DOCSLIB.ORG

Learning Based Visual Engagement and Self-Efficacy

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Learning Based Visual Engagement and Self-Efficacy LEARNING BASED VISUAL ENGAGEMENT AND SELF-EFFICACY by SVATI DHAMIJA B.Tech., Electronics and Communications Engineering, Punjab Technical University, India 2007 M.Tech., Electronics and Communications Engineering, Guru Gobind Singh Indraprastha University, India 2011 A dissertation submitted to the Graduate Faculty of the University of Colorado Colorado Springs in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Computer Science 2018 This dissertation for the Doctor of Philosophy degree by Svati Dhamija has been approved for the Department of Computer Science by Terrance E. Boult, Chair 1 Charles C. Benight 2 Manuel Gunther¨ 1 Rory A. Lewis 1 Jonathan Ventura 3 Walter J. Scheirer 4 Date: 12-03-2018 1 T.Boult, R.Lewis and M.Gunther¨ are with University of Colorado Colorado Springs 2 C.Benight is with Psychology Department in University of Colorado Colorado Springs 3 J.Ventura is with California Polytechnic State University 4 W.Scheirer is with University of Notre Dame ii Dhamija, Svati (Ph.D., Computer Science) Learning Based Visual Engagement and Self-Efficacy Dissertation directed by El Pomar Professor, Chair Terrance E. Boult Abstract With the advancements in the fields of Affective computing, Computer vision and Human-computer interaction, along with the exponential development of intelligent machines, the study of human mind and behavior is emerging as an unrivaled mystery. A plethora of experiments are being carried out each day to make machines capable enough to sense subtle verbal & non-verbal human behaviors and understand human needs in every sphere of life. Numerous applications ranging from virtual assistants for online-learning to socially-assistive robots for health care, are being designed. A frequent challenge in such applications is creating scalable systems that are tailored to an individual’s capabilities, preferences and response to interaction. Moreover, digital interventions that are tailored to induce cognitive and behavioral changes in individuals are common, but often lack tools for automated analysis and monitoring of user behavior while interacting with the application. Lack of automated monitoring is especially detrimental in aggravating societal problems like mental illnesses and disorders, where it is important to determine the psychological wellbeing of populations, to detect people at risk, to personalize and deliver interventions that work in real-world. Prior research indicates that web-interventions have the potential to advance patient-centered mental health care delivery and can estimate patient involvement by recognizing human emotions and affect. Personalized computing systems with capabilities of perception, rational and intelligent behavior can also be developed by observing bio-physiological responses and facial expressions. Though invention of such systems is slowly improving the prediction and detection of stress levels, it is farther away from adapting to an individual’s specific need of cure and advise coping methods. Given the stigma associated with mental disorders, people may not always seek assistance. Engagement or disengagement with applications can be an important indicator of individual’s iii desire to be treated and whether they are interested in coping with their situation. A scalable adaptive person-centered approach is therefore essential to non-invasively monitor user- engagement and response to the application. In this work, we present a novel vision and learning based approach for web-based interventions with the use case of trauma-recovery. This thesis comprises of my recent research which develops contextual engagement, mood-aware engagement and multi-modal machine learning algorithms in a non-intrusive setting with EASE (Engagement Arousal & Self-Efficacy), by calibrating Engagement from face videos and physiological Arousal leading towards a hidden state of Self-Efficacy. iv Dedicated to my family. Thank you for always being there for me. Acknowledgements First and foremost, I would like to express my sincerest regards to my advisor Dr. Terrance E. Boult. Without his guidance and supervision, this thesis would not be possible. His unique ways of mentoring and vision of the big picture inspired me to think objectively and taught me the fundamentals of research. I would also like to extend gratitude to my committee members: Dr. Benight, Dr. Gunther,¨ Dr. Ventura, Dr. Scheirer