Mohamed R. Amer

I am currently a Founder and Chief Science Officer at Robust.AI. Prior to Robust.AI, I was a Senior Technical Manager at the Center for Vision Technologies, SRI International. I received $20M in DARPA funding for AI and ML projects including Machine Common Sense, Communicating with Computers, and Explainable Artificial Intelligence. I received my PhD from Oregon State University under the advise of Prof. Sinisa Todorovic. During my PhD, my research focus was on the application of generative hierarchical graphical models to vision applications such as monocular extraction of 2.1D sketch, multi-view 3D reconstruction and activity recognition in large scale videos.

Download Full C.V.LinkedInContact MeScholarBlog

Research Interests

The current focus of my research are deep generative models and graph networks applied to language and vision problems such as: generative ranking, visual storytelling, commonsense grounding and visual question answering. The main idea is to visualize textual information and use it to search, animate, or communicate.

Selected Publications

X. Lin, I. Sur, S. Nastase, A. Divakaran, U. Hasson, M. R. Amer. Data-Efficient Mutual Information Neural Estimator. arXiv, 2019. PDF

B. Knyazev, G. W. Taylor, M. R. Amer. Understanding Attention in Graph Neural Networks. International Conference on Learning Representations, 2019. Oral presentationPDF 

T. J. Meo, C. Kim, A. Raghavan, A. Tozzo, D. A. Salter, A. Tamrakar, M. R. Amer. Aesop: A Visual Storytelling Platform for Conversational AI and Common Sense Grounding. AI Communications, 2019 Pre-Print.
 
X. Lin, M. R. Amer. Human Motion Generation from Text using Dense Validation Adversarial Networks. arXiv, 2018. PDF
 
B. Knyazev, G. W. Taylor, X. Lin, M. R. Amer. Discovering and Fusing Relations in Molecules with Spectral Graph Networks. Neural Information Processing Systems Workshops, 2018. PDF 

T. Meo, A. Raghavan, D. Salter, A. Tozzo, A. Tamrakar, and M. R. Amer. Aesop: A Visual Storytelling Platform for Conversational AI, International Joint Conference on Artificial Intelligence, 2018. Best demo award. PDF 
 
X. Lin, C. Kim, T. Meo, and M. R. Amer. Learn, Generate, Rank: Generative Ranking of Motion Capture. European Conference on Computer Vision, 2018. PDF
  
D. Ramachandram, M Lisicki, TJ Shields, M.R. Amer, GW Taylor. Bayesian optimization on graph-structured search spaces: Optimizing deep multimodal fusion architectures. Neurocomputing, 2018. PDF

M. Ehrlich, T. J. Shields, T. Almaev, M. R. Amer. Facial Attributes Classification using Multi-Task Representation Learning. IEEE Conference on Computer Vision and Pattern Recognition Workshops, 2016. PDF

M. R. Amer and S. Todorovic. Sum Product Networks for Activity Recognition in Videos. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016. PDF

M. R. Amer, S. Yousefi, R. Raich, and S. Todorovic. Monocular Extraction of 2.1D Sketch using Constrained Convex Optimization. International Journal of Computer Vision, 2015. PDF

M. R. Amer, A. Fern, S. Todorovic, S. Zhu. Monte Carlo Tree Search for Scheduling Activity Recognition. International Conference on Computer Vision, 2013. PDF

Full Publication List ...

Workshops

- ACM-MM Workshop on Computational Models of Social Interactions and Behaviors: Human-Computer-Media Communication, 2015
- CVPR Workshop on Computational Models of Social Interactions and Behaviors: Grounding, Sensing, and Applications, 2014
- ICCV Workshop on Understanding Human Activities: Context and Interactions, 2013

Professional Service

- Publications Chair: CVPR'19 
- Program Comittee: CVPR’14-19, ECCV’16-18, ICCV’15-17, NIPS’14-18, ICML’15-19, ICLR’17-18, PAMI'15, IJCV'15-16

Awards

- 07/2018 IJCAI - Best Demo Award 
- 12/2017 SRI - Most Creative Talk 
- 06/2010 OSU - EECS Teaching Assistant of the year
- 08/2010 INRIA - Best Poster Award

Talks

- 12/2017 SRI - Generative Adversarial Networks for Vision
- 12/2016 SRI - Unified framework for machine learning
- 12/2015 SRI - Social interaction modeling using dynamic hybrid models
- 03/2014 Amazon - Hybrid models for event detection in time-series data at 
- 06/2013 SRI - Sum-Product networks for activity recognition in videos
- 11/2012 ECCV - Cost-Sensitive inference for multi-scale activity recognition