Ciro Santilli @cirosantilli 35

TODO where to find it: www.kaggle.com/general/50987

Cited on original Generative adversarial network paper: proceedings.neurips.cc/paper_files/paper/2014/file/5ca3e9b122f61f8f06494c97b1afccf3-Paper.pdf

14 million images, more than 20k categories, typically denoting prominent objects in the image, either common daily objects, or a wild range of animals. About 1 million of them also have bounding boxes for the objects.

Each image appears to have a single label associated to it. Care must have been taken somehow with categories, since some images contain severl possible objects, e.g. a person and some object.

In practice however, the ILSVRC subset is more commonly used.

Official project page: www.image-net.org/

The data license is restrictive and forbids commercial usage: www.image-net.org/download.php.

The categories are all part of WordNet, which means that there are several parent/child categories such as dog vs type of dog available. ImageNet1k only appears to have leaf nodes however (i.e. no "dog" label, just specific types of dog).

From cocodataset.org/:

So they have relatively few object labels, but their focus seems to be putting a bunch of objects on the same image. E.g. they have 13 cat plus pizza photos. Searching for such weird combinations is kind of fun.

Their official dataset explorer is actually good: cocodataset.org/#explore

And the objects don't just have bounding boxes, but detailed polygons.

Also, images have captions describing the relation between objects:Epic.

This dataset is kind of cool.

Original 2014 paper by Microsoft: arxiv.org/abs/1405.0312

TODO vs COCO dataset.

As of v7:

The images and annotations are both under CC BY, with Google as the copyright holder.

atlas.brain-map.org/ omg some amazing things there.

A Drosophila melanogaster has about 135k neurons, and we only managed to reconstruct its connectome in 2023.

The human brain has 86 billion neurons, about 1 million times more. Therefore, it is obvious that we are very very far away from a full connectome.

Instead however, we could look at larger scales of connectome, and then try from that to extract modules, and then reverse engineer things module by module.

This is likely how we are going to "understand how the human brain works".

Some notable connectomes:

This is the most plausible way of obtaining a full connectome looking from 2020 forward. Then you'd observe the slices with an electron microscope + appropriate Staining. Superintelligence by Nick Bostrom (2014) really opened Ciro Santilli's eyes to this possibility.

Once this is done for a human, it will be one of the greatest milestone of humanities, coparable perhaps to the Human Genome Project. BUt of course, privacy issues are incrediby pressing in this case, even more than in the human genome project, as we would essentially be able to read the brain of the person after their death.

As of 2022, the Drosophila connectome had been almost fully extracted.

This is also a possible path towards post-mortem brain reading.

This is a simple hierarchical plaintext notation Ciro Santilli created to explain programs to himself.

It is usuall created by doing searches in an IDE, and then manually selecting the information of interest.

It attempts to capture intuitive information not only of the call graph itself, including callbacks, but of when things get called or not, by the addition of some context code.

For example, consider the following pseudocode:Supose that we are interested in determining what calls f1.

Then a reasonable call hierarchy for f1 would be:

Some general principles:

Most of these are going to be Whole-genome sequencing of some model organism:

Circa 2023, the feed is an unbearable list of stupid suggestions, never-ending idiotic memes, and you just end up missing posts you actually care about from people you actually follow.