Cheap Bandits
Authors: Manjesh Hanawal, Venkatesh Saligrama, Michal Valko, Remi Munos
ICML 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate and compare our algorithm with Spectral UCB which is shown to outperform its competitor Lin UCB for learning on graphs with large number of nodes. To demonstrate the potential of our algorithm in a more realistic scenario we also provide experiments on Forest Cover Type dataset. We set δ = 0.001, R = 0.01, and λ = 0.01. From Figure 1, we see that the cumulative regret performance of Cheap UCB is slightly worse than for Spectral UCB, but significantly better than for Lin UCB. However, in terms of the cost Cheap UCB provides a gain of at least 30 % as compared to both Spectral UCB and Lin UCB. |
| Researcher Affiliation | Collaboration | Manjesh Kumar Hanawal MHANAWAL@BU.EDU Department of ECE, Boston University, Boston, Massachusetts, 02215 USA Venkatesh Saligrama SRV@BU.EDU Department of ECE, Boston University, Boston, Massachusetts, 02215 USA Michal Valko MICHAL.VALKO@INRIA.FR INRIA Lille Nord Europe, Seque L team, 40 avenue Halley 59650, Villeneuve d Ascq, France R emi Munos REMI.MUNOS@INRIA.FR INRIA Lille Nord Europe, Seque L team, France and Google Deep Mind, United Kingdom |
| Pseudocode | Yes | Algorithm 1 Cheap UCB |
| Open Source Code | No | The paper does not provide any concrete links or statements about the public availability of its source code. |
| Open Datasets | Yes | Forest Cover Type data, a collection of 581021 labeled samples... This dataset was already used to evaluate a bandit setting by Filippi et al. (2010). |
| Dataset Splits | No | The paper describes a bandit problem setting where data is observed sequentially, and does not specify traditional train/validation/test dataset splits for reproduction. |
| Hardware Specification | No | The paper discusses computational issues and algorithms but does not specify any hardware details (e.g., CPU, GPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions matrix inversion using iterative update (Zhang, 2005) and fast symmetric diagonally dominant solvers as CMG (Koutis et al., 2011), but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | Yes | We set δ = 0.001, R = 0.01, and λ = 0.01. In the plots displayed we used N = 250, T = 150 and k = 5. We averaged the experiments over 100 runs. Algorithm 1 provides input parameters such as λ, δ, R, c. |