About Me

Through hands on experience and advanced academic training, I have developed strong expertise in research, applied machine learning and statistics, and am passionate about using my skills to create data driven solutions to solve real world problems. I completed my MSc in Computer Science at Univeristy of Toronto supervised by Joseph Jay Williams. During my masters, I worked on issues pertaining to statistical inference from data collected by Multi-Armed Bandit algorithms in online educational experiments. In particular, I developed methods which balance statistical inference properties and reward maximization, and validated these methods in real world online education settings. I also worked on designing, deploying, and analyzing randomized experiments in online education. In the past, I have worked on applications of Deep Learning in fall detection as well as agitation detection at IATSL, under the supervision of Shehroz Khan and Alex Mihailidis.

Publications and Workshops

  • Algorithms for Adaptive Experiments that Trade-off Statistical Analysis with Reward: Combining Uniform Random Assignment and Reward Maximization, Jacob Nogas, , Tong Li, Fernando Yanez, Arghavan Modiri, Nina Deliu, Ben Prystawski, Sofia Villar, Audrey Durand, Anna Rafferty, Joseph J. Williams, Presented at Causal Inference Challenges in Sequential Decision Making: Bridging Theory and Practice - a NeurIPS 2021 Workshop

  • Using Adaptive Experiments to Rapidly Help Students , Zavaleta-Bernuy, A., Zheng, Q. Y., Shaikh, H., Nogas, J. , Rafferty, A., Petersen, A., & Williams, J. J., (2021, June). International Conference on Artificial Intelligence in Education (pp. 422-426). Springer, Cham.

  • A pilot study to detect agitation in people living with dementia using multi-modal sensors. S. Spasojevic, J. Nogas , A.Iaboni, B. Ye, A. Mihailidis, A. Wang, S. J. Li, L. S. Martin, K. Newman, S. S. Khan. Journal of Healthcare Informatics Research (published online, May 2021). DOI: 10.1007/s41666-021-00095-7.

  • Challenges in Statistical Analysis of Data Collected by a Bandit Algorithm: An Empirical Exploration in Applications to Adaptively Randomized Experiments , Joseph Jay Williams, Jacob Nogas , Nina Deliu, Hammad Shaikh, Sofia Villar, Audrey Durand, Anna Rafferty, arXiv preprint

  • Improving Short and Long-term Learning in an Online Homework System, Ben Prystawski, Jacob Nogas , Andrew Petersen, Joseph Jay Williams, Educational Data Mining in Computer Science Education (CSEDM) Workshop @ EDM, 2020

  • DeepFall - Non-invasive Fall Detection with Deep Spatio-Temporal Convolutional Autoencoders, Jacob Nogas, Alex Mihailidis, Shehroz S. Khan, Journal of Health Informatic Research, vol. 4, pages 50–70, 2020 [Code]

  • Spatio-Temporal Adversarial Learning for Detecting Unseen Falls, Shehroz S. Khan, Jacob Nogas, Alex Mihailidis, Pattern Analysis and Applications, 2020 (impact factor: 1.512)

  • Agitation Detection in People Living with Dementia using Multimodal Sensors, Shehroz S. Khan, Sofija Spasojevic, Jacob Nogas, Bing Ye, Alex Mihailidis, Andrea Iaboni, Angel Wang, Lori Schindel Martin and Kristine Newman, 41st Engineering in Medicine and Biology Conference, Berlin, 2019

  • Fall Detection from Thermal Camera Using Convolutional LSTM Autoencoder, Jacob Nogas, Shehroz S. Khan, Alex Mihailidis, 2nd Workshop on AI for Aging, Rehabilitation and Independent Assisted Living at IJCAI, Federated Artificial Intelligence Meeting, Sweden, 2018 [PDF].
  • Awards

  • Machine Learning Summit 2018 Vector Institute Best Poster Award: DeepFall -- Non-Invasive Fall Detection with Deep Spatio-Temporal Convolutional Autoencoders