JordanMicahBennett/Live-Agile-Artificial-Neural-Network-Programming-Sessions

Agile neural network programming sessions, for reasonably rapid data science preparation.

Live-Agile-Artificial-Neural-Network-Programming-Sessions

Agile neural network programming sessions by God Bennett, for reasonably rapid data science preparation.

🎥🎥My live youtube session (from scratch/memory):

https://github.com/JordanMicahBennett/BASIC-ARTIFICIAL-NEURAL-NETWORK_FROM-LIVE-JAVA-SESSION

Neural nets normally concern signals bouncing around and being transformed, starting from some problem space like the numbers in an xortable, and ending in some guess about what some output should be given some inputs. (EG: XR GATE TASK)

1. Loops live inside each neuron, that enable communication/calculation with other neurons
2. Loops that organize the neurons together such that they can communicate
3. Loops that feed sample problem data to the neural network above that tend to generate single signals, i.e. an answer to some problem

Some of God Bennett's other works/roles:

• Manager of Artificial Intelligence at AICE via Adalabs Africa - May 2021 to present.
• Creator of the Supersymmetric-artificial-neural-network. (Reasonably observable as a new branch or field of Deep Learning in Artificial Intelligence, called Supersymmetric Deep Learning)
• Creator of Cryptosynth Aeons - 1st Ai powered beings in blockchain history that anyone can hold a conversation with
• Creator of NVDIA featured smart Ai pothole detector.
• Creator of World's first open source Covid19 Ai Diagnosis tool.
• Creator of Machine Learning Jamaica Institute Initiative.
• Author of Why is the purpose of human life to create Artificial General Intelligence?
• ...

Fun note: 1000+ lines of basic neural network code were written in Java from memory/scratch by God Bennett for the session below. God can also write the same 1000+ lines of neural network code at any given time from memory/scratch, after years of bi-annual practice. Note that memorizing this neural network code as God does, is reasonably not required, although writing one out step by step is beneficial, as discussed below.

Artificial General Intelligence (AGI), also known as human level AI, is known as humanity's last big invention.

1. Artificial General Intelligence (AGI), also known as human level AI, is known as humanity's last big invention, according to AI/AGI experts including winner of 2018 Turing award (similar to nobel prize but for computer science) Professor Yoshua Bengio, as well as Dr Ben Goertzel, and other AGI experts including Eray Ozkural, winner of 2015 Kurzweil AGI award.

   • Notably, AGI/human level AI is slated to emerge by 2029, based on Google/Kurzweil's law of accelerated change.

2. Google, Microsoft, as well as many major powers, are racing towards building AGI.

3. Some of the world's smartest people have changed their careers/focus from physics to focus on Artificial General Intelligence/Ai research, including world renknowned physicist Prof Max Tegmark, famous quantum engineer Dr. Suzanne Gildert, etc, and even chip makers have been leaving behind the old Von Neumann architecture in computing for a while now, with more focus on parallelizable, i.e. reasonably more brain like frameworks, like tpus, vpus, and neuromorphic chips.

This session aims to equip the programmer with fundamental artificial neural network programming skills. Neural networks (embedded inthe goal of AGI research) are universal function approximators aka universal tools that can approximate a wide variety of tasks, somewhat similar to how 1 human brain or natural neural network can do driving, teaching, nursing etc, reading, writing etc.

• Agi is likely humanity’s last invention as a type of meta solver to reasonably solve most of humanity’s issues, including the hardest Math problems, Physics problems, …. reasonably the entire landscape of cognitive thinking.

Sessions 2019 to /Dec 18, 2020/NCB

• Held at Conference ITS 2019. (See the third item entitled "ITS2019 Artificial Intelligence Code Tutorial"). (See also the github repository)
• Held at National Commercial Bank, with small QA team, December 18, 2020.
• Held at Jamaica Artificial Intelligence Association, March 21, 2021.

Sessions 2021, Universal Ai Diploma

• Universal Ai Diploma 2021 Neural Network from scratch Mandatory writing Sessions

Details:

a. Live Agile Artificial Neural Network Programming Sessions_v5.pdf

b. Quick technical overview: The "How"??

Introduction

In an agile setting where months may not be available for data science courses, this is a reasonably optimal way of quickly getting a developer of starting or intermediate coding skills to understand overall layout of machine learning.

Understanding these broad/fundamental neural network models from scratch without machine learning libraries, builds intuition in applying and debugging machine learning libraries.

Since artificial neural networks can represent a wide variety of utilities/functions as seen in Universal Approximation Theorem, and since neural networks power most cognitive/smart apps today, God's agile artificial intelligence/machine learning process concerns 2 topics:

1. Most basic Neural network programming in detail (in BlueJ or any IDE of choice. BlueJ is good for quickly visualizing how classes connect, as coding takes place.)
2. Data preprocessing.

Picture created in Gimp by God, by combining an empty brain stencil/outline, with a picture of a Galaxy.

Other notable sources/alternatives for basic neural network code lessons, though not strictly in Java:

1. A March post of mine underlining Microsoft's Joseph Albahari's basic neural network lesson.
2. Sentdex's Neural Network From scratch series.