and Dr. Lewis for setting high expectations and propelling me in fruitful directions. I thank the financial support of NSF Research Grant SCH-INT 1418520 “Learning and Sensory-based Modeling for Adaptive Web-Empowerment Trauma Treatment” for this work. I am grateful to numerous VAST lab members who made the workplace fun and engaging, Archana, Abhijit, Chloe, Steve, Manuel, Lalit, Gauri, Dubari, Khang, Ankita, Dan R., Caleb, James, Ethan, Marc, Bill, Yousef, Adria and Akshay. I would also like to acknowledge the support of my fellow researchers from psychology, Kotaro, Carrie, Austin, Pamela, Amanda and Shaun, who worked alongside me through the tardy and laborious process of data collection. Special thanks to Ginger for the numerous adventures we had together, including ski trips, hikes, city tours, fun-filled dinners, board games etc. She made colorado springs a home away from home. My encouraging friends Pooja, Reetu, Komal, Cherry, Radhika, Silvee, Himani and Archana, who make life happier in many ways despite the long distances and different time-zones that separate us. Last but not the least, I thank my parents and in-laws for being the source of inspiration and instilling me with courage. My grandparents for their unconditional love and blessings. Mumma and Papa for their countless sacrifices and prayers over the years. Thank you for keeping me grounded, making me aware of glass ceilings and trusting my abilities to break them. My brother, Akshay, for being my decisive and reassuring agony-uncle. My husband, Abhijit, for being my pillar of strength, friend and mentor. Thank you for braving this journey with me. Finally, I thank the almighty for watching over me through the activities of mind, senses and emotions. vi Contents 1 Introduction 1 1.1 Why Engagement? . 2 1.2 Why Self-Efficacy? . 3 1.3 Background of application space : Trauma-Recovery . 3 1.4 Limitations of prior work . 5 1.5 Scope of this research . 7 1.6 Contributions from this thesis . 8 1.6.1 Dataset Collection . 8 1.6.2 Contextual Engagement Prediction from video . 9 1.6.3 Mood-aware Engagement Prediction . 9 1.6.4 Automated Action Units vs. Expert Raters . 9 1.6.5 Self-Efficacy prediction: Problem formulation . 10 1.6.6 Predicting Change in Self-Efficacy using EA features . 10 1.7 Publications from this work . 11 1.7.1 First author publications . 11 1.7.2 Other work . 11 2 EASE Dataset Collection 13 2.1 Introduction . 13 vii 2.2 Related Work . 13 2.3 EASE Data Collection Setup . 15 2.4 Participants and Demographics . 16 2.5 The EASE space variables . 17 2.5.1 Engagement measures . 17 2.5.2 Arousal data . 19 2.5.3 Trauma-recovery outcomes and dependent variables . 20 2.6 Challenges . 21 3 Contextual Engagement Prediction from video 23 3.1 Introduction . 23 3.2 Related work . 26 3.3 Algorithms for Sequence Learning . 29 3.3.1 Recurrent Neural Network . 30 3.3.2 Gated Recurrent Units . 31 3.3.3 Long Short-Term Memory . 32 3.4 Experiments . 33 3.4.1 Facial action units extraction . 33 3.4.2 SVC model . 35 3.4.3 Sequence learning models and tuning . 35 3.4.4 Context-specific, Cross-context and Combined models . 36 3.4.5 Varying data-segment length . 36 3.5 Evaluation results . 38 3.5.1 Contextual engagement results . 38 3.5.2 Comparison results for sequence learning algorithms . 38 3.6 Discussion . 42 3.6.1 Performance of Contextual Engagement Models . 42 viii 3.6.2 Incorporating Broader Range of Data . 44 4 Mood-Aware Engagement Prediction 46 4.1 Introduction . 46 4.2 Related Work . 49 4.3 Experiments . 50 4.3.1 Profile of Mood States (POMS) . 50 4.3.2 Facial action units extraction . 51 4.3.3 LSTM Engagement and Mood Models . 52 4.3.4 Cross-Validation and Train-test pipeline . 54 4.4 Evaluation Results . 56 4.5 Broader Impact . 57 4.5.1 Robust Mood and Mood-Aware Engagement Models . 59 5 Automated Action Units Vs. Expert Raters 61 5.1 Introduction . 61 5.2 Related Work . 63 5.3 Sequential learning model using RNN: automated, expert raters . 65 5.4 EASE Subset & Annotations . 66 5.5 Machine Learning Models . 68 5.6 Model Training/Tuning . 71 5.7 Results . 73 5.8 Discussion . 75 6 Self-Efficacy: Problem formulation and Computation 77 6.1 Introduction . 77 6.2 EASE Coping Self-Efficacy Measures . 79 6.3 Effectiveness of CSE in estimating PTSD symptom severity . 80 ix 6.4 Observations from related work . 84 6.4.1 Engagement and Self-Efficacy . 86 6.4.2 Heart Rate Variability and Self-Efficacy . 87 6.5 Defining CSE as a Machine Learning problem . 89 6.6 CSE Prediction Pipeline . 90 6.7 EASE subset and Data distribution . 91 6.8 Experiments . ..
