Hardness of Learning Neural Networks with Natural Weights

Authors: Amit Daniely, Gal Vardi

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We prove negative results in this regard, and show that for depth-2 networks, and many natural" weights distributions such as the normal and the uniform distribution, most networks are hard to learn. Namely, there is no efficient learning algorithm that is provably successful for most weights, and every input distribution. It implies that there is no generic property that holds with high probability in such random networks and allows efficient learning.
Researcher Affiliation Collaboration Amit Daniely School of Computer Science and Engineering, The Hebrew University, Jerusalem, Israel and Google Research Tel-Aviv amit.daniely@mail.huji.ac.il Gal Vardi Weizmann Institute of Science gal.vardi@weizmann.ac.il
Pseudocode No The paper is theoretical, focusing on proofs and hardness results for learning neural networks. It does not include any pseudocode or algorithm blocks describing a computational procedure.
Open Source Code No The paper does not provide any statements about releasing source code for the described methodology, nor does it include links to a code repository.
Open Datasets No The paper is theoretical and does not involve empirical training of models on datasets. It discusses 'input distribution D' for theoretical analysis but does not refer to specific, publicly available datasets for training.
Dataset Splits No The paper is theoretical and does not describe empirical experiments. Therefore, it does not mention training, validation, or test dataset splits.
Hardware Specification No The paper is theoretical and does not involve running computational experiments that would require specific hardware. Therefore, no hardware specifications are mentioned.
Software Dependencies No The paper is theoretical and does not describe computational implementations or experiments. Consequently, it does not list any software dependencies with version numbers.
Experiment Setup No The paper is theoretical and focuses on mathematical proofs and hardness results rather than practical implementations or experiments. As such, it does not provide details on experimental setup, hyperparameters, or training configurations.