skipjacks

Client-Side JavaScript AI or Neural Net with CSide requests, training and output

To see the old README.md, it is still being worked on, because it is completing a different task than a standard readme. Originally it was going to be short, but as things progressed I realised I could do a whole lot more and offer alot more insight. But as you'll be able to see, from the starting few sections, that it was originally planned to summarise this project in terms of prior-research, need-for-development, roadmap, and novel-extensions-to-theory that are needed, and a start on the mathematics underlying the most important of one of those theories.

Install

I never know if 1, 2, 3 means options or steps. These are options.

1. With npm from npmjs

npm install skipjacks

2. With npm from GitHub

npm install git+https://github.com/JDonMc/skipjacks.git

3. From source for development, modifications, customisation.

git clone https://github.com/JDonMc/skipjacks cd skipjacks

4. With yarn from GitHub

yarn install [email protected]:JDonMc/skipjacks.git

How to Use

Currently a work in progress, building out from old files I have in Python, and in C++. I tried asking ChatGPT for help, but I think I ended up wasting my time and making things worse. She kept losing her memory of what was going on, talking in circles, and blatantly refusing to acknowledge my points.

The class / object structure is made to be representative of the abstract form of AI in it's most primitive sense.

Import statements

1. With require

const { Connection, Neuron, Layer, NeuralNet } = require('skipjacks');

2. With import

import { Connection, Neuron, Layer, NeuralNet } from 'skipjacks';

3. With <script> and node as an installed node_module

<script type="module" src="skipjacks/index.js"></script>

4. With <script> on the client-side, no custom installs, which is ultimately what it's built for.

<script src="https://unpkg.com/[email protected]/bundle.js"></script>

Initialising a NeuralNet and all it's components

Roadmap

So the task at hand is to build the base neural networks with set activation functions, and easy customisability of activation functions to include the potential for novel activation functions for say the hyperbolic arc sin, or the hyperbolic (representing how Mice brains have now been measured to record spatial data, according to Nature 2023), which can be done through ready class inheritence perhaps, or even a custom method storing a given equation for the activation function and a given equation for its derivative function (as we've learnt that these are not 1:1 guaranteed implicatives, and alcalculus will need further modification for given assumptional characteristics of known datatypes), as well as configurable layer subsets each with unique activation functions, as well as transformers and multi-modal tokenizers (words, characters, subwords, which we'll then have to modify and extend to include the notion that the same word can have different definitions in different contexts, even within a single sentence).