AI

A collection of python code exploring a variety of AI methods

Contents

• Neural Net

  • Decision Boundary
  • Output Functions

• Clustering

  • Data Formatting
  • Results

• Sentiment Analysis

  • Naive Bayes
  • Decision Tree

• Regression

Neural Net

A neural net implemented through Python to solve the XOR problem.

Classifiers such as uni layered perceptrons can only handle linearly separable data, making them unable to solve non-linear problems like the XOR problem. Neural nets (or Multi Layered Perceptrons), however, are capable of handling such non-linear problems, making the XOR problem a suitable demonstration of their applications.

XOR Illustration

Classifier Decision Boundary

Final and Hidden Output Functions

References

[1] https://medium.com/analytics-vidhya/coding-a-neural-network-for-xor-logic-classifier-from-scratch-b90543648e8a

[2] https://towardsdatascience.com/how-neural-networks-solve-the-xor-problem-59763136bdd7

[3] https://dev.to/jbahire/demystifying-the-xor-problem-1blk

Clustering

K-Means clustering implementation performed on a sample image of Obama.

Data Formatting

Results

Sentiment Analysishw1.pdf

A Python program that takes in a set of sentiment labeled reviews.

A Naive-Bayes classifier and Decision-Tree classifer are trained and tested using an 80-20 split on the provided data.

Data Sentiment Spread

Naive Bayes

Naive Bayes : Training Results

Smoothing	Accuracy	Weighted Recall	Weighted Precision	Weighted F1-Measure
1.00	0.869793306053	0.869793306053	0.870386391303	0.869789754171
0.25	0.871681880180	0.871681880180	0.87218039395	0.871682335029
0.10	0.871891721750	0.871891721750	0.872378943981	0.871892618677
0.05	0.871996642534	0.871996642534	0.872478265022	0.871997755730
0.025	0.872311404889	0.872311404889	0.872798854901	0.872312298878

Naive Bayes : Testing Results

Smoothing	Accuracy	Weighted Recall	Weighted Precision	Weighted F1-Measure
0.05	0.819135543432	0.819135543432	0.819237343523	0.819162061498

Decision Tree

Decision Tree : Training Results

Accuracy	Weighted Recall	Weighted Precision	Weighted F1-Measure
0.999895079215	0.999895079215	0.999895101610	0.999895079402

Base Decision Tree : Testing Results

Accuracy	Weighted Recall	Weighted Precision	Weighted F1-Measure
0.685270667226	0.685270667226	0.686562742126	0.685290619552

Best Decision Tree : Testing Results

Accuracy	Weighted Recall	Weighted Precision	Weighted F1-Measure
0.718422156945	0.718422156945	0.718679219277	0.718483773165

Regression

Linear and polynomial regression, theory and implementation, done as an assignment for Prof. Yang Wang at Concordia University Montreal, completed on the Winter 2023 semester.

Report

Polynomial Regression