Strategic Objectives
• Discover state-of-the-art architectures without manual intervention.
• Master the search spaces and algorithms that define modern NAS.
• Reduce computational overhead while maximizing model performance.
• Implement automated pipelines for specific hardware constraints.
The Core Challenge
Designing neural networks is a tedious, trial-and-error process that consumes months of expert human labor.
01
The Dawn of AutoML
02
Foundations of Neural Networks
03
Defining the Search Space
04
The NAS Search Strategy
05
Reinforcement Learning in NAS
06
Neuroevolutionary Approaches
07
Gradient-Based Search
08
Bayesian Optimization Techniques
09
Performance Estimation Strategies
10
One-Shot Architecture Search
11
Evolutionary Computation Basics
12
Multi-Objective NAS
13
Hardware-Aware Search
14
Convolutional Neural Networks (CNNs) and NAS
15
Recurrent Structures and Transformers
16
Hyperparameter Optimization vs. NAS
17
Meta-Learning and NAS
18
Graph Neural Networks and Search
19
Computational Complexity and Scalability
20
Benchmarking and Evaluation
21