basics
A subset of Machine Learning that uses neural nets.
Deep Learning
The basic unit that processes and passes info in a neural net.
Neuron
This powers self-driving cars.
Neural networks
First step in model training.
Forward propagation
This hardware is ideal for training deep models.
GPUs
Deep Learning models are inspired by this organ.
Human brain
These are the three types of layers in an ANN.
Input, hidden, and output layers
Deep Learning helps diagnose diseases from this
X-rays
What follows forward propagation in training
Compute loss
Deep Learning works best with this kind of dataset.
Large datasets
Deep Learning automates this, unlike traditional ML.
Feature extraction
This function decides if a neuron activates.
Activation function
This field uses DL for understanding human language.
Natural language understanding
: This step adjusts weights to reduce error.
Backpropagation
This process measures how wrong the prediction is.
Loss computation
The foundation model type for Deep Learning.
Artificial Neural Networks (ANNs)
All of these are valid activation functions.
ReLU, Softmax, Sigmoid
DL is widely used because it can solve these.
Complex problems
The loop: forward → loss → backprop → repeat.
Training loop
DL needs this due to complex data and layers.
Powerful hardware
DL is powerful because it learns from this.
Multiple complex layers
DL models are hard to interpret because they use many of these.
Multiple complex layers
DL can do all of these: image classification, speech & medical diagnosis.
All of the above
This full process enables a model to make accurate predictions.
Neural network training process
Key to mastering DL: neural nets, optimization, and this.
Large datasets