Load more

Recommended publications
  • HONGLAK LEE 2260 Hayward St, Beyster Building Room 3773, Ann Arbor, MI 48109 [email protected]
    HONGLAK LEE 2260 Hayward St, Beyster Building Room 3773, Ann Arbor, MI 48109 http://www.eecs.umich.edu/~honglak [email protected] EDUCATION 9/2010 Stanford University, Stanford, CA Ph.D. in Computer Science Thesis title: Unsupervised Feature Learning via Sparse Hierarchical Representations Thesis advisor: Professor Andrew Y. Ng 6/2006 Stanford University, Stanford, CA M.S. in Computer Science M.S. in Applied Physics 2/2003 Seoul National University, Seoul, Korea B.S. in Physics and Computer Science, GPA 4.14/4.30, Summa Cum Laude PROFESSIONAL EXPERIENCE 9/2010 – Present Assistant Professor, CSE division, EECS department, University of Michigan, Ann Arbor, MI 1/2005 – 8/2010 Research Assistant, Stanford AI Lab, Stanford University, Stanford, CA 8/1999 –1/2002 Software Engineer, ECO Co. Ltd., Seoul, Korea AWARDS & HONORS National Science Foundation CAREER Award, 2015 AI’s 10 to Watch, IEEE Intelligent Systems, 2013 Google Faculty Research Award, 2011 Research Highlights in Communications of the ACM, 2011 Best Paper Award: Best Application Paper, International Conference on Machine Learning (ICML), 2009 Best Student Paper Award, Conference on Email and Anti-Spam (CEAS), 2005 Stanford Graduate Fellowship, Stanford University, 2003–2006 Graduate Fellowship from Korea Foundation for Advanced Studies (KFAS), 2003–2008 College Student Scholarship from Korea Foundation for Advanced Studies (KFAS), 1997–2001 Honorable Mention in the 16th University Students Contest of Mathematics, 1997 Undergraduate Scholarship from Korea Science and Engineering Foundation (KOSEF), 1996–1999 The First Place in the Entrance Exam of College of Natural Sciences, Seoul National University, 1996 The 12th place out of 344,780 applicants in the College Scholastic Ability Test (CSAT), Korea, 1996 Silver Medal in the 26th International Physics Olympiad, Canberra, Australia, 1995 REFEREED CONFERENCE AND JOURNAL PUBLICATIONS 1.
    [Show full text]
  • Make3d: Learning 3D Scene Structure from a Single Still Image Ashutosh Saxena, Min Sun and Andrew Y
    1 Make3D: Learning 3D Scene Structure from a Single Still Image Ashutosh Saxena, Min Sun and Andrew Y. Ng Abstract— We consider the problem of estimating detailed 3-d structure from a single still image of an unstructured environment. Our goal is to create 3-d models which are both quantitatively accurate as well as visually pleasing. For each small homogeneous patch in the image, we use a Markov Random Field (MRF) to infer a set of “plane parame- ters” that capture both the 3-d location and 3-d orientation of the patch. The MRF, trained via supervised learning, models both image depth cues as well as the relationships between different parts of the image. Other than assuming that the environment is made up of a number of small planes, our model makes no explicit assumptions about the structure of the scene; this enables the algorithm to capture much more detailed 3-d structure than does prior art, and also give a much richer experience in the 3-d flythroughs created using image-based rendering, even for scenes with significant non-vertical structure. Using this approach, we have created qualitatively correct 3-d models for 64.9% of 588 images downloaded from the internet. We have also extended our model to produce large scale 3d models from a few images.1 Fig. 1. (a) An original image. (b) Oversegmentation of the image to obtain Index Terms— Machine learning, Monocular vision, Learning “superpixels”. (c) The 3-d model predicted by the algorithm. (d) A screenshot depth, Vision and Scene Understanding, Scene Analysis: Depth of the textured 3-d model.
