Optimization, Efficient Model Engineering,
and Distributed Systems Design

AI-OWLSAI-OWLS: Advancing Cutting-edge AI through Innovative Optimization, Efficient Model Engineering, and Distributed Systems Design

The goal of this cluster is to develop next-generation AI methodologies in optimization, graph problems, online learning, and deep neural networks and to establish Rice as a leader for practical, open-source AI algorithms, advancing state-of-the-art ML/AI across scientific fields.

Cluster Members

Student Collaborators

  • Afroditi Kolomvaki (Computer Science, Rice University)
  • Jasper Liao (Computer Science, Rice University)
  • Ria Stevens (Computer Science, Rice University)

Long-Term Goal

Our long-term vision is to establish Rice University as a premier academic hub for practical, state-of-the-art, open-source algorithms for machine learning (ML) and artificial intelligence (AI) by directly applying the team’s theories, practices, and methods with scientific applications in mind. This incentive is based on the fact that Rice Engineering School is a rare case where top-class academics from CS, ECE, Applied Math and Operations Research, and beyond interact actively within AI/ML, Optimization, Networks, and Decision Making. Yet, we often miss a shared vision and opportunity to jointly put "flesh" into attempts toward practical algorithms with crucial societal impact.

At the center of this vision lie Algorithms, Modeling, and Optimization, areas that heavily define the AI/ML time we live in but are also resilient to future engineering shifts, such as quantum computing. This core cluster includes expertise crucial in training deep neural networks, solving combinatorial problems, solving problems on graphs and using graphs, adapting to dynamic streaming data streams, solving large-scale problems in a distributed way, and more. Our interdisciplinary team combines expertise spanning convex/non-convex optimization, distributed algorithms, online learning, efficiency in training, and graph algorithms to develop next-generation recipes that can unlock the full potential of AI across application domains.

Given the opportunity this call gives us, we aim to generate AI-OWLS. This cutting-edge AI model suite integrates optimization and modeling techniques with modern AI settings and problems. Starting with small-to-medium ML models as our initial focus, we want to exploit the trend of what we can achieve using specialized solutions. The goal is to create a comprehensive family of highly optimized and efficiently scalable –open-source– models deployed on devices without needing the latest GPUs that span diverse domains like ML applications (computer vision, natural language processing), infrastructure, ecology, biology, wireless systems, graph analytics, and scientific computing.

Releasing these models as publicly available will democratize access to advanced AI capabilities and foster widespread adoption across industry, government, and academia (starting locally and slowly expanding on that). We will develop intuitive tooling and benchmarking frameworks to streamline model training, deployment, and evaluation. We aim to make Rice one of the academic open-source computational models capital, playing a pivotal role in shaping the next wave of AI innovation built on principled optimization foundations. This will cement Rice’s position at the forefront of machine learning research while benefiting society through accessible and scalable AI solutions.

Updates will be posted to the following web page, managed by the cluster: https://akyrillidis.github.io/aiowls/

VISIT


The Rice Ken Kennedy Institute is located on the campus of Rice University inside Duncan Hall. Click the map below for directions.


Rice Map

CONTACT


Rice University
Ken Kennedy Institute
6100 Main Street, MS-39
Houston, Texas 77005

CONNECT


Phone: 713-348-5823
Email: kenkennedy@rice.edu


Subscribe to our Newsletter.

Medium