Catching AI with its pants down: Demonstration of the Models in Action
| We will strip the mighty, massively hyped, highly dignified AI of its cloths, and bring its innermost details down to earth! |
Prologue
This is part 8 of this blog series, Catching AI with its pants down. This blog series aims to explore the inner workings of neural networks and show how to build a standard feedforward neural network from scratch.
This part is simply a central page for all the Jupyter Notebooks that showcase the artificial neuron model and the neural network model in action. Both models were designed straight from the mathematical formulations, and the complete information on how are detailed in this blog series.
Except for one demonstration with a toy dataset created for pedagogical purposes for this blog series, all the other demonstrations are done with real-world, research datasets.
| Parts | The complete index for Catching AI with its pants down |
|---|---|
| Pant 1 | Some Musings About AI |
| Pant 2 | Understand an Artificial Neuron from Scratch |
| Pant 3 | Optimize an Artificial Neuron from Scratch |
| Pant 4 | Implement an Artificial Neuron from Scratch |
| Pant 5 | Understand a Neural Network from Scratch |
| Pant 6 | Optimize a Neural Network from Scratch |
| Pant 7 | Implement a Neural Network from Scratch |
| Pant 8 | Demonstration of the Models in Action |
Demonstrations for the Neural Network Model
These are the demonstrations carried out for the model that involves just a network of artificial neuron.
US Adult Income Dataset
The US adult income dataset contains data on 48842 individuals, and includes fields containing data on their age, education, sex, income level (classed into 2 categories: over 50k or under 50k), etc.
The model (a neural network) learned to reach a 78% accuracy in predicting whether an individual makes over 50k a year or not, based on just data from other fields.
Full details can be found in this GitHub repo.
Chest X-Ray Pneumonia Image Dataset
This is a dataset containing 4273 chest x-ray images of children suffering from pneumonia (both bacterial and viral) and 1592 chest x-ray images of those not suffering from the lung infection.
The model (a standard feedforward neural network) learned to reach a 92% accuracy in predicting whether a patient has pneumonia or not, based on their chest x-ray image. This is despite being less suited for image learning when compared to convolutional neural networks.
Full details can be found at this GitHub repo.
Cat vs Not-Cat Dataset
This is a dataset of images of cats and those without cats.
The model (a standard feedforward neural network) learned to reach a 73% accuracy in predicting whether there is a cat in an image or not.
Will be released soon.
Demonstrations for the Single Unit Model
These are the demonstrations done for the model that involves just a single artificial neuron.
Breast Cancer Wisconsin Dataset
The Breast cancer Wisconsin dataset is a dataset that describe the characteristics of the cell nuclei for 569 breast lumps, and includes whether they are malignant or benign. Published by Clinical Science Center, University of Wisconsin.
The model (a single artificial neuron) was able to reach a 92% accuracy in distinguishing whether the tumour is malignant or benign.
Full details can be found in this GitHub repo.
Toy dataset
The toy dataset was originally introduce in part 2 of this blog series. It’s a dataset simulated using classical mechanics and random uniform noise. It had two features, velocity and mass, and a categorical binary output, energy level.
The model (a single artificial neuron) was able to reach a 89% accuracy in distinguishing whether the motion of a ball, characterized by its velocity and mass, was high or low energy level.
Full details can be found in this GitHub repo.
comments powered by Disqus