8 problems
- 0 votes0 replies1 view
The principle of optimality for autonomous agents
The problem is to infer an autonomous agent's decision-making process from observations of its decisions. Principle of optimality. Autonomous agents generally make decisions in som…
- 0 votes0 replies0 views
Conjecture on structure in multiple-constraint revealed-preference sets
Let denote the set associated with the observations at time , and suppose it corresponds to a problem with multiple constraints. Multiple-constraint structure co…
- 0 votes0 replies0 views
Critical-region analysis for learning from optimization data
Consider learning from data generated by an optimization problem, with external parameters , and let the associated value function or optimizer function have an underly…
- 0 votes0 replies0 views
Critical-region explanation for difficulty in learning convex parametric optimization objectives
Let a convex parametric optimization problem have external parameters , and let its solution or value function be described by critical regions in the parameter space.…
- 0 votes0 replies0 views
Conjecture on elliptical Pareto critical sets for polynomial multiobjective optimization problems
Elliptical Pareto-critical-set conjecture. The ellipse parametrized by can be represented as the Pareto critical set of this poly…
- 0 votes0 replies0 views
The local-minimum explanation for the two-point model's PTV performance
The 2 Pts@. model combines the two predictions 2-D GANCER-sc. and 3-D GANCER-sc. in an ensemble inverse-optimization model, and PTV satisfaction measures performance on planning-ta…
- 0 votes0 replies1 view
The objective-magnitude explanation for absolute-duality-gap model performance
The forward problem uses objective functions whose magnitudes vary widely. Conjecture. The fact that the absolute duality gap model is not well-suited to this application is due to…
- 0 votes0 replies0 views
The imputed risk measure's protection-against-overfitting conjecture
Let be the true optimized certainty equivalent risk measure, let be the reference spectral risk measure, and let be the imputed risk…