Select the Dataset(s) of your Choice You can use Data from Your Workplace, Create your own or Consider Getting Data: Specialist Diploma in Applied Artificial Intelligence Coursework, RP, Singapore

C3379C Pattern Recognition and Anomaly Detection

Requirements:

1) Select the dataset(s) of your choice. You can use data from your workplace, create your own or consider getting data from the recommended references listed below.

2) Pick four anomaly detection algorithms from the following and develop the anomaly detection models for your chosen dataset(s). Algorithms include:

a. OC-SVM
b. Isolation Forest
c. Deep Autoencoder
d. Convolutional Autoencoder
e. Denoising Autoencoder
f. Dilated Temporal Convolutional Network
g. Encoder-Decoder Temporal Convolutional Network
h. LSTM

3) For each algorithm, you should demonstrate the following tasks as covered in the module (15 marks for each algorithm). Your submission should include clear comments to explain the code used.

a. Apply data pre-processing techniques where necessary on your dataset for training and evaluation

b. Train and optimise each algorithm with two hyperparameter tuning techniques. Compare the model’s performance with each hyperparameter changed.

c. Evaluate and compare your models’ performance with a chosen metric where appropriate.

Recommended references for data (You may source for other datasets too):
• https://kwseow.github.io/
• http://deeplearning.net/datasets/
• https://www.datasetlist.com/
• https://datasetsearch.research.google.com/
• https://cloud.google.com/public-datasets/
• https://data.gov.sg/
• https://www.singstat.gov.sg/find-data
• https://data.world/