

[Повторение] Machine Learning: Attention Mechanisms in Deep Learning

IT общността
Спонсори на събитието:
За събитието
*The event will be in English.
This event is part of the DEV.BG’s Machine Learning user group.
>>> The construction and the idea behind the Attention Mechanism
>>> Types of Attention Mechanisms (Soft, Sparse, Hard)
>>> Attention Mechanisms in Natural Language Processing
>>> The Transformer Model
>>> The power of transfer learning from Attention-based NLP models
>>> Attention Mechanisms in Computer Vision
>>> The Vision Transformer Model (ViT)
>>> The family of ViT-like models
>>> ViT-like models in image classification
>>> ViT-like models in object detection
>>> Transfer Learning from ViT-like models
>>> Using Attention as an Interpretability tool in Natural Language Processing
>>> Using Attention as an Interpretability tool in Computer Vision
October 12th, 12:30 pm

За лектора
Stefan is Founder and CEO of Miracle Star Ltd., software engineer with around 7 years of experience in software development, the last 4-5 of which he spent as a Deep Learning Engineer. Stefan has worked on many projects involving application of Deep Learning in Computer Vision and Natural Language Processing. He holds a Bachelors degree in Mathematics and Computer Science from the American University in Bulgaria and a Masters degree in Advanced Computer Science (with a focus on Natural Language Processing) from the University of Cambridge. Recently, he co-authored an academic article with the title ‘Is Sparse Attention More Interpretable?’ which was published at the prestigious ACL-IJCNLP conference.
Нашите
Партньори
HyperScience brings AI to the office. Through ML solutions they help various enterprises and government institutions to automate the hectic back office work.
FactSet delivers superior analytics, service, content, and technology to help more than 88,000 users see and seize opportunity sooner.
Lab08 is one of the youngest software companies in Bulgaria, which develops products in the field of HR, consumer tests, social media and more. It is mainly made up of professionals with invaluable experience in building big data systems with multiple integrations.