    [Show full text]
  • AAAI-12 Conference Committees
    AAAI 2012 Conference Committees Chairs and Cochairs AAAI Conference Committee Chair Dieter Fox (University of Washington, USA) AAAI­12 Program Cochairs Jörg Hoffmann (Saarland University, Germany) Bart Selman (Cornell University, USA) IAAI­12 Conference Chair and Cochair Markus Fromherz (ACS, a Xerox Company, USA) Hector Munoz‐Avila (Lehigh University, USA) EAAI­12 Symposium Chair David Kauchak (Middlebury College, USA) Special Track on Artificial Intelligence and the Web Cochairs Denny Vrandecic (Institute of Applied Informatics and Formal Description Methods, Germany) Chris Welty (IBM Research, USA) Special Track on Cognitive Systems Cochairs Matthias Scheutz (Tufts University, USA) James Allen (University of Rochester, USA) Special Track on Computational Sustainability and Artificial Intelligence Cochairs Carla P. Gomes (Cornell University, USA) Brian C. Williams (Massachusetts Institute of Technology, USA) Special Track on Robotics Cochairs Kurt Konolige (Industrial Perception, Inc., USA) Siddhartha Srinivasa (Carnegie Mellon University, USA) Turing Centenary Events Chair Toby Walsh (NICTA and University of New South Wales, Australia) Tutorial Program Cochairs Carmel Domshlak (Technion Israel Institute of Technology, Israel) Patrick Pantel (Microsoft Research, USA) Workshop Program Cochairs Michael Beetz (University of Munich, Germany) Holger Hoos (University of British Columbia, Canada) Doctoral Consortium Cochairs Elizabeth Sklar (Brooklyn College, City University of New York, USA) Peter McBurney (King’s College London, United Kingdom)
    [Show full text]
  • Ashesh Jain 142 Gates Building, Stanford University, CA 94305 [email protected]
    Ashesh Jain 142 Gates Building, Stanford University, CA 94305 [email protected], www.cs.cornell.edu/~ashesh Interests My research interest lies at the intersection of machine learning, robotics and computer vision. Broadly, I build machine learning systems & algorithms for agents { such as robots, cars etc. { to learn from informative human signals at a large-scale. Most of my work has been in multi-modal sensor-rich robotic settings, for which I have developed sensory fusion deep learning architectures. I have developed and deployed algorithms on multiple robotic platforms (PR2, Baxter etc.), on cars, and crowd-sourcing systems. Education Cornell University, New York, USA (2012-2016 (Exptd)) Ph.D. student, Computer Science • PhD committee: Ashutosh Saxena (advisor), Thorsten Joachims, Doug James, Robert Kleinberg, and Bart Selman. • Learning from large-scale human signals for robots and assistive cars • First author papers in NIPS, ICCV, SIGKDD, ICRA, ISRR, and IJRR Indian Institute of Technology Delhi, India (2007-2012) B.Tech. Electrical Engineering & M.Tech. Information and Communication Technology • Advisor: Prof. Manik Varma and Prof. S.V.N. Vishwanathan • Thesis: Large-scale Algorithms for Multiple Kernel Learning Experience Stanford University, California, USA (2014-Present) Visiting Ph.D. student, Computer Science • Working with Prof. Ashutosh Saxena and Prof. Silvio Savarese • Leading Brain4Cars and Machine Learning lead on RoboBrain. Purdue University, Indiana, USA (Summer 2011) Research Intern, Statistics Department with Prof. S. V. N. Vishwanathan • Developed Multiple Kernel Learning algorithms to scale to Millions of kernels The Royal Bank of Scotland, India (Summer 2010) • Developed software to automate testing of web application GUI Software NeuralModels: A deep learning framework for quick prototyping of structures of Recurrent Neural Networks, Sensory-fusion architectures, and deep learning on graph structured data.
