The school will start on May 22, 2023, Monday. It will be divided into two halves of 5 days each, with a hiking day in the middle.
The Themes and topics for ANAIS 2023 are below. 

#ANAIS2023 | THEME & TOPICS

FOUNDATIONS:  

Probability and Statistics, Linear Algebra, Machine Learning, Neural Networks.

TRANSFORMERS: 

Foundations, history, promise, challenges.

MULTI-MODAL DATA: 

Models that leverage data from more than one domain, such as image and language, signals and images etc.; vision-language models.

DATA SCIENCE AND ENGINEERING: 

Majority of existing works focus on models, network architecture engineering, and proposing new models. However, the performances of most neural network methods have some level of dependence on the dataset itself. This theme will focus on teaching works on engineering the dataset and problem itself to improve performance from existing models and methods. 

Topics: Data augmentation, data synthesis, best data selection for pretraining, understanding the dependence of data and its characteristics on model performance, active learning etc.

GENERATIVE AI:

Diffusion models, NERF, ChatGPT etc. 

AI-SOCIETY DEMOCRATIZATION OF AI: 

We would like to focus on two themes this year: 

1. Trustworthy AI: Fairness, Usability, Explainability and Interpretability, Universality, Federated learning.

2. Low resource and affordable AI: Efficient computing and moving away from the dependence on heavy computing, Knowledge distillation, Domain adaptation, and Semi-supervised & Unsupervised learning (reducing the cost of annotation).
   
   

AI APPLICATIONS:

A wide range of applications depending on speakers' interests such as healthcare, agriculture, networks, finance, science etc.