Deep Feedforward Neural Networks (DFNN) or Multi Layer Perceptrons (MLP) for Classification #

• MLP as solution for XOR
• Hidden layers and Non-linearity
• Forward Propagation (use vectorization and algorithm for forward pass)
• Define loss functions and compute the error
• Backward Propagation (use vectorization and algorithm for backprop)
• Weight updation using gradients (use vectorization and algorithm for backprop)
• Impact of depth and width in DFNN
• Code implementation from scratch (webinar)

Reference #

• Dive into deep learning. Cambridge University Press.. (T1 – Ch 5)
• Jurafsky, D., & Martin, J. H. (Third Edition) Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition. (R4 - Ch 7.3, 7.5)

Home | Deep Learning