Tensorflow Probability

TensorFlow Probability (TFP) is a Python library that extends TensorFlow, focusing on probabilistic programming and its applications in machine learning. Current research emphasizes using TFP for Bayesian optimization, developing efficient and energy-aware deep learning models (including those based on architectures like BERT, ResNet, and various GNNs), and improving the reliability and security of TensorFlow itself through techniques like fuzzing and constraint-guided testing. This work is significant because it enhances the development, deployment, and trustworthiness of AI systems across diverse applications, from computer vision and natural language processing to robotics and scientific computing.