Explore deep learning techniques with differential privacy in this one-hour conference talk by Li Zhang at #ICBS2024. Delve into the challenges of training neural network models on sensitive, crowdsourced datasets while preserving privacy. Learn about innovative algorithmic techniques developed within the differential privacy framework to protect private information. Discover how to implement these methods to train deep neural networks with non-convex objectives under a modest privacy budget. Examine the trade-offs between software complexity, training efficiency, and model quality when incorporating privacy measures. Gain insights into the speaker's influential paper on privacy-preserving machine learning and its impact on the field. Explore follow-up work and current state-of-the-art approaches in this critical area of machine learning research.