← Back to the training program

Fernando Velasco


From faces to Language: Building an intelligent assistant capable of seeing and talking

Artificial Intelligence - Deep learning, Machine Learning



One of the key steps of the late ashtonishing growth AI has experienced lately is the development of new techniques that enable the possibility of facing issues that are closer to human intelligence. From very powerful usages such as forecasting in trading, markets or risk, we are now able to address different tasks like vision, hearing or language, that a priori seem simpler for the human being, but harder from a computational point of view.

On this training we will address two of these tasks. On the one side we will focus on vision and will create a facial recognition system. Those systems are widely used on the industry, from surveillance cameras to flow detection in airports, and including football player tracking during a match. On the other side we will make use of Natural Language Understanding techniques in order to build a conversational assistant. Those assistants are of great use when solving customers issues such as the ones that can arise on call centers or web-based.


·Internet Connection
·Machine Learning knowledge
·Basic Python
·Mac/Linux system

Nature of the training

As humans, our brains are pre-trained for automatically performing simple tasks such as identifying numbers, objects and faces. We make use of language as a natural way to represent thoughts and be able to communicate. However, these tasks are far from being trivial from a computer point of view.

During the training we will learn how to design a system which integrates both capabilities: vision and natural language interaction. Finally, we will put them together and see how that works. The training will include both a theoretical part and code fragment implementations.

Designing a facial recognition system involves a pipeline of actions, such as detecting every face, identifying the key features that make them unique and identify them. For a real-time video input, we will also need to adapt to the movement, as faces might change, move or be hidden. We will address those steps during the first part of the training.

On the second part of the training we will build a conversational assistant. Such an assistant involves understanding, smoothness, memory and conversation guidance for a limited domain. We will address those pars one by one, making use of NLU tools on our way.


Big Data Spain will issue the certificate for this course to prove subject matter competency


Developers, Data Scientists. Mainly technical people, with interests on Artificial Intelligence, NLU and Computer Vision

Bio of the instructor - Fernando Velasco

Fernando Velasco Lozano is a Data Scientist at Stratio, enthusiastic on mathematical modelling, data and innovation. His experience includes academic research in the Algebraic Geometry area, and his focus and passion: Data Science. His area of expertise includes Deep Learning and Behavioral Algorithms.