    [Show full text]
  • Unsupervised Structured Learning of Human Activities for Robot Perception
    UNSUPERVISED STRUCTURED LEARNING OF HUMAN ACTIVITIES FOR ROBOT PERCEPTION A Dissertation Presented to the Faculty of the Graduate School of Cornell University in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy by Chenxia Wu August 2016 © 2016 Chenxia Wu ALL RIGHTS RESERVED UNSUPERVISED STRUCTURED LEARNING OF HUMAN ACTIVITIES FOR ROBOT PERCEPTION Chenxia Wu, Ph.D. Cornell University 2016 Learning human activities and environments is important for robot perception. Human activities and environments comprise many aspects, including a wide variety of human actions and various objects that interact with humans, which make their modeling very challenging. We observe that these aspects are re- lated to each other spatially, temporally and semantically. They form sequen- tial, hierarchical or graph structures. Understanding these structures is key to the learning algorithms and systems of robot perception. Therefore, this thesis focuses on structured modeling of these complex hu- man activities and environments using unsupervised learning. Our unsupervised learning approaches can detect hidden structures from the data itself, without the need for human annotations. In this way, we enable more useful applica- tions, such as forgotten action detection and object co-segmentation. While structured models in supervised settings have been well-studied and widely used in various domains, discovering latent structures is still a chal- lenging problem in unsupervised learning. In this work, we propose unsuper- vised structured learning models, including causal topic models and fully con- nected Conditional Random Field (CRF) auto-encoders, which have the ability to model more complex relations with less independence. We also design effi- cient learning and inference optimizations that maintain the tractability of com- putations.
    [Show full text]
  • Ashesh Jain 142 Gates Building, Stanford University, CA 94305 [email protected]
    Ashesh Jain 142 Gates Building, Stanford University, CA 94305 [email protected], www.cs.cornell.edu/~ashesh Interests My research interest lies at the intersection of machine learning, robotics, and computer vision. Broadly, I build machine learning systems & algorithms for agents { such as robots, cars etc. { to learn from informative human signals at a large-scale. Most of my work has been in multi-modal sensor-rich robotic settings, for which I have developed sensory fusion deep learning architectures. I have developed and deployed algorithms on multiple robotic platforms (PR2, Baxter etc.), on cars, and crowd-sourcing systems. Education Cornell University, New York, USA (2012-2016) Ph.D. Computer Science • Specializing in Machine Learning, Robotics, and Computer Vision • Learning from large-scale human signals for robots and assistive cars • First author papers in NIPS, CVPR, ICCV, SIGKDD, ICRA, ISRR, and IJRR • PhD committee: Ashutosh Saxena, Bart Selman, Thorsten Joachims, Doug James, Robert Kleinberg. Indian Institute of Technology Delhi, India (2007-2012) B.Tech. Electrical Engineering & M.Tech. Information and Communication Technology • Advisor: Prof. Manik Varma and Prof. S.V.N. Vishwanathan • Thesis: Large-scale Algorithms for Multiple Kernel Learning Experience Stanford University, California, USA (2014-2016) Visiting Ph.D. Computer Science • Working with Prof. Ashutosh Saxena and Prof. Silvio Savarese • Leading Brain4Cars and Machine Learning lead on RoboBrain. Purdue University, Indiana, USA (Summer 2011) Research Intern, Statistics Department with Prof. S. V. N. Vishwanathan • Developed Multiple Kernel Learning algorithms to scale to Millions of kernels The Royal Bank of Scotland, India (Summer 2010) Software NeuralModels: A deep learning framework for quick prototyping of structures of Recurrent Neural Networks, Sensory-fusion architectures, and deep learning on graph structured data.
    [Show full text]
  • AAAI Organization
    AAAI Organization Officers AAAI President Eric Horvitz (Microsoft Corporation) AAAI President-Elect Martha E. Pollack (University of Michigan) Past President Alan Mackworth (University of British Columbia) Secretary-Treasurer Ted Senator (SAIC) Councilors (through 2008): Maria Gini (University of Minnesota) Kevin Knight (USC/Information Sciences Institute) Peter Stone (University of Texas at Austin) Sebastian Thrun (Stanford University) (through 2009): David Aha (U.S. Naval Research Laboratory) David Musliner (Honeywell Laboratories) Michael Pazzani (Rutgers University) Holly Yanco (University of Massachusetts Lowell) (through 2010): Sheila McIlraith (University of Toronto) Rich Sutton (University of Alberta) David E. Smith (NASA Ames Research Center) Cynthia Breazeal (Massachusetts Institute of Technology) Standing Committees Conference Chair Yolanda Gil (USC/Information Sciences Institute) Fellows and Nominating Chair Alan Mackworth (University of British Columbia) xxvi Finance Chair Ted Senator (SAIC) Conference Outreach Chair Peter Stone (University of Texas at Austin) Membership Chair Holly Yanco (University of Massachusetts Lowell) Publications Chair David Leake (Indiana University) Symposium Chair and Associate Chair Alan C. Schultz (Naval Research Laboratory) Marjorie Skubic (University of Missouri-Columbia) Symposium Associate Chair Holly Yanco (University of Massachusetts Lowell) AAAI Press Editor-in-Chief Anthony Cohn (University of Leeds) General Manager David M. Hamilton (The Live Oak Press, LLC) Press Editorial Board Aaron
