Learning under uncertainty: a comparison between R-W and Bayesian approach
Authors: He Huang, Martin Paulus
NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We applied the above models to two sets of data in a visual search task: (1) stable condition with no change points (from Yu & Huang, 2014) and (2) low volatility condition with change points based on N(30, 1).Subjects (N=207) were grouped into poor learners (optimal choice%< .5, n = 63), good learners (.5 optimal choice% 9/13, n = 108) and expert learners (optimal choice% > 9/13, n = 36) based on their performance (percentage of trials started from the most likely rewarded location). |
| Researcher Affiliation | Academia | He Huang Laureate Institute for Brain Research Tulsa, OK, 74133 crane081@gmail.com Martin Paulus Laureate Institute for Brain Research Tulsa, OK, 74133 mpaulus@laureateinstitute.org |
| Pseudocode | No | The paper provides mathematical equations for the models (e.g., Equation 1 and 2) but does not include any pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain any statements about releasing code, providing a repository link, or making the source code available for the described methodology. |
| Open Datasets | Yes | We applied the above models to two sets of data in a visual search task: (1) stable condition with no change points (from Yu & Huang, 2014) and (2) low volatility condition with change points based on N(30, 1).We thank Angela Yu for sharing the data in Yu et al. 2014, and for allowing us to use it in this paper. |
| Dataset Splits | No | The paper mentions subjects grouped into performance categories ('poor learners', 'good learners', 'expert learners') for analysis, but does not specify training, validation, or test dataset splits for model reproduction. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., GPU/CPU models, memory) used for simulations or data analysis. |
| Software Dependencies | No | The paper does not mention any specific software dependencies or library versions used for the implementation or analysis. |
| Experiment Setup | Yes | For each condition, we simulated 100 runs (90 trial per run) of agents choices with η ranges from 0 to 1 with an increment of 0.1 and fixed β = 20.For each simulated behavior type, R-W model was fitted using Maximum Likelihood Estimation with η ranges from 0 to 1 with an increment of .025 and β = 20.For each simulated condition (stable, low and high volatility), we simulated 100 runs (90 trials per run) for agents choices with α ranges from 0 to 1 and fixed β = 20. |