Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

An SVD and Derivative Kernel Approach to Learning from Geometric Data

Authors: Eric Wong, J. Zico Kolter

AAAI 2015 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we evaluated the method in the context of molecular energy prediction, showing good performance for modeling previously unseen molecular configurations.
Researcher Affiliation Academia Eric Wong School of Computer Science Carnegie Mellon University EMAIL J. Zico Kolter School of Computer Science Carnegie Mellon University EMAIL
Pseudocode No The paper describes algorithms in text and through mathematical derivations but does not include structured pseudocode or algorithm blocks.
Open Source Code No The paper does not state that source code for the described methodology is publicly available.
Open Datasets No For both molecules, we generated 100 data points by adding noise to the original coordinates.
Dataset Splits Yes The kernel hyperparameters, namely exponential parameter γ and regularization parameter λ, were chosen by a grid-search with an inner 4-fold cross validation and optimized for the root mean squared error.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, memory) used for experiments are mentioned in the paper.
Software Dependencies No all computations were carried out using the GPAW numerical code (Mortensen, Hansen, and Jacobsen 2005), a gridbased implementation of the DFT calculator. We also use the Python Atomic Simulation Environment (ASE) (Bahn and Jacobsen 2002) to set up the computations and later to perform the molecular optimization.
Experiment Setup Yes The kernel hyperparameters, namely exponential parameter γ and regularization parameter λ, were chosen by a grid-search with an inner 4-fold cross validation and optimized for the root mean squared error. We introduce an additional regularization parameter λderiv to account for the difference in magnitude between the energies and derivatives.