Deep Learning without PhD, masters, graduation
Mayur Bhangale StoreKey
A Python based scientific computing package targeted at two sets of audiences:
• A replacement for numpy to use the power of GPUs • A deep learning research platform that provides maximum flexibility
and speed
Basics
Variables, Tensors, Autograd
Predictive models
Linear Regression
• Fit a line to a data set of observations • Use this line to predict unobserved values
Predictive models
Linear Regression
Input
x
Output
y
mx + c = y
Linear Function
10 20 100 200
1.3 1.2 4.5 4.8
Predictive models
Logistic Regression
• Predicts the probability of occurrence of an event by fitting data to a logit function.
• Used to predict a binary outcome (1 / 0, Yes / No, True / False) given a set of independent variables.
Predictive models
Logistic Regression
Inputx
Logits
y
mx + c = y
Linear Function
10 20
100 200
1.3 1.2 4.5 4.8
S = g(y)
SoftmaxFunction
g(y)
Softmax
0.1 0.1 0.4 0.4
D(S, L)
CrossEntropyFunction
0 0 1 1
L
True Labels
Predictive models
Input
x
Logits
y
Linear Function
SoftmaxFunction
g(y)
Softmax
D(S, L)
CrossEntropyFunction
Labels
L
True Labels
Readout LayerInput Layer Output Layer
Logistic Regression
Predictive models
Input
x
Logits
y
Linear Function
SoftmaxFunction
g(y)
Softmax
CrossEntropyFunction
Labels
L
True Labels
Readout LayerInput Layer Output Layer
ExampleSize: 784
Size: 10
Convolutional Neural Network