Neural network gradient-based learning of black-box function interfaces

Authors: Alon Jacovi, Guy Hadash, Einat Kermany, Boaz Carmeli, Ofer Lavi, George Kour, Jonathan Berant

ICLR 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We present four experiments in increasing complexity to test the Estimate and Replace approach and compare its performance against existing solutions. Specifically, the experiments demonstrate that by leveraging external black-box functions, we achieve better generalization and better learning efficiency in comparison with existing competing solutions, without using intermediate labels.
Researcher Affiliation Collaboration Alon Jacovi1,2 , Guy Hadash1 , Einat Kermany1 , Boaz Carmeli1 , Ofer Lavi1, George Kour1, Jonathan Berant3,4 1 IBM Research, Israel 2 Bar Ilan University, Israel 3 Tel Aviv University, Israel 4 Allen Institute for Artificial Intelligence
Pseudocode No The paper describes the training procedures and architecture in prose and diagrams, but does not include any explicit pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statement about open-sourcing its code or provide a link to a code repository.
Open Datasets Yes The Image-Addition and Image-Lookup tasks use the MNIST training and test sets.
Dataset Splits No The paper mentions 'training and test sets' for MNIST and TLL, but does not explicitly specify a distinct validation dataset split or how data is partitioned for validation purposes to reproduce the experiments.
Hardware Specification No The paper states 'The implementation was done in Py Torch' but does not provide any specific hardware details such as GPU/CPU models or cloud instances used for the experiments.
Software Dependencies No The paper mentions using PyTorch, TensorFlow, A3C, and Adam optimization but does not provide specific version numbers for any of these software dependencies.
Experiment Setup Yes The hyper-parameters of the model are in Table 6.