Dive into a comprehensive exploration of cutting-edge Machine Learning concepts in this 53-minute conference talk from NDC Conferences. Demystify complex ML buzzwords and gain a clear understanding of the latest trends in the field, from Reinforcement Learning and AutoML to ML bias, without delving into complex formulas. Learn about passing the Turing test, Isaac Asimov's contributions, and the implications of AI-driven decision-making. Explore topics such as debiasing techniques, model interpretability, and the potential for automated data science. Discover the power of transfer learning, generative adversarial networks (GANs), and their applications in various domains. Delve into deep reinforcement learning, its successes, and real-world applications. Conclude with insights on causality in machine learning and the importance of distinguishing correlation from causation. By the end of this talk, gain a comprehensive overview of the current state of Machine Learning and its future directions.
Overview
Syllabus
Intro
Passing the Turing test
Isaac Asimov
Everything under control, someone's
Choices made for you
Devices reflecting stereotypes
Aggression detected?
Debiasing by post-processing
Debiasing by removing features
Correcting bias before training
Model Agnostic Interpretation
Model Interpretability?
Can data scientist be replaced?
Framing a project
Building data pipelines
Model development
Automated Machine Learning : Azure ML
Automated ML future
Transfer skills and knowledge
Norwegian Moose Detector
Transfer Learning: replace prediction unit
Transfer Learning: freeze other layers
Pretrained networks
GAN: Generate the data?
GAN: Architecture
GAN applications : data augmentation
GAN : Image reconstruction
Image to Image transition
Al creates modern art
GAN : Adversarial attacks
Reward and Punishment
Feedback from reality
Deep Reinforcement Learning
Alpha Go Victory
Boston Dynamics: back to basics
Adaptability
Correlation is not Causation
Cause-Effect connection
Causal Learning
Taught by
NDC Conferences
