Data

Tools

Indexes

Applications

Experiments

Open Source Science

https://github.com/ahmetpala/artificial-neural-networks-bu-ee550

https://github.com/ahmetpala/artificial-neural-networks-bu-ee550

Science Score: 26.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (8.6%) to scientific vocabulary
Last synced: 10 months ago · JSON representation

Repository

Basic Info
  • Host: GitHub
  • Owner: ahmetpala
  • Language: Python
  • Default Branch: main
  • Size: 1010 KB
Statistics
  • Stars: 1
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created about 5 years ago · Last pushed about 1 year ago
Metadata Files
Readme

README.md

Machine Learning Implementation Projects

This repository contains a collection of small machine learning projects. All code is written in Python using NumPy, pandas, and Matplotlib. These are educational and experimental implementations.

Projects

1. Recursive Least Squares (RLS) Regression

  • Generates 15 noisy data points from a simple polynomial.
  • Fits four models of increasing complexity using the RLS algorithm:
    • Model 1: constant
    • Model 2: linear
    • Model 3: quadratic
    • Model 4: cubic
  • Calculates error for each model.
  • Plots each estimated curve.

2. Hopfield Network

  • Creates 8x8 binary images for digits 1, 4, 7, and 9.
  • Adds noise with three different noise levels.
  • Trains a Hopfield network to memorize the original patterns.
  • Tests recovery of original patterns from noisy inputs.
  • Plots each iteration until convergence.

3. XOR Neural Network and Function Approximation

3.1 XOR Problem

  • Neural network with:
    • 2 input neurons
    • 1 hidden layer (5 neurons)
    • 1 output neuron
  • Sigmoid activation, MSE loss.
  • Trains with backpropagation to solve XOR.
  • Plots training loss.

3.2 Function Approximation

  • Target: y = sin(x) + 2*cos(x)
  • Neural network with:
    • 1 input
    • 2 hidden layers (4 and 5 neurons)
    • 1 output
  • Uses tanh in hidden layers, linear output.
  • Trains with MSE loss.
  • Plots predictions and error.

3.3 Handwritten Digit Recognition

  • Uses optdigits dataset (CSV).
  • Neural network with:
    • 64 input neurons
    • 2 hidden layers (32 and 24 neurons)
    • 10 output neurons (softmax)
  • Cross-entropy loss.
  • Classifies digits 0–9.
  • Shows accuracy and predictions.

4. Radial Basis Function (RBF) Network

  • 1D function: y = 0.4*cos(2πx) + 0.6*sin(7πx) + 0.5 + noise
  • Uses 5 Gaussian basis functions (centers chosen manually).
  • Trains using gradient descent.
  • Plots predictions and loss.

Requirements

  • Python 3.x
  • NumPy
  • pandas
  • matplotlib
  • scikit-learn (only used for KMeans in RBF)

How to Run

Clone the repo and run any .py or .ipynb file. Each file is self-contained.

Owner

  • Login: ahmetpala
  • Kind: user

GitHub Events

Total
  • Push event: 40
  • Pull request event: 23
  • Create event: 2
Last Year
  • Push event: 40
  • Pull request event: 23
  • Create event: 2