understanding-ai icon indicating copy to clipboard operation
understanding-ai copied to clipboard

personal repository

Results 32 understanding-ai issues
Sort by recently updated
recently updated
newest added

get doc -> ocr -> summary each page -> final summary from all pages

# Summary All-Seeing 1B (AS-1B) dataset: we propose a new large-scale dataset (AS-1B) for open-world panoptic visual recognition and understanding, using an economical semi-automatic data engine that combines the power...

Repository for sarcasm. Not a real deep vector DB (just a python code). Related Post: - https://www.linkedin.com/posts/dhruv-anand-ainorthstartech_ainorthstartech-vectorsearch-vectordatabases-activity-7053961131685490688-_UFB?utm_source=share&utm_medium=member_desktop

Summary - uses blip/blip2 to generate a simple caption - uses grit/detectron2 to generate a dense caption - uses segment anything to generate a region_semantic information - unify all above...

https://arxiv.org/abs/1807.03748 ~big fan of Aaron van den Oord~ # Abstract - propose universal unsupervised learning approach to extract useful representations from high-dimensional data, **CPC(Contrastive Predictive Coding)** - use probabilistic contrastive...

Unsupervised Learning
Representation Learning
Autoregressive

https://arxiv.org/abs/1901.03909 This follows [ Adding One Neuron Can Eliminate All Bad Local Minima](https://arxiv.org/abs/1805.08671) and [Elimination of All Bad Local Minima in Deep Learning](https://arxiv.org/abs/1901.00279) # Key point ![image](https://user-images.githubusercontent.com/2807595/51187986-01c9fe00-18ab-11e9-8d0e-c4cf10132666.png) ![image](https://user-images.githubusercontent.com/2807595/51188023-14dcce00-18ab-11e9-8b11-5b0def1dae37.png) ***a*** and...

Optimization
Local Minima Optimization

https://arxiv.org/pdf/1704.04861.pdf # Summary ![image](https://user-images.githubusercontent.com/2807595/50852921-e2811d00-134e-11e9-93bd-6c33b986e66f.png) ![image](https://user-images.githubusercontent.com/2807595/50852931-ead95800-134e-11e9-8dc8-c41784752a97.png) ![image](https://user-images.githubusercontent.com/2807595/50853151-79e67000-134f-11e9-9cc6-552ea2ce592d.png) ![image](https://user-images.githubusercontent.com/2807595/50853176-8a96e600-134f-11e9-8305-a98e47b5b468.png)

Convolution
Optimization

http://arxiv.org/abs/1812.10464 # Notes - Model learn joint multilingual sentence representations ![image](https://user-images.githubusercontent.com/2807595/50990459-0fb10500-14e0-11e9-8385-567c21da6418.png) ## Pre-training techniques using Moses tools - punctuation normalization - non printing chars removal - Tokenization ## Paper summary...

Zero-Shot
NMT

https://arxiv.org/pdf/1901.00884v1.pdf ![image](https://user-images.githubusercontent.com/2807595/50796418-75627e80-129f-11e9-9a98-74e19603f349.png) ![image](https://user-images.githubusercontent.com/2807595/50796424-7b585f80-129f-11e9-8026-4479b8a904e3.png) # 4. Conclusion 1. It is easy to construct examples where the intermediate representations have no match, even when the outputs of the network are identical. 2....

Representation Learning

https://arxiv.org/abs/1711.00937 # Abstract - paper proposes model(**VQ-VAE**) that learns "discrete representations" - differs from VAEs - encode network outputs discrete (means not continuous) - prior learnt than static - circumvent...

VAE
Representation Learning