Learning Robust Rule Representations for Abstract Reasoning via Internal Inferences

Authors: Wenbo Zhang, likai tang, Site Mo, Xianggen Liu, Sen Song

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this work, we propose a novel framework, ARII, that learns rule representations for Abstract Reasoning via Internal Inferences. ... We evaluate ARII on two benchmark datasets, including PGM and I-RAVEN. We observe that ARII achieves new state-of-the-art records on the majority of the reasoning tasks, including most of the generalization tests in PGM. Our codes are available at https://github.com/Zhangwenbo0324/ARII.
Researcher Affiliation Academia 1College of Computer Science, Sichuan University 2Department of Biomedical Engineering, Tsinghua University 3Laboratory of Brain and Intelligence, Tsinghua University 4College of Electrical Engineering, Sichuan University
Pseudocode No The paper does not contain any clearly labeled pseudocode or algorithm blocks.
Open Source Code Yes Our codes are available at https://github.com/Zhangwenbo0324/ARII.
Open Datasets Yes We evaluate ARII on two benchmark datasets, including PGM and I-RAVEN. ... The PGM [5] together with the RAVEN dataset [7] expands the size of RPM instances through automatic generation algorithms, serving as the benchmark datasets for deep learning network to study abstract reasoning. However, Hu et al. [8] found severe defects (a short cut for predictions) existing in the RAVEN dataset and created an unbiased version called I-RAVEN dataset to solve it. Therefore, we evaluate our method on the PGM and I-RAVEN rather than RAVEN.
Dataset Splits No The paper mentions training and testing sets for PGM ("neutral where the training and test set are sampled from the same distribution") but does not explicitly state the use of a validation set or specific percentages for any data splits in the provided text.
Hardware Specification No The paper does not explicitly describe the hardware used to run its experiments, such as specific GPU or CPU models.
Software Dependencies No The paper mentions using convolutional neural networks (CNNs) and neural network models but does not provide specific version numbers for any software dependencies, libraries, or programming languages used.
Experiment Setup No The paper does not explicitly provide specific hyperparameter values (e.g., learning rate, batch size, number of epochs) or detailed system-level training settings in the main text. It mentions lambda values for optimization but not their specific numerical values.