Foundational research in AI and optimization involves the interplay of finding the right representations, developing the right machine learning methods, and advancing optimization methods.

The design of chip systems has long inspired optimization innovations, from simulated annealing to randomized rounding and spectral embedding. Challenges include hierarchical-system context, extreme cost of training data, constrained optimization, and pervasive security aspects.

Collaborative and contextual robotics brings four core challenges to optimization: generation of safe motion in an environment, efficient interpretation of sensory data, planning of tasks across impossibly large task spaces, and adapting over time.

Network design must address decentralization of data acquisition, control, and decision over an inherently stochastic wireless medium (with physics of signal generation, propagation, and detection) at multiple time scales.