AutoML Seminars Courses

Large Language Models to Enhance Bayesian Optimization

Discover how Large Language Models enhance Bayesian optimization through LLAMBO, improving hyperparameter tuning and black-box function optimization with natural language integration.

• YouTube
• 31 minutes
• On-Demand
• Free Video

The AI Scientist - Towards Fully Automated Open-Ended Scientific Discovery

Explore groundbreaking research on autonomous AI systems capable of conducting end-to-end scientific discovery, from generating ideas to writing papers and performing peer reviews in machine learning.

• YouTube
• 52 minutes
• On-Demand
• Free Video

Searching for Neural Architectures from Fundamental Operations - An Introduction to einspace

Explore neural architecture search through einspace, a novel approach using fundamental operations to discover diverse and competitive network architectures for machine learning tasks.

• YouTube
• 38 minutes
• On-Demand
• Free Video

Mixture-of-Supernets - Improving Weight-Sharing Supernet Training with Architecture-Routed MoE

Explore cutting-edge research on Mixture-of-Supernets architecture that enhances neural architecture search efficiency and improves BERT and MT model performance through innovative weight-sharing techniques.

• YouTube
• 35 minutes
• On-Demand
• Free Video

Vanilla Bayesian Optimization Performs Great in High Dimensions

Discover how scaling Gaussian process lengthscale prior with dimensionality enables vanilla Bayesian optimization to excel in high-dimensional optimization problems, outperforming complex alternatives.

• YouTube
• 35 minutes
• On-Demand
• Free Video

Multi-objective Differentiable Neural Architecture Search

Explore advanced techniques in neural architecture optimization across multiple objectives and hardware devices, focusing on Pareto front profiling and hypernetwork-based solutions.

• YouTube
• 43 minutes
• On-Demand
• Free Video

Neural Fine-Tuning Search for Few-Shot Learning

Explore advanced neural architecture search techniques for optimizing few-shot learning adaptation strategies, focusing on meta-training and meta-testing trade-offs in multi-domain contexts.

• YouTube
• 30 minutes
• On-Demand
• Free Video

Is Mamba Capable of In-Context Learning? - An Analysis of State Space Models

Explore how Mamba, a state space model, demonstrates in-context learning capabilities comparable to transformers while handling longer input sequences more efficiently in AI applications.

• YouTube
• 44 minutes
• On-Demand
• Free Video

Quick-Tune: Quickly Learning Which Pretrained Model to Finetune and How

Discover a methodology for efficiently selecting and fine-tuning pretrained models through meta-learning, optimizing model selection and hyperparameter configurations for new datasets.

• YouTube
• 40 minutes
• On-Demand
• Free Video

OmniPred - Towards Universal Regressors with Language Models

Explore groundbreaking research on OmniPred's innovative approach to using language models as universal regressors, demonstrating superior performance over traditional regression methods across diverse experiments.

• YouTube
• 49 minutes
• On-Demand
• Free Video

AutoGluon 1.0 - Shattering SOTA with Zeroshot-HPO and Dynamic Stacking

Discover how AutoGluon 1.0 revolutionizes machine learning with Zeroshot-HPO and Dynamic Stacking, achieving superior performance in automated model training and ensemble stability.

• YouTube
• 45 minutes
• On-Demand
• Free Video

Rethinking Bias Mitigation: Fairer Architectures Make for Fairer Face-Recognition

Explore groundbreaking research on addressing bias in face recognition systems through neural architecture optimization, revealing how architectural choices impact fairness more than data preprocessing alone.

• YouTube
• 41 minutes
• On-Demand
• Free Video

MotherNet: A Foundational Hypernetwork for Tabular Classification

Discover how MotherNet hypernetwork architecture revolutionizes tabular classification by generating optimized neural networks through in-context learning, offering faster training and competitive performance against XGBoost.

• YouTube
• 58 minutes
• On-Demand
• Free Video

Multi-Objective Population Based Training for Hyperparameter Optimization

Discover advanced hyperparameter optimization techniques through Population Based Training (PBT) and its multi-objective evolution, focusing on real-world applications in precision, fairness, and robustness.

• YouTube
• 26 minutes
• On-Demand
• Free Video

LLMs for Automated Data Science - Introducing CAAFE for Context-Aware Automated Feature Engineering

Discover how CAAFE leverages large language models to enhance automated feature engineering in data science, improving model performance through context-aware feature generation and semantic understanding.

• YouTube
• 30 minutes
• On-Demand
• Free Video