Online learning algorithms often have the issue of exhibiting poor performance during initial stages optimization procedure, which in practical applications might dissuade potential users from deploying such solutions. In this paper, we study a novel setting, namely conservative online convex optimization, are optimizing sequence loss functions under constraint that to perform at least as well ...

In this paper we propose an alternative solution to rl-blodr 1' problems. This altemativeis based upon the idea of transforming the I' problem into an equivalent (in the sense of having the same solution) mixed ll/'Hm problem that can be solved using convex optimieation techniques. The proposed algorithm has the advantage of generating, at each step, an upper bound of the cost that converges un...

Applications abound in which optimization problems must be repeatedly solved, each time with new (but similar) data. Analytic algorithms can hand-designed to provably solve these an iterative fashion. On one hand, data-driven "learn optimize" (L2O) much fewer iterations and similar cost per iteration as general-purpose algorithms. the other unfortunately, many L2O lack converge guarantees. To f...

A convex optimization model predicts an output from input by solving a problem. The class of models is large, and includes as special cases many well-known like linear logistic regression. We propose heuristic for learning the parameters in given dataset input-output pairs, using recently developed methods differentiating solution problem with respect to its parameters. describe three general c...

We optimize a general model of bioprocesses, which is nonconvex due to the microbial growth in biochemical reactors. formulate convex relaxation and give conditions guaranteeing its exactness both transient steady-state cases. When kinetics are modeled by Contois or, under constant biomass, Monod or Powell functions, second-order cone program, can be solved efficiently at large scales. implemen...

The network localization problem with convex and non-convex distance constraints may be modeled as a nonlinear optimization problem. The existing localization techniques are mainly based on convex optimization. In those techniques, the non-convex distance constraints are either ignored or relaxed into convex constraints for using the convex optimization methods like SDP, least square approximat...

introduction to nonlinear optimization theory algorithms introduction to nonlinear optimization theory algorithms introduction to nonlinear optimization theory algorithms introduction to nonlinear optimization theory algorithms chapter 16: introduction to nonlinear programming nonlinear programming: concepts, algorithms and applications theory, algorithms, and applications with matlab introduct...