In this project, you will explore the use of image gradients by applying it on the following three tasks:  o Adversarial Attack: “Intriguing properties of neural networks”
- Class Visualization: “Deep Inside Convolutional Networks: Visualising Image Classification
Models and Saliency Maps” and “Understanding Neural Networks Through Deep Visualization” o Style Transfer: “Image style transfer using convolutional neural networks” Â
Please go through the above reference papers very carefully. You are expected to understand how to accomplish each task using the methods presented in the papers and write code to implement them by completing Project1.ipynb which is available in Project1.zip from LumiNUS.
Please take careful note of the following (failure to do so could incur penalties):Â o your code MUST be kept within “TODO” and “END OF YOUR CODE”;
- do NOT modify the definitions of the functions in Project1.ipynb; o write “clean” (easily readable) code and check to make sure it runs/executes;
- do NOT share your solution code with others; submit your own work/code; there will be penalties for cheating and late submission.
