John Gideon

About Me

John Gideon is a doctoral candidate in Computer Science and Engineering at the University of Michigan. He received his B.S in Electrical Engineering (Magna Cum Laude) and M.S. in Computer Engineering from the University of Cincinnati in 2012 and 2013, respectively. His undergraduate and Master's had a strong emphasis on signal processing, embedded systems, and parallel processing. He has also had professional experience working for Toyota Research Institue (TRI), General Electric Aviation, and the University of Cincinnati Simulation Center contracting with Proctor and Gamble.

John is currently a Graduate Student Researcher working with Professor Emily Mower Provost in the Computational Human-Centered Analysis and Integration (CHAI) Lab. His current research as part of the PRIORI project, aims to use mobile phone calls from individuals with bipolar disorder to determine when they are symptomatic. In particular, he is interested in discovering methods of improving emotion and mood classification by reducing the effect of other acoustic factors. Some of these include device characteristics, dataset differences, and subject demographics. Techniques used include deep learning, clinician directed feature creation, and audio signal processing. His research goals are aimed at the improvement of medical care and everday tasks with assisstive technologies. In his free time, John enjoys the outdoors (seen above at Glacier National Park), board games, singing, and travel.

Research

Broadly speaking, I am interested in the improvement of assisstive technologies for medical care and everyday tasks. In particular, mental health is one area that could be greatly improved through the understanding of subtle changes in emotion and mood. Currently I am exploring how this can be accomplished with machine learning using everyday cues such as speech, body language, and physical and social activity. My background in systems development and parallel processing has enabled me to not only form novel research questions but also effectively carry them out. This fusion of low level computer understanding with a focus on human-computer interaction has allowed me to work on many interesting projects. Below are highlights of few projects on which I've worked:

Computational Human-Centered Analysis and Integration Lab

University of MichiganFall 2013 - Present
  • Investigating the automatic detection of emotion and mood from speech for the improvement of medical care with Professor Emily Mower Provost.
  • Researcher on the PRIORI project, which aims to use mobile phone calls from individuals with bipolar disorder to determine when they are symptomatic.
  • Exploring methods to improve emotion and mood classification by reducing the contribution of other acoustic factors including device characteristics, dataset differences, and subject demographics.
  • Employing a variety of techniques including deep learning, clinically directed feature creation, and signal processing to create working systems.

Virtualized Beowulf Clusters with Low Latency Messaging

University of CincinnatiFall 2012 - Summer 2013
  • Showed a custom virtualized operating system with much lower native network latency.
  • Designed and implemented llamaMPI – a combination of the Message Passing Interface (MPI) and llamaOS, a custom low-latency operating system.
  • This allowed for integration of already existing applications with a speedup of up to 70%. The final result was tested with the NAS Parallel Benchmarks and the WARPED parallel discrete event simulator.

Environment Tracking and Augmentation Senior Project

University of CincinnatiFall 2011 - Spring 2012
  • Constructed a model of various housing interior rooms using an online custom tracking algorithm employing the Kinect device.
  • The models were then augmented in real time on the normal video display with various informative computer graphics.

Publications

  • John Gideon, Melvin McInnis, and Emily Mower Provost. “Barking up the Right Tree: Improving Cross-Corpus Speech Emotion Recognition with Adversarial Discriminative Domain Generalization (ADDoG).” IEEE Transactions on Affective Computing. 2018. (under review) - View
  • Soheil Khorram, Mimansa Jaiswal, John Gideon, Melvin McInnis, Emily Mower Provost. “The PRIORI Emotion Dataset: Linking Mood to Emotion Detected In-the-Wild.” INTERSPEECH. 2018. (oral presentation) - View
  • John Gideon, Simon Stent, Luke Fletcher. "A Multi-Camera Deep Neural Network for Detecting Elevated Alertness in Drivers." ICASSP. 2018. (poster presentation) - View - Project Page
  • John Gideon, Soheil Khorram, Zakaria Aldeneh, Dimitrios Dimitriadis, and Emily Mower Provost. "Progressive Neural Networks for Transfer Learning in Emotion Recognition." INTERSPEECH. 2017. (oral presentation) - View
  • John Gideon, Biqiao Zhang, Zakaria Aldeneh, Yelin Kim, Soheil Khorram, Duc Le, and Emily Mower Provost. "Wild Wild Emotion: A Multimodal Ensemble Approach." ICMI. EmotiW Challenge. 2016. (oral presentation and poster) - View
  • Soheil Khorram, John Gideon, Melvin McInnis, and Emily Mower Provost. "Recognition of Depression in Bipolar Disorder: Leveraging Cohort and Person-Specific Knowledge." INTERSPEECH. 2016. (oral presentation) - View
  • John Gideon, Emily Mower Provost, and Melvin McInnis. "Mood State Prediction from Speech of Varying Acoustic Quality for Individuals with Bipolar Disorder." ICASSP. Shanghai, China. March, 2016. (oral presentation) - View
  • John Gideon. “The Integration of Llamaos for Fine-Grained Parallel Simulation.” University of Cincinnati. OhioLINK Electronic Theses and Dissertations Center. 2013. (thesis) - View

Experience

I have worked in a variety of fields ranging from engineering research, student instruction, and cutting edge embedded system design.

Intern Research Scientist

Toyota Research Institue6/17 – 9/17
  • Investigated the automatic detection of surprise in drivers.
  • Collected a dataset of people reacting to dashcam footage of crashes.
  • Sensors included three cameras, a microphone, gaze tracking glasses, and steering wheel and pedal input.
  • Developed a novel multi-view neural network for surprise detection.

Graduate Student Instructor

University of Michigan9/13 – 4/14
  • Supported the teaching of the introductory algorithms and data structures course with mainly sophomore level students.
  • Led discussion and lab section, for which he prepared his own presentations. Mentored students during office hours, allowing for more directed advice.

Engineering Researcher

University of Cincinnati Simulation Center4/13 – 8/13
  • Performed research and improved key software as a contractor to Procter and Gamble.
  • Revamped existing R scripts to run simulations orders of magnitude faster.
  • Completed extensive analysis of a new image processing algorithm in MATLAB.

Engineering Co-op

General Electric Aviation1/10 – 12/12
  • Conducted research and development on a new embedded testing system.
  • Utilized state of the art multicore digital signal processing and FPGA technologies.
  • Used formal development processes including coding standards and version control.
  • Reduced testing costs and increased data accuracy versus previous system.

Technical Skills

  • Deep learning using Pytorch and Keras
  • Signal processing and acoustical analysis
  • Most familiar with C++, Python, and MATLAB
  • Some experience with Perl, Bash, Java, C, and C#
  • Dabbled in Verilog, VHDL, R, and Ada
  • Version control systems such as Git and Subversion

Curriculum Vitae

Contact Me

Please feel free to drop me a line if you are interested in talking.