Друзья, хотели бы анонсировать новую встречу AI club.
Встреча пройдет в субботу 18 январе в 11:00. Пожалуйста, обратите внимание, что в этот раз она пройдет в гостеприимном новом офисе AltexSoft, который находится по адресу ул. Гражданская 15. Для того, чтобы попасть в офис, необходимо нажать звоночек у входа 🙂
Вас ждёт 3 интересных доклада:
1. Online Learning of Activation Functions in Multilayer Neural Networks
Докладчики: Yevgeniy Bodyanskiy, Prof. of Artificial Intelligence department at NURE,
Bohdan Kolchygin, Senior ML Consultant
Artificial neural networks (ANN) have been presently broadly applied to working out issues in the areas of Data Mining, Intelligent Control and Image Processing due to their multipurpose fitting approximate qualities and capabilities of training-learning by experimental data that characterize the functioning of a studied object or a studied phenomenon. A case is getting substantially more complicated if data come sequentially for processing.
2. Model Explainability in Text Classification
Докладчик: Alexander Buyko, Machine Learning Engineer at GlobalLogic
ML community became concerned about model transparency in last years and doesn’t tend to use AI as a black box anymore.
In this report, I will share the history of our experiments in building robust and trustworthy text classification model.
Also, we will discuss explainability techniques in Deep Learning and its application for model debugging.
3. Building R&D team and process. Lessons learned
Докладчик: Alexander Obiednikov, Software Development Manager at Ring Ukraine
The talk will be about what SDLC looks like and how to build product features based on ML. How to build a team in R&D scenarios, what roles are important, and why? How to deal with pure scientific research vs research and development tasks? Also, I’m going to share lessons learned during getting to production.
Вход свободный, но необходима регистрация, т.к. в зале ограниченное количество мест и мы закроем регистрацию после заполнения зала.
Немає коментарів
Додати коментар Підписатись на коментаріВідписатись від коментарів