Unsupervised Abstract Reasoning for Raven’s Problem Matrices

This code is the implementation of our TIP paper.

This is the first unsupervised abstract reasoning method on Raven's Progressive Matrices, it is an extention of our arxiv preprint.

Comparision with some supervised methods.

Average testing accuracy on the RAVEN, I-RAVEN, and PGM dataset

Method	Raven	I-RAVEN	PGM
CNN	36.97	13.26	33.00
ResNet50	86.26	-	42.00
DCNet (ICLR2021)	93.58	49.36	68.57
NCD (Ours)	36.99	48.22	47.62

Generalization test results on PGM dataset

Method	neutral	interpolation	extrapolation
WReN (ICML2018)	62.6	64.4	17.2
DCNet (ICLR2021)	68.6	59.7	17.8
MXGNet (ICLR2020)	89.6	84.6	18.4
NCD (Ours)	47.6	47.0	24.9

Citation

If our code is useful for your research, please cite the following papers.

@article{zhuo2021unsup,
  title={Unsupervised Abstract Reasoning for Raven’s Problem Matrices},
  author={Tao Zhuo, Qiang Huang, and Mohan Kankanhalli},
  journal={IEEE Transactions on Image Processing},
  year={2021}
}

@article{zhuo2020solving,
  title={Solving Raven's Progressive Matrices with Neural Networks},
  author={Tao Zhuo and Mohan Kankanhalli},
  journal={arXiv preprint arXiv:2002.01646},
  year={2020}
}

@inproceedings{iclr2021,  
    author={Tao Zhuo and Mohan Kankanhalli},  
    title={Effective Abstract Reasoning with Dual-Contrast Network},  
    booktitle={International Conference on Learning Representations (ICLR)},      
    year={2021}
}

Unsupervised Abstract Reasoning for Raven’s Problem Matrices

Related tags

Overview

Unsupervised Abstract Reasoning for Raven’s Problem Matrices

Comparision with some supervised methods.

Average testing accuracy on the RAVEN, I-RAVEN, and PGM dataset

Generalization test results on PGM dataset

Citation

Owner

Tao Zhuo

jel - Japanese Entity Linker - is Bi-encoder based entity linker for japanese.

My Implementation for the paper EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Tasks using Tensorflow

A list of NLP(Natural Language Processing) tutorials

Official code for "Parser-Free Virtual Try-on via Distilling Appearance Flows", CVPR 2021

Simple Speech to Text, Text to Speech

A repository to run gpt-j-6b on low vram machines (4.2 gb minimum vram for 2000 token context, 3.5 gb for 1000 token context). Model loading takes 12gb free ram.

Grapheme-to-phoneme (G2P) conversion is the process of generating pronunciation for words based on their written form.

Code for EmBERT, a transformer model for embodied, language-guided visual task completion.

ttslearn: Library for Pythonで学ぶ音声合成 (Text-to-speech with Python)

Official source for spanish Language Models and resources made @ BSC-TEMU within the "Plan de las Tecnologías del Lenguaje" (Plan-TL).

Autoregressive Entity Retrieval

Natural Language Processing with transformers

This library is testing the ethics of language models by using natural adversarial texts.

Trained T5 and T5-large model for creating keywords from text

An evaluation toolkit for voice conversion models.

Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS)

Maix Speech AI lib, including ASR, chat, TTS etc.

CVSS: A Massively Multilingual Speech-to-Speech Translation Corpus

NeurIPS'21: Probabilistic Margins for Instance Reweighting in Adversarial Training (Pytorch implementation).

Ceaser-Cipher - The Caesar Cipher technique is one of the earliest and simplest method of encryption technique