Probing the Compositionality of Intuitive Functions
Authors: Eric Schulz, Josh Tenenbaum, David K. Duvenaud, Maarten Speekenbrink, Samuel J. Gershman
NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We show that (a) participants prefer compositional pattern extrapolations in both forced choice and manual drawing tasks; (b) samples elicited from participants priors over functions are more consistent with the compositional grammar; and (c) participants perceive compositional functions as more predictable than non-compositional ones. |
| Researcher Affiliation | Academia | Eric Schulz University College London e.schulz@cs.ucl.ac.uk Joshua B. Tenenbaum MIT jbt@mit.edu David Duvenaud University of Toronto duvenaud@cs.toronto.edu Maarten Speekenbrink University College London m.speekenbrink@ucl.ac.uk Samuel J. Gershman Harvard University gershman@fas.harvard.edu |
| Pseudocode | No | The paper does not contain any pseudocode or algorithm blocks. |
| Open Source Code | No | No statement regarding open-source code release for the methodology described in the paper is found. |
| Open Datasets | No | The paper mentions specific datasets like 'an airline passenger data set, volcano CO2 emission data, the number of gym memberships over 5 years, and the number of times people googled the band Wham! over the last 8 years' (Section 5.2). It also provides a link to 'http://learning.eng.cam.ac.uk/carl/mauna' for an example related to the periodic kernel. However, no specific access information (DOI, formal citation with author/year for public access, or direct repository link) is provided for the primary datasets used in the experiments. |
| Dataset Splits | No | The paper states 'the input space was split into training and test sets' in Experiment 2a and 'the outputs for xlearn = [0, 0.1, , 7] were used as a training set to which all three kernels were fitted and then used to generate predictions for the test set xtest = [7.1, 7.2, , 10]' in Experiment 1a. However, it does not provide specific percentages for these splits or explicit mention of a validation set. |
| Hardware Specification | No | The paper does not provide any specific hardware details (e.g., CPU, GPU models, memory, or cloud instance types) used for running the experiments. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers (e.g., programming languages, libraries, or frameworks with their versions). |
| Experiment Setup | Yes | In Experiment 3: Manual function completion, the paper states: 'functions were sampled from the compositional grammar, the number of points to be presented on each trial was sampled uniformly between 100 and 200, and the noise variance was sampled uniformly between 0 and 25. Finally, the size of an unobserved region of the function was sampled to lie between 5 and 50.' |