About: Teresa Escrig - Managing Director - AI Global lead
Dr. Teresa Escrig is an AI expert with over 25 years of distinguished accomplishments in the field. She has been a professor on AI related topics at several universities in Europe and the US, founded a couple of AI startups and is an AI global lead at a large consulting corporation. A visionary, out-of-the-box thinker, passionate and effective leader and speaker, Dr. Escrig’s mission is to make sure that we develop transparent, unbiased, safe and responsible AI.
1. Please explain your expertise in Artificial Intelligence and Robotics research?
I have over two decades of distinguished accomplishments in the AI field. I’ve hold a University tenure position and have taught AI related topics at several universities in Europe and the U.S. I lead a university research group on Cognitive Robotics, published more than one hundred peer-review articles, three books, four patents, and received several awards including the best PhD award and the National Prize on Science and Technology. I’ve been founder and CEO of two startups in AI – “Cognitive Robots” and “Qualitative Artificial Intelligence” and lead creator of four products from concept to delivery in diverse areas of application “Cognitive Robot Navigation”, “Cognitive Vision System”, “DNA sequencing”, and “Transparent Autonomous Vehicles”. I hold now the position of AI global lead in Cognitive Computing & Computer Vision at Accenture Technology. Recently, I’ve created the Decalogue of Responsible AI enterprise to harness the benefits of AI without the unintended consequences.
2. What is the role of Artificial Intelligence (AI) and Cognitive Vision System in delivering the best customer experience?
The best customer experience comes from smart technologies that know our needs as they are happening and are proactively offering solutions to enhance our productivity, not only at work but also in daily tasks, like for instance, tapping into our TO-DO list and organizing the best way to accomplish our errands.
3. Please explain how Artificial Intelligence is being applied in cyber security?
I was involved for one year in cyber security research at U. Washington (University of Washington) a few years ago, and I discovered that they were making the same type of mistakes than the researchers on Robotics, i.e. applying only ML algorithms, with no significant results. I wrote a couple of research papers on how to use a more cognitive approach, using ontologies, to understand and visualize known attacks and have a sandbox to test unknown attacks.
4. Can you share the list of books you pen-down and also, share your experience on receiving numerous awards?
List of books:
- Escrig, T., Toledo, F., Qualitative Spatial Reasoning: Theory and Practice. Application to Robot Navigation, IOS Press, Frontiers in Artificial Intelligence and Applications, ISBN 90 5199 4125, 1998.
- Escrig, T., Robots that Reason (Spanish), April 2005. National Prize on Science and Technology 2004.
- Escrig, T., How to program in Prolog, students book at Jaume I university.
I’ve received several awards:
- PAAMS Award of Scientific Excellence, 2012
- National Prize on Science and Technology, 2004
- Best PhD Award, 1997/98
- 3 best paper awards at different conferences
5. Kindly enlighten our readers on your brainchild products “Cognitive Brain for Service Robotics” and the “Cognitive Vision System”.
The “Cognitive Brain” incorporates four aspects of human intelligence: 1) perception (object recognition), 2) reasoning, 3) learning and 4) decision-making, by formalizing commonsense reasoning with Qualitative Modeling, which is transparent, do not require massive amounts of data, provide insights from the first data sample, and adapts to new environments.
The competitive advantages of the Cognitive Brain are:
PORTABILITY - The same Cognitive Brain can be installed, with minor changes, into any service vehicle or robot to transform it into a completely intelligent, autonomous vehicle or robot. No need to be retrained.
INTELLIGENT PERCEPTION - Our technology performs sensor integration and extracts the most significant information from the environment, assigning names and concepts to objects.
MAP BUILDING - Our technology automatically creates the map of any new environment, as well as the ability to adapt itself to changes in a known environment.
HIGH-LEVEL DECISION MAKING - With intelligent aspects of perception, reasoning, and learning incorporated into our Cognitive Brain, the system is able to make proactive decisions, including when to start an activity and what to do with unexpected obstacles.
MONITORING - The position and performance of any autonomous vehicle or robot can be monitored and controlled remotely with any smart device.
The Intelligent Perception of the Cognitive Brain is obtained with the “Cognitive Vision System”, which also contains several Qualitative Models to distinguish shape, color, topological relationships of the objects within the image, and relative and absolute orientation. From any image containing thousands of pixels with no semantics, we are able to obtain dozens of tags with meaning. That is then stored in an ontology of objects. It is an automatic tagging of objects in images, recognizing features of objects, much like people do, using common sense.
6. How does Cognitive Brain help in transforming any vehicle into an autonomous and intelligent vehicle?
It is the brain with the minimum hardware possible (computer, sensors and bus can), ready to be installed in any vehicle.
7. What is your research roadmap on Cognitive Vision System? Kindly please share your insights.
We have seen that using an automatic tagging before using ML algorithms reduces the time, and therefore the cost, of algorithm development up to 70%. The roadmap is to use this automatic tagging of any images stored in ontologies and integrate it with ML algorithms. We have coined the term Holistic AI as the integration of Knowledge Representation and Reasoning (which includes Qualitative Modeling, Ontologies and Data Graphs) with Machine Learning. Holistic AI is the first pillar of Responsible AI. We have used this concept to develop a “Transparent Autonomous Vehicle” solution, which includes the following benefits: increase trust in all stakeholders and reduces time and cost of algorithm development.