Description:
In the world of machine learning, uncertainty estimation is crucial for applications where predictions need to be accompanied by a measure of confidence. Current methods for uncertainty estimation, such as Bayesian Neural Networks, mainly focus on introducing randomness in the weights of the network. This project aims to take a novel approach by introducing stochasticity directly into the activation functions of a Binarized Neural Network (BNN), creating a Stochastic Binarized Activation (SBA) function.

The primary objectives of this thesis project will be:
1. Development: Define a novel SBA function that introduces stochasticity into the activation process of a BNN, with the goal of improving uncertainty estimation.
2. Evaluation: Design and conduct experiments to assess the effectiveness of the SBA, comparing its performance against conventional BNNs and other baseline models.
3. Optimization: Investigate methods to optimize the training process and computational efficiency of BNNs employing the SBA function.
4. Application: Apply the developed method to real-world tasks where uncertainty estimation is crucial, such as medical diagnosis or financial forecasting.

Requirements:
• Good background in machine learning and deep learning
• Familiarity with Binarized Neural Networks and Bayesian methods for uncertainty estimation or interest to learn
• Proficiency in Python and deep learning libraries such as PyTorch or TensorFlow
• Good mathematical and statistical skills
• Interest in novel research and pushing the boundaries of current machine learning techniques

Benefits:
• Opportunity to contribute to the cuZng edge of machine learning research
• Gain experience in developing and evaluating novel neural network architectures
• Potential to publish findings in a top-tier machine learning/EDA conference or journal

Interested candidates should contact Soyed Tuhin Ahmed and Prof. Mehdi Tahoori with their CV, transcript, and a brief description of their interest in the project.

Reference:
1. Xia, Yufeng, et al. "RBUE: a ReLU-based uncertainty estimation method for convolutional neural networks." Complex & Intelligent Systems (2023): 1-15.
2. Abdar, Moloud, et al. "A review of uncertainty quantification in deep learning: Techniques, applications and challenges." Information fusion 76 (2021): 243-297.