Building a Neural Network from Zero

Course Overview

• Course Title: Building a Neural Network from Zero
• Instructor: Nick Ovchinnikov
• Target Audience: Aspiring machine learning engineers, curious programmers
• Prerequisites:
  • Basic knowledge of Python programming
  • Familiarity with linear algebra concepts like vectors and matrices
  • An interest in understanding neural networks at a fundamental level

Curriculum Highlights

• Key Topics Covered:
  • Implementing neural networks from scratch
  • Gradient descent and optimization techniques
  • Building custom layers, activation functions, and loss functions
  • Solving the Fashion-MNIST classification challenge
  • Numerical differentiation and gradient computation
  • Stochastic Gradient Descent (SGD) with momentum
  • Cross-entropy loss and activation functions like Sigmoid
  • Weight initialization methods (He and Xavier)
  • Building a Feedforward Neural Network (FFNN) from scratch
• Key Skills Learned:
  • Implementing forward and backward propagation
  • Mastering gradient descent and other optimization techniques
  • Creating custom layers, activation functions, and loss functions
  • Applying neural networks to classification challenges

Course Format

• Duration: 4 hours on-demand video
• Format: Self-paced online course
• Resources:
  • Access on mobile and TV
  • Certificate of completion