    [Show full text]
  • Robobarista: Learning to Manipulate Novel Objects Via Deep Multimodal Embedding
    Robobarista: Learning to Manipulate Novel Objects via Deep Multimodal Embedding Jaeyong Sung, Seok Hyun Jin, Ian Lenz, and Ashutosh Saxena Abstract— There is a large variety of objects and appliances in human environments, such as stoves, coffee dispensers, juice extractors, and so on. It is challenging for a roboticist to program a robot for each of these object types and for each of their instantiations. In this work, we present a novel approach to manipulation planning based on the idea that many household objects share similarly-operated object parts. We formulate the manipulation planning as a structured prediction problem and learn to transfer manipulation strategy across different objects by embedding point-cloud, natural language, and manipulation trajectory data into a shared embedding space using a deep neural network. In order to learn semantically meaningful spaces throughout our network, we introduce a method for pre-training its lower layers for multimodal feature embedding and a method for fine-tuning this embedding space using a loss-based margin. In order to collect a large number of manipulation demonstrations for different objects, we develop a new crowd-sourcing platform called Robobarista. We test our model on our dataset consisting of 116 objects and appliances with 249 parts along with 250 language instructions, for which there are 1225 crowd-sourced manipulation demonstrations. We further show that our robot with our model can even prepare a cup of a latte with appliances it has never seen before.1 Fig. 1. First encounter of an espresso machine by our PR2 robot. Without ever having seen the machine before, given language instructions I.
    [Show full text]
  • Large-Scale Knowledge Base for Robots
    RoboBrain: Large-Scale Knowledge Engine for Robots Ashutosh Saxena, Ashesh Jain, Ozan Sener, Aditya Jami, Dipendra K Misra, Hema S Koppula Abstract In this paper we introduce a knowledge engine, which learns and shares knowledge representations, for robots to carry out a variety of tasks. Building such an engine brings with it the challenge of dealing with multiple data modalities including symbols, natural language, haptic senses, robot trajectories, visual features and many others. The knowledge stored in the engine comes from multiple sources including physical interactions that robots have while performing tasks (perception, planning and control), knowledge bases from the Internet and learned representations from several robotics research groups. We discuss various technical aspects and associated challenges such as modeling the correctness of knowledge, inferring latent information and formulating different robotic tasks as queries to the knowledge engine. We describe the system architecture and how it supports different mechanisms for users and robots to interact with the engine. Finally, we demonstrate its use in three important research areas: grounding natural language, perception, and planning, which are the key building blocks for many robotic tasks. This knowledge engine is a collaborative effort and we call it RoboBrain: http://www.robobrain.me Keywords—Cloud Robotics, Robot Learning, Systems, knowledge bases. 1 Introduction Over the last decade, we have seen many successful applications of large-scale knowledge systems. Examples include Google knowledge graph [13], IBM Wat- son [16], Wikipedia, and many others. These systems know answers to many of our day-to-day questions, and not crafted for a specific task, which makes them valuable for humans.
    [Show full text]
  • Avi Singh – Curriculum Vitae
    Avi Singh H +91-8853544535 B [email protected] Curriculum Vitae Í https://avisingh599.github.io Research Interests Computer Vision, Machine Learning, Robotics Education 2012–2016 Bachelor of Technology Indian Institute of Technology Kanpur GPA 9.4/10. Major: Electrical Engineering, Minor: Computer Science (Artifical Intelligence) Publications arXiv link Recurrent Neural Networks for Driver Activity Anticipation via Sensory- Fusion Architecture Ashesh Jain, Avi Singh, Hema Koppula, Shane Soh, Ashutosh Saxena. ICRA 2016 extended Brain4Cars: Sensory-Fusion Recurrent Neural Networks for Driver Activity abstract Anticipation Ashesh Jain, Shane Soh, Bharad Raghvan, Avi Singh, Hema Kopulla, Ashutosh Saxena. Full Oral at BayLearn 2015 Research Experience May-July 2015 Research Intern, Cornell University project page Brain4Cars: Anticipating Maneuvers via Learning Temporal Driving Models under Prof. Ashutosh Saxena, Department of Computer Science. Brain4Cars addresses the problem of anticipating driver maneuvers several seconds before they happen. It fuses the information from driver-facing and road-facing cameras with data from other sensors to make its predictions. These predictions can then be passed to driver assistance systems that can warn the driver if the maneuver is deemed to be dangerous. My contributions to the project are listed below: { The KLT face tracker in the project was replaced with a facial landmark localization pipeline based on Constrained Local Neural Fields. This provided robust tracking and allowed the computation of head pose, which then served as a strong feature for maneuver anticipation. { Implemented Gaussian Mixture Model-based initialization, and LBFGS optimization for training Autoregressive Input Output Hidden Markov Models (AIOHMM), which are a modification of HMMs and used for anticipation in Brain4Cars.
