Reference no: EM132388514
Assignment -
This assignment contains:
- Python programming
- Traditional machine learning methods, and
- Deep learning methods
1. Python programming
Answer short questions -
What are the major Python versions? Are they inter-compatible?
What are the advantages and disadvantages of a scripting language like Python?
Name one Python package each for
- numeric calculation,
- image processing,
- data visualisation,
- machine learning.
Explain the difference between mutable and immutable objects. Given an example for each.
Functions -
How to indicate a code block in Python?
Explain the difference between local scopes and global scopes;
Refer to Code 1 - 3 attached and infer the difference between keywords global and nonlocal;
Compare Code 4 and Code 5 (attached), explain why Code 5 is able to change the value of x. Hint: is it a new assignment or an in-place change?
Numpy arrays -
Provide three reasons why numpy arrays are better options than native Python lists:
Define a numpy array arr as follows (provide code):
array([[[ 1, 2],
[ 3, 4]],
[[ 5, 6],
[ 7, 8]],
[ [ 9, 10],
[11, 12]]] )
Get the following properties of arr (provide code):
- dimensions;
- total number of elements;
- average;
- standard deviation;
- transpose;
Modify arr to the following array (provide code):
array([[ 1, 2, 3, 4],
[ 5, 6, 7, 8],
[ 9, 10, 11, 12]])
2. Traditional machine learning methods
Explain terminologies
- Supervised vs unsupervised learning
- Training vs testing
- Overfilling
- Gradient descent
Data partitioning
- Propose a data partitioning plan if the model has severl hyper-parameter to be determined.
3. Deep learning methods
Explain terminologies
- Neural networks
- Weights and biases
- Backpropagation
- Activation functions
Convolutional neural networks -
- What are the major components of a convolutional neural network?
- What does each of these components do?
- When should you use a softmax activation function?
- What loss can you use for a multi-class classification task?
Attachment:- Assignment File.rar