Applied cloud deep semantic recognition advanced anomaly detection

Applied Cloud Deep Semantic Recognition Advanced Anomaly Detection Applied Cloud Deep Semantic Recognition Advanced Anomaly Detection Edited by Mehdi Roopaei Paul Rad CRC Press Taylor & Francis Group 6000 Broken Sound Parkway NW, Suite 300 Boca Raton, FL 33487-2742 c 2018 by Taylor & Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group, an Informa business No claim to original U.S Government works Printed on acid-free paper International Standard Book Number-13: 978-1-138-30222-8 (Hardback) This book contains information obtained from authentic and highly regarded sources Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use The authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint Except as permitted under U.S Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400 CCC is a not-for-profit organization that provides licenses and registration for a variety of users For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com and the CRC Press Web site at http://www.crcpress.com Contents Contributors vii Introduction .ix Large-Scale Video Event Detection Using Deep Neural Networks GUANGNAN YE Leveraging Selectional Preferences for Anomaly Detection in Newswire Events 25 PRADEEP DASIGI AND EDUARD HOVY Abnormal Event Recognition in Crowd Environments .37 MOIN NABI, HOSSEIN MOUSAVI, HAMIDREZA RABIEE, MAHDYAR RAVANBAKHSH, VITTORIO MURINO, AND NICU SEBE Cognitive Sensing: Adaptive Anomalies Detection with Deep Networks .57 CHAO WU AND YIKE GUO Language-Guided Visual Recognition 87 MOHAMED ELHOSEINY, YIZHE (ETHAN) ZHU, AND AHMED ELGAMMAL Deep Learning for Font Recognition and Retrieval 109 ZHANGYANG WANG, JIANCHAO YANG, HAILIN JIN, ZHAOWEN WANG, ELI SHECHTMAN, ASEEM AGARWALA, JONATHAN BRANDT, AND THOMAS S HUANG A Distributed Secure Machine-Learning Cloud Architecture for Semantic Analysis 131 ARUN DAS, WEI-MING LIN, AND PAUL RAD A Practical Look at Anomaly Detection Using Autoencoders with H2O and the R Programming Language 161 MANUEL AMUNATEGUI Index 179 v Contributors Aseem Agarwala Google Inc Seattle, Washington Manuel Amunategui SpringML, Inc Portland, Oregon Jonathan Brandt Adobe Research San Jose, California Arun Das Electrical and Computer Engineering University of Texas at San Antonio San Antonio, Texas Pradeep Dasigi Language Technologies Institute Carnegie Mellon University Pittsburgh, Pennsylvania Thomas S Huang Beckman Institute University of Illinois at Urbana-Champaign Champaign, Illinois Hailin Jin Adobe Research San Jose, California Wei-Ming Lin Electrical and Computer Engineering University of Texas at San Antonio San Antonio, Texas Hossein Mousavi Polytechnique Montréal Montréal, Québec, Canada Vittorio Murino PAVIS Department Istituto Italiano di Tecnologia Genoa, Italy Ahmed Elgammal Rutgers University Piscataway, New Jersey Moin Nabi SAP SE Berlin, Germany Mohamed Elhoseiny Facebook AI Research Monterey Park, California Hamidreza Rabiee Azad University of Karaj Karaj, Iran Yike Guo Data Science Institute Imperial College London London, United Kingdom Paul Rad Electrical and Computer Engineering University of Texas at San Antonio San Antonio, Texas Eduard Hovy Language Technologies Institute Carnegie Mellon University Pittsburgh, Pennsylvania Mahdyar Ravanbakhsh DITEN University of Genova Genova, Italy vii viii Contributors Mehdi Roopaei University of Texas at San Antonio San Antonio, Texas Nicu Sebe DISI University of Trento Trentino, Italy Eli Shechtman Adobe Research San Jose, California Zhangyang Wang Department of Computer Science and Engineering Texas A&M University College Station, Texas Zhaowen Wang Adobe Research San Jose, California Chao Wu Data Science Institute Imperial College London London, United Kingdom Jianchao Yang Snapchat Inc Venice, California Guangnan Ye IBM T.J Watson Research Center Yorktown Heights, New York Yizhe (Ethan) Zhu Rutgers University Piscataway, New Jersey Introduction Anomaly Detection and Situational Awareness In data analytics, anomaly detection is discussed as the discovery of objects, actions, behavior, or events that not conform to an expected pattern in a dataset Anomaly detection has extensive applications in a wide variety of domains such as biometrics spoofing, healthcare, fraud detection for credit cards, network intrusion detection, malware threat detection, and military surveillance for adversary threats While anomalies might be induced in the data for a variety of motives, all of the motives have the common trait that they are interesting to data scientists and cyber analysts Anomaly detection has been researched within diverse research areas such as computer science, engineering, information systems, and cyber security Many anomaly detection algorithms have been presented for certain domains, while others are more generic In the past, many anomaly detection algorithms have been designed for specific applications, while others are more generic This book tries to provide a comprehensive overview of the research on anomaly detection with respect to context and situational awareness that aims to get a better understanding of how context information influences anomaly detection We have grouped scalars from industry and academic with vast practical knowledge into different In each chapter, advanced anomaly detection and key assumptions have been identified, which are used by the model to differentiate between normal and anomalous behavior When applying a given model to a particular application, the assumptions can be used as guidelines to assess the effectiveness of the model in that domain In each chapter, we provide an advanced deep content understanding and anomaly detection algorithm, and then we show how the proposed approach deviates from basic techniques Further, for each chapter, we describe the advantages and disadvantages of the algorithm Last but not least, in the final chapters, we also provide a discussion on the computational complexity of the models and graph computational frameworks such as Google Tensorflow and H2O, since it is an important issue in real application domains We hope that this book will provide a better understanding of the different directions in which research has been done on deep semantic analysis and situational assessment using deep learning for anomalous detection, and how methods developed in one area can be applied to applications in other domains This book seeks to provide both cyber analytics practitioners and researchers with an up-todate and advanced knowledge in cloud-based frameworks for deep semantic analysis and advanced anomaly detection using cognitive and artificial intelligence (AI) models The structure of the remainder of this book is as follows ix 174 Applied Cloud Deep Semantic Recognition 8.10.1 Base GLMNET Model The UCI diabetes dataset we downloaded earlier is designed for classification as it contains a binary outcome of whether or not a patient was readmitted to the hospital for the same problem Whenever the “readmitted” feature is equal to “< 30,” this means that the patient was readmitted in less than 30 days and will represent our positive outcome; all other cases will represent our negative outcome # prep outcome variable to those readmitted under 30 days diabetes$readmitted