    [Show full text]
  • Dipendra Misra [email protected]
    Dipendra Misra [email protected] Address: Cornell Tech, 2 West Loop Road, New York, NY 10044, US Phone: (607)-793-7571 http://www.dipendramisra.com PhD Candidate, Cornell University (Computer Science). Research Interest I am interested in developing models and learning algorithms with emphasis on applications in natural language understanding. For example, building agents that can follow instructions, answer questions or hold a conversation. I am currently active in the following research areas: grounded natural language understanding, language and vision problems, semantic parsing, deep reinforcement learning, PAC reinforcement learning theory and model based reinforcement learning. Education • PhD (Computer Science), Cornell University (Aug 2013 - May 2019 Expected) Advisor: Yoav Artzi. Other committee members: Noah Snavely, Hadas Kress-Gazit. (GPA 4.0) • B-Tech (Computer Science and Engineering), IIT Kanpur (Aug 2009 - May 2013) Second Rank in Graduating Batch (CPI 9.8/10) Awards and Fellowships • Amazon AWS research grant (2014-2015). • Cornell University Fellowship (2013-2014). • Received OPJEMS Merit scholarship for 2011-12 and 2012-13 which is awarded to around 20 students in the same batch from all over India. • Academic Excellence Award for all years at IIT Kanpur. • Certificate of merit in National Standard Examination in Physics 2007. Publications • Towards a Simple Approach to Multi-step Model-based Reinforcement Learning. Kavosh Asadi, Evan Carter, Dipendra Misra, Michael Littman. (arXiv preprint), 2018. Preliminary version accepted at Deep RL workshop at NeurIPS 2018. • Touchdown: Natural Language Navigation and Spatial Reasoning in Visual Street Environments. Howard Chen, Alane Suhr, Dipendra Misra, Noah Snavely, Yoav Artzi. (arXiv preprint), 2018. Preliminary version accepted at VIGIL workshop at NeurIPS 2018.
    [Show full text]
  • CHAI 2020 Progress Report 9/30 2/38 Research Towards Solving the Problem of Control
    PROGRESS REPORT | 2020 Prof. Stuart J. Russell, CHAI Faculty Director and staff September 30, 2020 TABLE OF CONTENTS Research towards solving the problem of control 3 Organizational Chart 4 CHAI Research 5 Overview 5 Characteristics of the new model 5 Promoting the new model 6 Other research outputs 7 Specific outputs 7 CHAI alumni outputs 13 How CHAI contributes to student training in general 16 Other impacts on the AI research community 17 Other impacts 20 Contributions to Public Awareness of AI Existential Risk 20 Contributions to World Leaders’ Awareness 21 Connecting with China 22 Future plans 23 Basic AG theory 23 Theory of embedded agency 24 Cooperation with multiple AI systems 24 Making the new model practical 24 Social and human sciences: many humans, real humans 25 Training, field-building, policy, thought leadership 25 Appendix: Publications 26 CHAI 2020 Progress Report 9/30 2/38 Research towards solving the problem of control Founded in 2016, CHAI is a multi-site research center headquartered at UC Berkeley with branches at Michigan and Cornell. CHAI’s aim is to reorient AI research towards provably beneficial systems, over which humans can retain control even as they approach or exceed human-level decision-making capabilities. This document reports on CHAI’s activities and accomplishments in its first four years and its plans for the future. CHAI currently has 9 faculty investigators, 18 affiliate faculty, around 30 additional graduate and postdoctoral researchers (including roughly 25 PhD students), many undergraduate researchers and interns, and a staff of 5. CHAI’s primary support comes in the form of gifts from donor organizations and individuals.
    [Show full text]