Building a 100% accurate NEURAL NETWORK from scratch??!!

Cád é an craic, a chairde?!
In this video, I take you through the process of building a neural network from scratch using KDB+/Q. I’ll show the code, explain the functions, and walk through the logic step-by-step and we'll run it live to see what accuracy we can get. Watch to find out as we dive into neural network theory, coding tips, and real-time visualizations of the network’s progress. Slán libh 💚🤍🧡

*this is NOT a fully connected network for unexplained reasons - you'll see in another video why... if I ever get around to it... ;)*

My latest live project: https://mkeenan-kdb.github.io/gaeilge...
My github: https://github.com/mkeenan-kdb/

00:00 Intro
00:37 Visualizing a simple network
01:21 Describing a forward pass
02:44 Neuron activation
03:26 Connection strengths/weights
04:29 Describing the network's output
05:11 Output gradients and back-propagation
06:20 Intro to the code - what's first
08:05 Biases description
09:37 Hyperparameter optimization
10:02 Back to script start-up description
10:25 Looking at some input images
11:35 Defining expected outputs
12:49 Training code description
15:20 RUNNING THE TRAINING
15:44 Visualizing live model performance
16:40 100% accuracy??
17:09 Evaluating true model performance
20:42 Summary == "pretty cool"
21:10 Notes
21:36 Ideas, what's next
22:40 Outro - have fun!