One of the key challenges in artificial intelligence is the symbol grounding problem and the creation of "embodied" architectures for intelligent agent control.
In this direction, we pay the most attention to solving the VQA problem - answering questions about the image, with which we can demonstrate the capabilities of our iconic architecture. The very task of VQA is also very important in practice, allowing you to simulate the processes of understanding images.
Выступление на конференции AGI-2020
Доклад А.К. Ковалева на тему "Hyperdimensional Representations in Semiotic Approach to AGI"
Gupta N. et al. Neural module networks for reasoning over text // ICLR 2020. 2020. P. 1–17.
Dalu Guo et al. Bilinear Graph Networks for Visual Question Answering ArXiv
Kovalev A.K., Panov A.I., Osipov E. Hyperdimensional Representations in Semiotic Approach to AGI // Artificial General Intelligence. AGI 2020. Lecture Notes in Computer Science. Springer, 2020. Vol. 12177. P. 231–241. Springer
Kovalev A.K., Panov A.I. Mental Actions and Modelling of Reasoning in Semiotic Approach to AGI // Artificial General Intelligence. AGI 2019. Lecture Notes in Computer Science / ed. Hammer P. et al. Springer, 2019. Vol. 11654. P. 121–131. Springer