Practical Deep Reinforcement Learning (PDRL)

Over the course of 7 weeks, participants will delve into deep reinforcement learning (DRL), an advanced area of machine learning that has transformed domains like game playing, robotic control, healthcare, supply chain optimization, and smart building management. By leveraging PyTorch, one of the most widely used deep learning frameworks, you will gain practical skills and the confidence needed to build and deploy DRL models effectively.

This certificate is ideal for students, researchers, and professionals with a basic background in machine learning and Python. Participants are expected to be familiar with basic statistics and Python.

Designed with you in Mind

Participants will achieve the following learning objectives:

• Understanding of DRL Concepts
  Acquire a foundational understanding of DRL principles, including key algorithms like deep Q-networks (DQN), policy gradients, and actor-critic architectures.
• PyTorch Proficiency
  Gain hands-on experience in coding and implementing DRL algorithms using PyTorch.
• Application of DRL Techniques
  Explore how DRL techniques are applied in different fields, such as game playing, robotics, healthcare, and smart energy systems.
• Model Optimization Skills
  Learn to evaluate and optimize DRL models, focusing on reward shaping, exploration strategies, and hyperparameter tuning.
• Deployment Knowledge
  Understand the practical considerations for deploying DRL models, emphasizing real-world implementation challenges.

Upon completing the PDRL program, participants will:

• Master the fundamentals of deep reinforcement learning, including deep Q-networks (DQN), policy gradients, and actor-critic methods.
• Develop proficiency in using PyTorch to implement DRL algorithms efficiently.
• Understand practical applications of DRL across multiple industries and domains.
• Evaluate and optimize DRL models using advanced techniques, including reward shaping and hyperparameter tuning.
• Gain insight into real-world deployment challenges such as scalability and safety.
• Demonstrate the ability to solve real-world problems using DRL models.

Meet Your Instructor

Image

Victor Palacios serves as the Director of Data Science and Artificial Intelligence Partnerships at the University of San Francisco, where he develops strategic industry collaborations and supports applied AI innovation across the Data Institute. He holds three graduate degrees: an MS in Information Science from Nagoya University in Japan, an MS in Data Science from USF, and he is currently completing a third master's in Computer Science.

Full Profile