Ana-Maria Olteteanu
I work at the intersection of artificial intelligence and cognitive science, trying to understand how humans solve various tasks and build computational approaches which enable artificial intelligence systems to solve similar classes of tasks.
My background is interdisciplinary. I was a pianist in my childhood and teen years, I graduated from a Bachelor in Musical Performance. I was always curious about how the human mind works, and good at maths. That curiosity re-emerged while undergoing my first PhD thesis, in musicology, which analysed the links between musical form and musical expression, with cognitive underpinnings. At the same time I took a MSc course in Cognitive Computing, which paved the way to my second PhD thesis in Cognitive Systems, at Universitat Bremen, where I now work as a postdoc. My current focus is on creative problem solving.
As part of my current work, I use various computational tools to make cognitive systems which can answer creative problem solving tasks and/or help investigate how humans solve such tasks. These systems are designed under a theoretical framework of creative problem solving (CreaCogs), e.g.:
- comRAT-C is a system that can solve the compound Remote Associates Test
- OROC is a prototype which can do creative object replacement and object composition
- comRAT-V can solve a visual variant of the Remote Associates Test which we created
- comRAT-G creates Remote Associates Test items (17 million items created from noun terms alone).
Other approaches aim to build the way to computationally understanding re-representation, classes of creative problem solving, etc.
I also design and perform empirical investigations (psychological experiments) to understand how humans solve such tasks and evaluate the computational cognitive systems built.
For a more complete publication list see https://www.researchgate.net/profile/Ana-Maria_Olteteanu
For more details on the projects see http://creacogcomp.com/
https://www.researchgate.net/profile/Ana-Maria_Olteteanu/contributions
My background is interdisciplinary. I was a pianist in my childhood and teen years, I graduated from a Bachelor in Musical Performance. I was always curious about how the human mind works, and good at maths. That curiosity re-emerged while undergoing my first PhD thesis, in musicology, which analysed the links between musical form and musical expression, with cognitive underpinnings. At the same time I took a MSc course in Cognitive Computing, which paved the way to my second PhD thesis in Cognitive Systems, at Universitat Bremen, where I now work as a postdoc. My current focus is on creative problem solving.
As part of my current work, I use various computational tools to make cognitive systems which can answer creative problem solving tasks and/or help investigate how humans solve such tasks. These systems are designed under a theoretical framework of creative problem solving (CreaCogs), e.g.:
- comRAT-C is a system that can solve the compound Remote Associates Test
- OROC is a prototype which can do creative object replacement and object composition
- comRAT-V can solve a visual variant of the Remote Associates Test which we created
- comRAT-G creates Remote Associates Test items (17 million items created from noun terms alone).
Other approaches aim to build the way to computationally understanding re-representation, classes of creative problem solving, etc.
I also design and perform empirical investigations (psychological experiments) to understand how humans solve such tasks and evaluate the computational cognitive systems built.
For a more complete publication list see https://www.researchgate.net/profile/Ana-Maria_Olteteanu
For more details on the projects see http://creacogcomp.com/
https://www.researchgate.net/profile/Ana-Maria_Olteteanu/contributions
less
InterestsView All (9)
Uploads
Journal Articles by Ana-Maria Olteteanu
normative data on 144 items provided by Bowden and Jung-Beeman. comRAT is a computational solver which has been used to solve the compound RAT in linguistic and visual forms, showing correlation to human performance over the normative data provided by Bowden and Jung-Beeman. This paper describes using a variant of comRAT, comRAT-G, to generate and construct a repository of compound RAT items for use in the cognitive psychology and cognitive modelling community. Around 17 million compound Remote Associates Test items are created from nouns alone, aiming to provide control over (i) frequency of occurrence of query items, (ii) answer items, (iii) the probability of coming up with an answer, (iv) keeping one or more query items constant and (v) keeping the answer constant. Queries produced by comRAT-G are evaluated in a study in comparison with queries from the normative dataset of Bowden and Jung-Beeman, showing that comRAT-G queries are similar to the established query set.
topology and location are used for describing any object in the scene. Two kinds of domain knowledge are provided: (i) categorizations of objects according to their qualitative descriptors, and (ii) semantics for describing the affordances, mobility and other functional properties of target objects. First order logics are obtained for reasoning and scene understanding. Tests were carried out at the Interact@Cartesium scenario and promising results were obtained.
Conference Papers by Ana-Maria Olteteanu
normative data on 144 items provided by Bowden and Jung-Beeman. comRAT is a computational solver which has been used to solve the compound RAT in linguistic and visual forms, showing correlation to human performance over the normative data provided by Bowden and Jung-Beeman. This paper describes using a variant of comRAT, comRAT-G, to generate and construct a repository of compound RAT items for use in the cognitive psychology and cognitive modelling community. Around 17 million compound Remote Associates Test items are created from nouns alone, aiming to provide control over (i) frequency of occurrence of query items, (ii) answer items, (iii) the probability of coming up with an answer, (iv) keeping one or more query items constant and (v) keeping the answer constant. Queries produced by comRAT-G are evaluated in a study in comparison with queries from the normative dataset of Bowden and Jung-Beeman, showing that comRAT-G queries are similar to the established query set.
topology and location are used for describing any object in the scene. Two kinds of domain knowledge are provided: (i) categorizations of objects according to their qualitative descriptors, and (ii) semantics for describing the affordances, mobility and other functional properties of target objects. First order logics are obtained for reasoning and scene understanding. Tests were carried out at the Interact@Cartesium scenario and promising results were obtained.
extracted from the Corpus of Contemporary American English. The results provided by the computational RAT are compared to those obtained in previous RATs carried out on human
participants in order to hypothesize on an associationist creative cognition paradigm.