Curiosity Cloning - Neural Modelling for Image Analysis
Is it possible to detect the scientific curiosity from human brain activity? If so, can we then teach to a computer to exhibit a similar behaviour?
Imagine a planetary rover having a vivid scientific curiosity and thus autonomously selecting the scientifically valuable data that are worth to be analysed by scientists back on Earth. Imagine such a rover to be programmed to exhibit curiosity patterns cloning those of particular human beings. Back on Earth, we would select from a menu the name of some famous scientist and upload his ‘curiosity module’ on board the rover. The rover would then exhibit naturally inquisitive behaviour by imitating the selected scientist’s scientific curiosity when selecting the experiments to be performed or the direction to explore.
Curiosity Cloning - Neural Modelling for Image Analysis study aims to be the first step towards realising this daring vision. In order to create artificial models of human curiosity, a large amount of training data will be undoubtely required. Such training data would consist of sample images associated with information about the amount of scientific interest they present to the subject. However, it is rather evident that presenting large numbers of images to someone and asking him to express his opinion is too troublesome and time-consuming. Is there a way to tackle this problem?
A possible solution may be to directly monitor the subject’s brain activity during a fast visual presentation of images and classify the subject’s brain waves. Starting from the assumption that a higher attention level stimulated by curiosity can be detected in the P300 brain wave, this study aims at studying the following questions:
- Up to what image presentation speed is the detection and quantitative analysis of brain waves reliable?
- Can we discriminate between different types of curiosity?
- Can we use the P300 to build a probabilistic ranking of the most interesting images as percieved by the subject?
Experiment I (Reliability vs. Speed)
In this first experiment we tried to see how the P300 brain wave detection performance degrades with the image presentation speed increasing. We performed the same experiment using inter image display periods of 500, 300, 150, 100 and 50 ms and evaluated the performance of the classification scheme adopted.
Experiment II (Beyond the Oddball paradigm)
In the second experiment we tried to demonstrate that one can detect more than two classes of brain waves associated to different visual stimuli: oddball, non-oddball and non-obvious oddball.
Experiment III (Scientific Curiosity)
The third experiment was aimed at building a ranking of images according to the subject interest. The underlying assumption here was that a P300 brain wave classifier (trained on whatever data) can later be used to detect curiosity or interest levels.
Final DEMO at ESTEC
In order to meet the needs of experiments conducted during the study, which involve display of the visual stimuli, the Advanced Concepts Team has developed a high-fidelity image display software, named Curiosity Cloning Image Viewer. The program takes advantage of modern multimedia technologies to achieve unprecedented display timing accuracy. The software is freely available to everyone, as it has been released as an open-source project, under the BSD license. If you would like to obtain either the source code or a binary release or the User’s Manual, you will find them under the following addresses:
The study was proposed under the Ariadna Call 2008/01 and was carried out by the Advanced Concepts Team in cooperation with the Multimedia Signal Processing Group of the Swiss Federal Institute of Technology and the School of Computing of the Dublin City University. Following is the list of the researchers participating in the study.
- Dario Izzo
- Luca Rossini
- Christos Ampatzis
- Marek Ruciński
- Touradj Ebrahimi
- Ashkan Yazdani
- Alan Francis Smeaton
- Peter Wilkins
- Graham Healy
- Izzo, D., Rossini, L., Rucinski, M., Ampatzis, C., Healy, G., Wilkins, P., Smeaton, A.F., Yazdani, A., and Ebrahimi, T., On the EEG footprint of image saliency, Internal report of the project, 2008. (link)
- Izzo, D., Rossini, L., Rucinski, M., Ampatzis, C., Healy, G., Wilkins, P., Smeaton, A.F., Yazdani, A., and Ebrahimi, T., Curiosity CLoning: neural analysis of scientific interest, Proceedings of the International Joint Conference on Artificial Intelligence 2009, Workshop on Artificial Intelligence in Space, 2009. (link)
- Smeaton, A.F., Wilkins, P., Healy, G., Ampatzis, C., Rucinski, M., and Izzo, D., Neurological Modeling of What Experts vs. Non-Experts Find Interesting, Neuroscience 2009: session number 687, Computation, Modeling, and Simulation III, 2009. (link)
Healy, G., P. Wilkins, A.F. Smeaton, D. Izzo, M. Rucinski, C. Ampatzis, and E.M. Moraud. 2010. “Curiosity Cloning: Neural Modelling for Image Analysis.” 08-8201b European Space Agency, the Advanced Concepts Team. [link]
Yazdani, A., F. Dufaux, T.M. Ha, T. Ebrahimi, D. Izzo, and C. Ampatzis. 2010. “Curiosity Cloning: Neural Modelling for Image Analysis.” 08-8201a European Space Agency, the Advanced Concepts Team. [link]
Izzo, D., L. Rossini, M. Rucinski, C. Ampatzis, G. Healy, P. Wilkins, A.F. Smeaton, A. Yazdani, and T. Ebrahimi. 2009. “Curiosity CLoning: Neural Analysis of Scientific Interest.” In Proceedings of the International Joint Conference on Artificial Intelligence 2009, Workshop on Artificial Intelligence in Space. [link]
Rucinski, M. 2008. “Curiosity Cloning Image Viewer User’s Manual.” CCIVUM01 European Space Agency, the Advanced Concepts Team. [link]