep Learning Architecture for Univariate Time Series Forecasting Dmitry Vengertsev 1 Abstract This paper studies the problem of applying machine learning with deep architecture

More information

Status of the physics validation studies using Geant4 in ATLAS

Status of the physics validation studies using Geant4 in ATLAS Status of the physics validation studies using Geant4 in ATLAS On behalf of the ATLAS Geant4 Validation Team A.Dell Acqua CERN EP/SFT, Geneva, CH dellacqu@mail.cern.ch The new simulation for the ATLAS

More information

PCA and Autoencoders

PCA and Autoencoders PCA and Autoencoders Tyler Manning-Dahan INSE 6220 - Fall 2017 Concordia University Abstract In this paper, I compare two dimensionality reduction techniques for processing images before learning a multinomial

More information

Jets in the 21st Century

Jets in the 21st Century LoLa Jets in the 2st Century Universität Heidelberg Mainz /28 LoLa Rise of the Machines Brief history of jets 994 jet-algo W -tagger for heavy Higgs [Seymour] 994 jet-algo top tagger for fun [Seymour]

More information

Jet energy measurement in the ATLAS detector

Jet energy measurement in the ATLAS detector Jet energy measurement in the ATLAS detector Jet energy scale uncertainties are usually among largest experimental uncertainties Need precise jet energy measurements and provide uncertainties and theit

More information

arxiv: v1 [stat.ml] 24 Nov 2016

arxiv: v1 [stat.ml] 24 Nov 2016 EPJ Web of Conferences will be set by the publisher DOI: will be set by the publisher c Owned by the authors, published by EDP Sciences, 28 arxiv:6.8256v [stat.ml] 24 Nov 26 The Algorithm: Anomaly Detection

More information

Collider Physics Analysis Procedures

Collider Physics Analysis Procedures Collider Physics Analysis Procedures Alex Tapper Slides available at: http://www.hep.ph.ic.ac.uk/~tapper/lecture.html Aim Overview of analysis techniques at CMS Contrast with Tevatron (see DØ lecture)

More information

Recurrent Neural Networks with Flexible Gates using Kernel Activation Functions

Recurrent Neural Networks with Flexible Gates using Kernel Activation Functions 2018 IEEE International Workshop on Machine Learning for Signal Processing (MLSP 18) Recurrent Neural Networks with Flexible Gates using Kernel Activation Functions Authors: S. Scardapane, S. Van Vaerenbergh,

More information

Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks

Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks Emily Denton 1, Soumith Chintala 2, Arthur Szlam 2, Rob Fergus 2 1 New York University 2 Facebook AI Research Denotes equal

More information

Characterization of Jet Charge at the LHC

Characterization of Jet Charge at the LHC Characterization of Jet Charge at the LHC Thomas Dylan Rueter, Krishna Soni Abstract The Large Hadron Collider (LHC) produces a staggering amount of data - about 30 petabytes annually. One of the largest

More information

Neural Networks. Single-layer neural network. CSE 446: Machine Learning Emily Fox University of Washington March 10, /9/17

Neural Networks. Single-layer neural network. CSE 446: Machine Learning Emily Fox University of Washington March 10, /9/17 3/9/7 Neural Networks Emily Fox University of Washington March 0, 207 Slides adapted from Ali Farhadi (via Carlos Guestrin and Luke Zettlemoyer) Single-layer neural network 3/9/7 Perceptron as a neural

More information

CSC321 Lecture 20: Reversible and Autoregressive Models

CSC321 Lecture 20: Reversible and Autoregressive Models CSC321 Lecture 20: Reversible and Autoregressive Models Roger Grosse Roger Grosse CSC321 Lecture 20: Reversible and Autoregressive Models 1 / 23 Overview Four modern approaches to generative modeling:

More information

UNSUPERVISED LEARNING

UNSUPERVISED LEARNING UNSUPERVISED LEARNING Topics Layer-wise (unsupervised) pre-training Restricted Boltzmann Machines Auto-encoders LAYER-WISE (UNSUPERVISED) PRE-TRAINING Breakthrough in 2006 Layer-wise (unsupervised) pre-training

More information

GraphRNN: A Deep Generative Model for Graphs (24 Feb 2018)

GraphRNN: A Deep Generative Model for Graphs (24 Feb 2018) GraphRNN: A Deep Generative Model for Graphs (24 Feb 2018) Jiaxuan You, Rex Ying, Xiang Ren, William L. Hamilton, Jure Leskovec Presented by: Jesse Bettencourt and Harris Chan March 9, 2018 University

More information

<Special Topics in VLSI> Learning for Deep Neural Networks (Back-propagation)

<Special Topics in VLSI> Learning for Deep Neural Networks (Back-propagation) Learning for Deep Neural Networks (Back-propagation) Outline Summary of Previous Standford Lecture Universal Approximation Theorem Inference vs Training Gradient Descent Back-Propagation

More information

Deep Convolutional Neural Networks for Pairwise Causality

Deep Convolutional Neural Networks for Pairwise Causality Deep Convolutional Neural Networks for Pairwise Causality Karamjit Singh, Garima Gupta, Lovekesh Vig, Gautam Shroff, and Puneet Agarwal TCS Research, Delhi Tata Consultancy Services Ltd. {karamjit.singh,

More information

Top quarks objects definition and performance at ATLAS

Top quarks objects definition and performance at ATLAS 5th International Workshop on op Quark Physics (OP) doi:.88/74-6596/45// op quarks objects definition and performance at ALAS V. Boisvert on behalf of the ALAS Collaboration Royal Holloway University of

More information

Lecture 5 Neural models for NLP

Lecture 5 Neural models for NLP CS546: Machine Learning in NLP (Spring 2018) http://courses.engr.illinois.edu/cs546/ Lecture 5 Neural models for NLP Julia Hockenmaier juliahmr@illinois.edu 3324 Siebel Center Office hours: Tue/Thu 2pm-3pm

