Cautious Regret Minimization: Online Optimization with Long-Term Budget Constraints
Authors: Nikolaos Liakopoulos, Apostolos Destounis, Georgios Paschos, Thrasyvoulos Spyropoulos, Panayotis Mertikopoulos
ICML 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our theoretical findings are also validated by a series of numerical experiments which suggest that increasing K that is, enlarging the window over which the budget must be balanced makes the K-benchmark guarantee tighter. and 5. Numerical results |
| Researcher Affiliation | Collaboration | 1Paris Research Center, Huawei Technologies, Paris, France 2EURECOM, Sophia-Antipolis, France 3Univ. Grenoble Alpes, CNRS, Inria, Grenoble INP, LIG, Grenoble, France. |
| Pseudocode | Yes | Cautious Online Lagrangian Descent (COLD) For t = 1, . . . , T: xt 1 V f t 1(xt 1) + Q(t)g t 1(xt 1) 2 Q(t + 1) = [Q(t) + ̂gt(xt)]+. |
| Open Source Code | No | The paper does not provide an explicit statement about open-sourcing the code for the described methodology, nor does it include any links to a code repository. |
| Open Datasets | No | We simulate a scenario with one website, where xt [0, ), ft(xt) = wtxt, and gt(xt) = ptxt b T /T, where wt, pt are generated by exponential distributions wt Exp(11) and pt Exp(10). This is a simulated environment, not a named public dataset with access info. |
| Dataset Splits | No | The paper describes a simulated scenario where data is generated. It does not mention explicit train/validation/test dataset splits. |
| Hardware Specification | No | The paper does not specify any particular hardware (e.g., CPU, GPU models) used for running the experiments or simulations. |
| Software Dependencies | No | The paper does not specify any software dependencies with version numbers used for the experiments. |
| Experiment Setup | Yes | We run the experiment for different horizons T = {2000, 4000, . . . , 10000}, budget b T = 300T, and parameters set to = max{T, V } and V = T 0.99 for each of the experiments. and we choose K = T 3/4 and V = {T 1/2, T 3/4, T 0.99, T 5/4} |