The 2023 INFORMS Annual Meeting took place in Phoenix, Arizona from October 15 to 18. It showcased a packed program, with a strong emphasis on the crossing paths of artificial intelligence (AI) and operations research (OR). Among the various sessions and presentations, members of the NSF AI Institute for Advances in Optimization (AI4OPT) made a significant impact throughout the event, underlining the commitment of AI-driven optimization across a range of real-world fields.

Optimizing Truck Fleet Scheduling for Fuel Deliveries

One of the key presentations at the conference was titled "Optimizing Truck Fleet Scheduling for Fuel Deliveries," which discussed a novel approach to optimize the delivery schedules for a fleet of trucks with specified availability time windows for fuel replenishment at various sites. The authors, Vahid Eghbal Akhlaghi, Hongzhao Guan, Jason Lu, and Pascal Van Hentenryck, proposed a mixed-integer programming model that combines a two-stage optimization algorithm. The first stage ensures sufficient deliveries to maintain site operations, while the second stage focuses on enhancing driver shift and truck utilization while minimizing total traveled distance.

The results of their research, based on a case study in Estonia, showed significant benefits, including cost savings, improved customer satisfaction, enhanced business resilience, and reduced fuel consumption and emissions. This approach underscores the potential for automation in fleet scheduling and the advantages of AI-driven optimization in the logistics and transportation sector.

Equitable Infrastructure Investment Decisions Amid High Wildfire Risk

In the session titled "Optimizing Equitable Infrastructure Investment Decisions Under High Wildfire Risk Scenarios," authors Swati Gupta, Alyssa Kody, Daniel K. Molzahn, Ryan Piansky, and Madeleine Pollack will present their research findings and insights. This session, scheduled from 4 PM to 5:15 PM at CC-North 127B, promises to shed light on the critical intersection of infrastructure investment and wildfire risk management, offering valuable perspectives for equitable decision-making in this challenging scenario.

Best Paper on Data Mining and Decision Analytics

Jie Wang, a 4th-year Ph.D. student in the Industrial Engineering program at Georgia Tech, supervised by Prof. Prof. Yao Xie wins INFORMS Workshop on Data Mining and Decision Analytics Best Paper Award Theoretical Track for paper titled "Entropic Regularization for Adversarial Robust Learning." The full list of authors: Jie Wang, Yifan Lin, Song Wei, Rui Gao and Yao Xie.

Additionally, Wang was recognized as a finalist in the "2023 INFORMS BSS Data Challenge in INFORMS Workshop on Data Mining and Decision Analytics" for report on "Reliable Offline Pricing in eCommerce Decision-Making: A Distributionally Robust Viewpoint."

AI4OPT Members Win INFORMS Optimization Society Prizes

The AI4OPT community celebrated two of its members, Renato Monteiro and Alper Atamturk, who were recognized with prestigious INFORMS Optimization Society Prizes. Renato Monteiro was awarded the Khachiyan Prize for his work in the field of continuous optimization. His presentation, "A Journey Through Many Exciting Topics of Continuous Optimization," highlighted his contributions to the field and their potential impact.

Alper Atamturk received the Farkas Prize for his work in strengthening mixed-integer programming (MIP) formulations of hybrid model predictive control. His presentation, "Strengthening MIP Formulations of Hybrid Model Predictive Control," showcased his innovative approaches to enhancing the accuracy and efficiency of MIP-based optimization in control systems.

AI4OPT Leader Receives Book Recognition With Award

George Lan received the the 2023 INFORMS Lanchester prize for his book titled ”First-order and Stochastic Optimization for Machine Learning,” published three years ago.

Panel Discussion: OR/AI - Integrating Operations Research with Artificial Intelligence

The INFORMS Annual Meeting also featured a panel session titled "OR/AI: Integrating Operations Research with Artificial Intelligence." The panel, moderated by Ahmed Abbasi from the University of Notre Dame, brought together experts in the field, including Ramayya Krishnan, Bistra Dilkina, Nan Zhang, and Cole Smith. They discussed research and funding opportunities at the intersection of operations research and artificial intelligence, highlighting the potential for synergy between these two disciplines in solving complex real-world problems.

K-12 Education Outreach in a Post-COVID World

Another panel session, "K-12 Education Outreach in a Post-COVID World," addressed the challenges and opportunities in K-12 education outreach, especially in the context of the COVID-19 pandemic. The panel, chaired by Zihan Zhang from Georgia Tech, featured experts like Afrooz Jalilzadeh, Pascal Van Hentenryck, Mary Ogidigben, Paul Hand, and Neil Desnoyers. They discussed the profound impact of the pandemic on K-12 education and the need for innovative outreach programs to adapt to changing circumstances while ensuring quality education for students.

Advancements in Machine Learning and Optimization

The AI4OPT community also presented research on the application of machine learning and optimization in various domains. Presentations included "Bucketized Active Sampling for Acopf Optimization Proxies," "End-To-End Feasible Optimization Proxies for Large-Scale Economic Dispatch," "Multiperiod Stochastic Security-Constrained Economic Dispatch," and "Online Risk Assessment with Optimization Proxies." These presentations highlighted how AI-driven optimization techniques are transforming industries such as power systems, logistics, and transportation.

On-Demand Multimodal Transit Systems

The conference addressed the challenges and solutions in transit systems, specifically focusing on "Bus Line Design for On-Demand Multimodal Transit Systems." This presentation discussed a novel mixed-integer program (MIP) formulation to incorporate bus line design into the network design problem for On-Demand Multimodal Transit Systems, improving service quality, accessibility, and coverage within budget constraints.

AI in Energy Markets and Supply Chains

Energy markets and supply chains were also a focus of the conference. A presentation titled "Value-Oriented Loss Function Tuning for Timeseries Forecasting in Energy Markets" discussed techniques for training time series forecasters in complex, uncertain energy markets. Another session, "Plenary Panel: Harnessing the Data Revolution in Supply Chains," explored the potential of AI and data analytics to optimize supply chain operations.

Innovations in Combinatorial Optimization

Combinatorial optimization was a significant theme at the conference. A presentation on "Optimization Proxies for Continuous and Discrete Optimization" showcased the usefulness of optimization proxies in real-time applications. Another presentation, "Portfolio Approximations and Fairness in Combinatorial Optimization," explored the application of fairness notions in combinatorial optimization, focusing on various discrete optimization problems.

Stochastic Optimization and Machine Learning

Stochastic optimization and machine learning techniques were highlighted in the presentation "Stochastic First-Order Algorithms for Constrained Distributionally Robust Optimization." This session discussed the use of stochastic first-order methods to solve distributionally robust optimization problems with multiple expectation constraints.

Semiconductor Manufacturing and Non-Smooth Optimization

The conference also addressed optimization in semiconductor manufacturing, as well as the challenges of non-smooth optimization problems. These presentations underscored the applicability of AI-driven optimization in diverse fields.

Keynote: Network Flows and Minimum Cuts

A keynote presentation by Dorit Simona Hochbaum on "Network Flows and Minimum Cuts." This keynote highlighted the importance of network flows and minimum cuts in a wide range of optimization problems, from ranking and clustering to machine learning and imaging.

About INFORMS

With over 12,500 members from around the globe, The Institute for Operations Research and the Management Sciences (INFORMS) is the leading international association for professionals in operations research, analytics, management science, economics, behavioral science, statistics, artificial intelligence, data science, applied mathematics, and other relevant fields.

5521 Research Park Drive, Suite 200

Catonsville, MD 21228 USA

Phone 1 443-757-3500

Phone 2 800-4INFORMS (800-446-3676)

Fax 443-757-3515

Email informs@informs.org