Monday, August 18, 2025

Previous | Next

• Creative Studio projects win international recognition

• Aseem Baranwal receives 2025 Cheriton Distinguished Dissertation Award

• From the ground up

• Monday's notes

Creative Studio projects win international recognition

By Christine Goucher.

Three Creative Studio projects have earned Awards of Excellence at the 31^st Annual Communicator Awards, an international competition that celebrates outstanding marketing and communication work across industries.

This is the third year in a row Creative Studio has been recognized, joining this year’s list of winners alongside Netflix, NASA+, Apple, Disney, Getty Museum, and NBCUniversal.

Award-winning projects

• Engineering Recruitment Posters – In partnership with the Engineering Undergraduate Office, a single poster request evolved into a student-focused suite of 16 designs plus collectible stickers after student feedback shaped the creative approach.
• Donor Support for Student Athletes – Leah Herrfort – Created with the Advancement team to support the digital Report on Giving, this video shares the impact of donor generosity through Leah’s voice and experience.
• Mission, Vision, Values – Waterloo at 100 – Sponsored by the Office of the President, this campus-wide campaign translated the University’s strategic direction into compelling visuals that connect people to our shared values.

These wins highlight the power of collaboration across campus to create purposeful, research-driven work that moves conversations forward.

Aseem Baranwal receives 2025 Cheriton Distinguished Dissertation Award

Aseem Baranwal with Professor Kimon Fountoulakis at spring 2025 convocation.

This article was originally published on the Cheriton School of Computer Science website.

PhD graduate Aseem Baranwal has received the 2025 Cheriton Distinguished Dissertation Award. Established in 2019, the annual award recognizes outstanding doctoral research at the Cheriton School of Computer Science and includes a $1,000 prize.

Aseem completed his doctoral studies co-supervised by Professor Kimon Fountoulakis from the Cheriton School of Computer Science and Professor Aukosh Jagannath from the Department of Statistics and Actuarial Science. His thesis, Statistical Foundations for Learning on Graphs, which he defended in October 2024, investigates the theoretical underpinnings of machine learning methods for graph-structured data.

“Congratulations to Aseem on receiving this year’s Cheriton Distinguished Dissertation Award,” said Professor Fountoulakis. “His thesis makes groundbreaking contributions to the field of graph neural networks and statistical learning on graphs, offering novel solutions with provable guarantees.”

“This is a well-deserved recognition,” Professor Jagannath added. “Aseem’s work introduces novel theoretical frameworks that deepen our understanding of graph neural network performance. His research offers fundamental insights into the optimality of message-passing architectures and is supported by rigorous empirical validation.”

Innovations and novel contributions

Aseem’s thesis explores how machine learning models can better interpret and classify data structured as a graph. His work focuses on graph neural networks — or GNNs — a class of models that integrate both node features and relational structures to make predictions. His research makes several novel contributions.

It develops a statistical framework to understand node classification in feature-rich relational data, offering a rigorous foundation for understanding the generalization performance and robustness to noise of GNNs compared with architectures that do not use relational information.

It introduces a notion of asymptotic local Bayes optimality for node classification, enabling the design of optimal GNN architectures for sparse relational data, a property that is often satisfied in practice.

It provides the first comprehensive analysis of the effects of graph convolutions, identifying fundamental classification thresholds and optimal placement of graph convolutions in multi-layer networks.

Finally, Aseem’s thesis is the first to theoretically analyze the performance of graph attention neural networks, which are among the most popular architectures used for machine learning on graphs. His research provides a mathematically precise characterization of graph attention mechanisms and their limitations in distinguishing intra- and inter-class edges, advancing understanding of GNN performance.

Impact on the research community

Aseem’s research addresses several key challenges in GNN research, including oversmoothing, generalization error, and message-passing optimality, offering novel solutions with provable guarantees. His findings have been presented at the International Conference on Learning Representations and the Conference on Neural Information Processing Systems, leading machine learning venues. His research has helped establish a new trend in understanding the performance of GNNs using statistical tools, work that has since been cited by leading researchers across multiple disciplines.

The code for all experiments is open-sourced and available on GitHub.

About the Cheriton Distinguished Dissertation Award

Aseem Baranwal is the eighth doctoral graduate to receive a Cheriton Distinguished Dissertation Award. Previous recipients are Amine Mhedhbi (2024), Michael Abebe (2023), Akshay Ramachandran (2022), Mike Schaekermann (2021, tie), Hong Zhou (2021, tie), Fiodar Kazhamiaka (2020), and Md Faizul Bari (2019).

From the ground up

By Sarah Fullerton. This article was originally published on Waterloo News.