Duration
21 hours (usually 3 days including breaks)
Requirements
Any programming language knowledge is required. Familiarity with Machine Learning is not required but beneficial.
Overview
Audience
This course is suitable for Deep Learning researchers and engineers interested in utilizing available tools (mostly open source) for analyzing computer images
This course provide working examples.
Course Outline
Deep Learning vs Machine Learning vs Other Methods
- When Deep Learning is suitable
- Limits of Deep Learning
- Comparing accuracy and cost of different methods
Methods Overview
- Nets and Layers
- Forward / Backward: the essential computations of layered compositional models.
- Loss: the task to be learned is defined by the loss.
- Solver: the solver coordinates model optimization.
- Layer Catalogue: the layer is the fundamental unit of modeling and computation
- Convolution​
Methods and models
- Backprop, modular models
- Logsum module
- RBF Net
- MAP/MLE loss
- Parameter Space Transforms
- Convolutional Module
- Gradient-Based Learning
- Energy for inference,
- Objective for learning
- PCA; NLL:
- Latent Variable Models
- Probabilistic LVM
- Loss Function
- Detection with Fast R-CNN
- Sequences with LSTMs and Vision + Language with LRCN
- Pixelwise prediction with FCNs
- Framework design and future
Tools
- Caffe
- Tensorflow
- R
- Matlab
- Others…