My Scientific Challenges for 2016-2022

A.

Study a data classifier with ability to deliver some form of additional information and not just a firm decision results. The attempts were MF ARTMAP using Fuzzy Sets and Fuzzy Logic and Accumulated Fuzzy Sets Criterion for classification. It seems to have something to do with explainable AI as pointed out by prof. Holzinger

Challenges:

How to handel Dynamic Feature space?

What form should have Semantic forms of explanation?

What is a Human User oriented classifiers?

Study a challenge IF we want to create a strong classifier for any kind of data 

or 

more narrow for particular kind of data sets ?

Expected scientific output:

New form of Classification tool for Humans

B. 

Study a new and faster punishment/reward system using Reinforcement Learning concept with crowdsourcing environment for with faster learning for selected application in Social Robotics. The concept of Virtual Robot as a digital twin for multiple physical robot is under extensive study and investigation. 

The basic concept of COWOZ (Cloud based Wizard of Oz) is a key element for further study and design of the knowledge form in incremental form is a major question for a future. This research is envisioning unlimited computer and storage power and unlimited (physically limited) WIFI internet connection.

Challenges:

What will be a structure of the Virtual Robot ?

What will be a innovative and faster Reward learning concept for 

Virtual Robot and Integrated Knowledge Based in crowdsourcing environment?

How to create a behavioral model of the Human in Human Social Robot interaction?

How to create a behavioral model of Wizard for COWOZ ?

Come up with selected application for Virtual Robot and Cowoz in Social Robotics

Expected scientific output:

New COWOZ concept with learning ability for HSR-I 

 

C

Within HSR-I special Study with special Focus an Emotion/Empathy issues. I would like to investigate if all existing commercial tools for emotion assessment and explicitly made clear if accuracy of assessment is equal no each of assessed emotions from FACE using image recognition. 

Study a notion of Human mood in HSR-I as time line of emotions of the human. Results of these scientific goal is important for getting information from environment and partially multilateral information to judge a Human emotion and mood during HSR-I

 

 

Also concepts of internal and external synthetic Robot emotions and their impact to human emotions in the HUMAN-Social ROBOT LOOP are under the interest of affective computing and empathic computing. 

These forms of syntactic emotions will be influences by number of human emotional models including Ekman and other models. Human emotions as reactions to synthetic Social Robot emotions will be under investigation and personalization.

 

Challenges:

How to measure an accuracy of Emotion assessment?

How to determine a human mood based on timeline of human emotions?

How to enrich Facial recognition to make more accurate emotion recognition of humans?

 

How to define a synthetic Robot Emotions inspired by human emotional models?

How to make personalization of Human Social Robot interaction based on empathy?

Can a Virtual Robot have Emotional Status and how it can be connected real Robots in interaction?

 

                 Expected scientific output:

Novel tool for emotion assessment accuracy

Definition of human mood in Human Social Robot Interaction

Definition of theoretical background for synthetic emotions for Social Robots

Definition of Emotional “life” of Virtual Robot in personalization process of Human Social Robot Interactions