Changes between Version 8 and Version 9 of Other/Summer/2020/AdvML


Ignore:
Timestamp:
Jun 8, 2020, 2:42:45 PM (5 years ago)
Author:
yb220
Comment:

—

Legend:

Unmodified
Added
Removed
Modified

  • Other/Summer/2020/AdvML

    v8 v9  
    3333- Jupyter notebook tutorial: https://www.dataquest.io/blog/jupyter-notebook-tutorial/
    3434- Video tutorial (Optional): Neural Networks and Deep Learning: https://www.coursera.org/learn/neural-networks-deep-learning
    35 
    36 == Week 3 Activities ==
    37 - Setup the TensorFlow environment and run the Python code sample for a basic neural network.
    38 - Read the paper “X-Vectors: Robust DNN Embeddings for Speaker Recognition” (IEEE ICASSP 2018).
    39 
    40 
    41 == Week 4 Activities ==
    42 - Understand the speaker recognition system (X-Vector) and time-delay neural network.
    43 - Learn MFCC feature and extract the MFCC feature using TensorFlow.
    44 
    45 == Week 5 Activities ==
    46 - Study the Python code samples for X-Vector and implement X-Vector.
    47 - Learn how to use X-Vector and feed the extracted MFCC features into X-Vector.
    48 
    49 == Week 6 Activities ==
    50 - Read the paper “Practical Adversarial Attacks Against Speaker Recognition Systems” (HotMobile 2020).
    51 - Understand the untargeted and targeted attacks against speaker recognition systems.
    52 
    53 == Week 7 Activities ==
    54 - Understand the Fast Gradient Sign Method (FGSM) for the untargeted attack.
    55 - Study the code samples for Practical Adversarial Attacks Against Speaker Recognition Systems.
    56 
    57 == Week 8 Activities ==
    58 - Develop an untargeted attack that can generate adversarial samples based on the sample code and tutorial.
    59 - Evaluate the performance of the adversarial samples on the voice assistant system (X-Vector).
    60 
    61 == Week 9 Activities ==
    62 - Debug and fine-tune the untargeted adversarial machine learning algorithm to achieve better performance.
    63 - Develop a targeted attack that can spoof the X-Vector and misclassify the input audio signals as targeted speakers.
    64 
    65 == Week 10 Activities ==
    66 - Debug and fine-tune the developed targeted attack method.
    67 - If time allows, simulate the room impulse response (RIR) and integrate it into the developed attack methods.
    68 
    69 == Week 11 Activities ==
    70 - Fine-tune the developed targeted and untargeted attack methods.
    71 - Summarize and prepare for the open house presentation.