More information

Evidence for tth production at ATLAS

Evidence for tth production at ATLAS Evidence for tth production at ATLAS Rohin T Narayan on behalf of the ATLAS collaboration Nov-6-2017, Higgs couplings Heidelberg Motivation Fermion masses are generated through Yukawa interaction Heaviest

More information

Thesis. Wei Tang. 1 Abstract 3. 3 Experiment Background The Large Hadron Collider The ATLAS Detector... 4

Thesis. Wei Tang. 1 Abstract 3. 3 Experiment Background The Large Hadron Collider The ATLAS Detector... 4 Thesis Wei Tang Contents 1 Abstract 3 2 Introduction 3 3 Experiment Background 4 3.1 The Large Hadron Collider........................... 4 3.2 The ATLAS Detector.............................. 4 4 Search

More information

Preparations for the ATLAS Heavy Ion Physics Program at the LHC. Deepak Kar IKTP, TU Dresden On behalf of the ATLAS Collaboration

Preparations for the ATLAS Heavy Ion Physics Program at the LHC. Deepak Kar IKTP, TU Dresden On behalf of the ATLAS Collaboration Preparations for the ATLAS Heavy Ion Physics Program at the LHC Deepak Kar IKTP, TU Dresden On behalf of the ATLAS Collaboration 1 QCD Hadronic phase: Bound states of quark and gluon pp collisions Heavy

More information

Artificial Neural Networks. Introduction to Computational Neuroscience Tambet Matiisen

Artificial Neural Networks. Introduction to Computational Neuroscience Tambet Matiisen Artificial Neural Networks Introduction to Computational Neuroscience Tambet Matiisen 2.04.2018 Artificial neural network NB! Inspired by biology, not based on biology! Applications Automatic speech recognition

More information

Jet Substructure In ATLAS

Jet Substructure In ATLAS Jet Substructure In ATLAS INFN Pisa & University of Arizona Parton Showers & Event Structure At The LHC (Northwest Terascale Research Projects Workshop) University of Oregon February 23-27, 2009 Overview

More information

An EoS-meter of QCD transition from deep learning

An EoS-meter of QCD transition from deep learning An EoS-meter of QCD transition from deep learning Nan Su Frankfurt Institute for Advanced Studies with Long-Gang Pang, Kai Zhou (FIAS), Hannah Petersen, Horst Stöcker (FIAS/Uni Frankfurt/GSI), Xin-Nian

More information

Direct measurement of the W boson production charge asymmetry at CDF

Direct measurement of the W boson production charge asymmetry at CDF Direct measurement of the boson production charge asymmetry at CDF Eva Halkiadakis Rutgers University For the CDF collaboration Joint Experimental-Theoretical Physics Seminar Fermilab May 22 2009 Outline

More information

Dimensionality Reduction and Principle Components Analysis

Dimensionality Reduction and Principle Components Analysis Dimensionality Reduction and Principle Components Analysis 1 Outline What is dimensionality reduction? Principle Components Analysis (PCA) Example (Bishop, ch 12) PCA vs linear regression PCA as a mixture

More information

Physics with Tau Lepton Final States in ATLAS. Felix Friedrich on behalf of the ATLAS Collaboration

Physics with Tau Lepton Final States in ATLAS. Felix Friedrich on behalf of the ATLAS Collaboration Physics with Tau Lepton Final States in ATLAS on behalf of the ATLAS Collaboration HEP 2012, Valparaiso (Chile), 06.01.2012 The Tau Lepton m τ = 1.8 GeV, heaviest lepton cτ = 87 μm, short lifetime hadronic

More information

Feedforward Neural Networks. Michael Collins, Columbia University

Feedforward Neural Networks. Michael Collins, Columbia University Feedforward Neural Networks Michael Collins, Columbia University Recap: Log-linear Models A log-linear model takes the following form: p(y x; v) = exp (v f(x, y)) y Y exp (v f(x, y )) f(x, y) is the representation

More information

Neural Architectures for Image, Language, and Speech Processing

Neural Architectures for Image, Language, and Speech Processing Neural Architectures for Image, Language, and Speech Processing Karl Stratos June 26, 2018 1 / 31 Overview Feedforward Networks Need for Specialized Architectures Convolutional Neural Networks (CNNs) Recurrent

More information

Demystifying Multivariate Searches

Demystifying Multivariate Searches Demystifying Multivariate Searches and the Matthew Schwartz Harvard University Work down with Jason Gallicchio, PRL, 105:022001,2010 and with Gallicchio, Tweedie, Huth, Kagan and Black in preparation Johns

More information

Cosmic Ray Physics with the ARGO-YBJ experiment

Cosmic Ray Physics with the ARGO-YBJ experiment Cosmic Ray Physics with the ARGO-YBJ experiment Ivan De Mitri University of Salento and Istituto Nazionale di Fisica Nucleare Lecce, Italy On behalf of the ARGO-YBJ Collaboration Second Roma International

More information

Performance of the FastTracKer in ATLAS

Performance of the FastTracKer in ATLAS Performance of the FastTracKer in ATLAS Maddalena Giulini under supervision of Prof. A. Schöning and Dr. T. Klimkovich Physikalisches Institut, Universität Heidelberg DPG Spring Conference, Mainz March

More information

Search for tt(h bb) using the ATLAS detector at 8 TeV

Search for tt(h bb) using the ATLAS detector at 8 TeV Search for tt(h bb) using the ATLAS detector at 8 TeV on behalf of the ATLAS Collaboration University of Göttingen Motivation, Strategy & Introduction Overview: tt(h bb) leptonic analysis Summary & Outlook

More information