Introduction to Deep Learning and Convolutional Neural Networks (CNNs)

GDG Oxford

Dives into the world of deep learning. Shows why deep learning excels in image tasks, demystifies the architecture of CNNs (convolutions, pooling, etc.), and explains the training process.

Mar 2, 2:00 – 3:00 PM

5 RSVP'd

Key Themes

AIBuild with AIMachine Learning

About this event

Feature Extraction:

  • Hand-crafted features: SIFT, HOG, LBP
  • Feature engineering and selection

Classification Algorithms:

  • k-Nearest Neighbors (kNN)
  • Support Vector Machines (SVM)
  • Decision Trees and Random Forests
  • Case Study: Building an image classifier using the techniques above (e.g., handwritten digit recognition)

The limitations of traditional approaches

  • Biological inspiration of neural networks
  • Building blocks of CNNs:
  • Convolutional layers
  • Pooling layers
  • Activation functions (ReLU, sigmoid)
  • Fully connected layers

Training CNNs:

  • Loss functions (cross-entropy)
  • Optimization algorithms (stochastic gradient descent, Adam)
  • Backpropagation


  • Gregory McGann

    GDG Oxford

Contact Us