Mark Crowley | Lab News

The year in review

see news for March 2022 and September 2022 for updates from earlier on in the year!

Attending the Canadian Wildfire Conference

Just in time for Hallowe’en this year I visited Edmonton, Alberta along with my MASc student Xiang Fang to attend the Wildland Fire Canada Conference (WFCC) to meet with researchers and practitioners dealing with Forest Wildfires in Canada and around the world. Part of our research uses wildfire as a challenging test and application domain for prediction, planning, risk assessment and visualization. This work is funded through Canada Wildfire, a nationally funded strategic network, aiming to connect researchers from planning, computer science, engineerings with more traditional forestry connected fields. We had a lot of good leads on datatsets and problems right from the source.

AI Seminar : “A Dream of Cause and Fire”

While I was at Edmonton, I was also invited to visit to friends and colleagues at the Alberta Machine Intelligence Institute (AMII)

My talk was some high-level musings about things I’ve been thinking about recently like causality, Reinforcement Learning, and since it was after the conference, some ideas about how Machine Learning research can help and be helped by Forest Wildfire as well. It will be up on 2022 AMII AI Seminar page at some point.

Mark Crowley. “A Dream of Cause and Fire: Musings on the Current and Future Uses of AI for Science”. Alberta Machine Intelligence Institute (AMII). Host: Mathew Taylor, UAlberta. Nov 4, 2022.

Publications

In the last part of the year we also had a good set of

Using Affect as a Communication Modality to Improve Human-Robot Communication in Robot-Assisted Search and Rescue Scenarios

Akgun, S. A., Ghafurian, M., Crowley, M., & Dautenhahn, K. (2022). Using Affect as a Communication Modality to Improve Human-Robot Communication in Robot-Assisted Search and Rescue Scenarios. IEEE Transactions on Affective Computing, 18. https://doi.org/https://doi.org/10.1109/TAFFC.2022.3221922

Through a collaboration with Prof. Kerstin Dautenhahn this exciting paper is on it’s way to being published showcasing useful evaluations of the usefulness of emotional signalling to improve communication speed and clarify for earch And Rescue (SAR) robots in noisy environments.

Predicting Aggressive Human Driver Behaviour

Camlica, Z., Quesenberry, J., Carballo, D., & Crowley, M. (2022). Aggressive Driver Behavior Detection using Parallel Convolutional Neural Networks on Simulated and Real Driving Data. 9th International Conference on Internet of Things: Systems, Management and Security (IOTSMS), 8. https://youtu.be/ORSPRSEJ5NQ

In Nov 2022 we had a workshop paper (Camlica et al., 2022) accepted on this topic as part of an international IOT conference. It presented new results from my PhD student Zehra Camlica on her work modelling aggressive/passive human driver behaviour styles using deep learning. We used simulated and real collected driver user data from the DBL project to validate the models. I presented it virtually in Dec 2022.

Manifold Learning Book finally released

Ghojogh, B., Crowley, M., Karray, F., & Ghodsi, A. (2023). Elements of Dimensionality Reduction and Manifold Learning (p. 363). Springer Nature. https://link.springer.com/book/10.1007/978-3-031-10602-6

Our long-awaited, much discussed, by us at least, textbook on Manifold Learning and Dimensionality Reduction is now at the final stage before being published via Springer Nature. The book is in the topic area of my former PhD student Benyamin Ghojogh’s thesis. Benyamin led the massive effort involved in the completion of this textbook. We’re confident it will be a fantastic resource for students and researchers wanting a wide view of this field but also one built from the ground up from the fundamentals at each step.

Another year, after all

So with all that, another year has ended. May the next one bring us all to greater joy and knowledge.

The novel method proposed in this paper is com- promised of application of two Convolutional Neural Networks (CNN) working in parallel to simultaneously classify driver be- haviors while classifying maneuvers by using time series data. We claim that the Parallel Convolutional Neural Network (PCNN) not only speeds-up training time but also increases performance since having information about the maneuver helps to improve behavior classification performance and vice versa. In this study, both simulation and real-world driving datasets are utilized for driver behavior analysis. As simulation data, mobile phone sensor data are simulated as a time series using a combination of a traffic simulator (SUMO) and a car simulation system (Webots). The same type of data is collected with a specially designed vehicle traveled on a defined route around a predefined region. The collected data are then separately utilized as training and testing data for classification of both maneuvers (e.g turns and lane changes) and driver behaviors (e.g aggressive, non-aggressive) applying a novel method using deep learning on time series data. In addition, other methods which are commonly used for time series analysis, Hidden Markov Models(HMMs) and Recurrent Neural Networks (RNN), are applied to the same datasets to compare with PCNN. According to the results, the CNN classifiers perform efficiently for a single task and PCNN outperforms both single task-CNN and RNN with an average accuracy of 86%.

Lab News - Fall 2022