Understanding Neural Networks - Speech Recognition & Measuring Quality of ML

The explosion of ChatGPT and GPT4 into mainstream applications may seem like we have solved all there is regarding language. However, there are still challenges that AI struggles with.

In collaboration with Better AI Meetup, we will be talking about speech processing as well as designing ways to evaluate machine learning algorithms for accuracy and speed and choose a proper one for a particular use case.

Evaluating Machine Learning Algorithms (Jakub Sochor)
Imagine you want to use a machine learning model but not sure which one. How can you know which algorithm is the best one for a particular use case? Jakub will present options that you might have and will showcase US NIST benchmarks that are around for almost 20 years now.

About speaker:
Jakub studied Ph.D. at FIT VUT in Brno with a focus on machine learning and computer vision. During the studies, Jakub wrote several first-author top-level publications (e.g. CVPR, CVIU). Afterwards, he worked on machine learning applications at multiple companies (PTC, Google) in Zurich, Switzerland. Currently, Jakub leads the R&D team of biometric solutions at Innovatrics.

Two revolutions in the life of a senior speech recognition guy (Jan Černocký)
When I started my career in speech processing around 2000, neural networks were at the edge of interest and the world was dominated by Gaussian models. Thanks to the work of Geoffrey Hinton, Tomas Mikolov and many others, NNs prevailed after 2010 and defined a new paradigm not only for speech but for many domains of machine learning. 2020's witness another revolution - the world is revolving around large pre-trained models that bring a disruptive improvement in performances, but also problems not seen before. This talk will evaluate our options in the current situation.

About speaker:
Jan "Honza" Černocký (Ing. BUT 1993, Dr. Universite Paris XI Orsay and BUT 1998) is full professor and Head of the Department of Computer Graphics and Multimedia, Faculty of Information Technology, Brno University of Technology. He also serves as managing director of BUT Speech@FIT research group. His research interests include artificial intelligence, signal processing and speech, speaker and language recognition. At FIT BUT, he is responsible for signal and speech processing courses. In 2006, he co-founded Phonexia. He was general chair of Interspeech 2021 in Brno.

## AGENDA

_5:00pm Welcome
_5:10pm Presentation - Evaluating Machine Learning Algorithms - Jakub Sochor, Innovatrics
_5:40pm Presentation - Two revolutions in the life of a senior speech recognition guy - Prof. Jan Černocký, FIT VUT
_6:10pm Live Q&A – _Slido #better_ai