HOUDINI: Escaping from Moderately Constrained Saddles

Authors: Dmitrii Avdiukhin, Grigory Yaroslavtsev

IJCAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We provide experimental results in the full version.
Researcher Affiliation Academia 1Indiana University, Bloomington, IN 2George Mason University, Fairfax, VA
Pseudocode Yes Algorithm 1: HOUDINIESCAPECORNER: Escaping from a corner for a quadratic function, Algorithm 2: FINDINSIDECORNER(x, A), Algorithm 3: HOUDINIESCAPE(x, S, δ): Escaping from a saddle point, Algorithm 4: FINDINSIDE(x, δ, (M , v ), (r , S ))
Open Source Code No The paper mentions providing "experimental results in the full version" but does not state that source code for the methodology is openly available or provide a link.
Open Datasets No The paper is theoretical and does not describe experiments with specific datasets for training.
Dataset Splits No The paper is theoretical and does not describe experiments or dataset splits for validation.
Hardware Specification No The paper is theoretical and does not describe any specific hardware used for experiments.
Software Dependencies No The paper is theoretical and does not mention any specific software dependencies or versions.
Experiment Setup No The paper is theoretical and does not describe specific experimental setup details, hyperparameters, or training configurations.