Abstract
Retina Implants for blind subjects with retinal degenerations are under development to regain a modest amount of vision. A spatiotemporal Retina Encoder (RE) simulates retinal information processing in real time and generates pulse signals for electrical stimulation of retinal ganglion cells (GC). For each GC channel, RE must be optimized with regard to the patient's perception. For implementation of a learning RE, we applied reinforcement learning (RL) based on evaluative feedback in a suitable training environment, which can be tested in subjects with normal vision for future use with blind subjects. We demonstrate successful learning on a hardware retina encoder.
Supported by Federal Ministry for Education, Science, -Research, and Technology (BMBF).
Preview
Unable to display preview. Download preview PDF.
References
Barto GA, Sutton RS, Anderson CW (1983) Neuronlike elements that can solve difficult learning control problems. IEEE Transactions on Systems, Man and Cybernetics 13:835–846
Becker M, Hünermann, R., Eckmiller E (1997) System identification of a learning retina encoder for a retina implant. Invest Ophthal & Vis Sci (Suppl.)38:S41
Eckmiller R, Becker M, Hünermann R (1997) Dialog concepts for learning retina encoders. Proc. ICNN'97, Houston, Texas, June 1997
Fleet DJ, Hallett PE, Jepson AD (1985) Spatiotemporal Inseparability in Early Visual Processing. Biol Cybern 52:153–164
Hünermann R, Becker M, Eckmiller R (1997) Real time implementation of a tunable retina encoder. Invest Ophthal & Vis Sci (Suppl.)38:S41
Humayun MS, de Juan E, Dagnelie G, Greenberg R, Probst R (1996) Artificial Vision. Invest Ophthal & Vis Sci (Suppl.)37:S451
Klein M (1996) Development and test of a retina encoder hardware system with two adaptive receptive field channels for retina implants. (in German) Masters Thesis, Universität Bonn, November 1996
Lee BB, Pokorny J, Smith VC, Kremers J (1994) Responses to pulses and sinusoids in Macaque ganglion cells. Vision Res 34:3081–3096
Massof RF, Finkelstein D (1987) A two-stage hypothesis for the natural course of Retinitis Pigmentosa. Adv in the Biosci 62:29–58
Michie D, Chambers' RA' (1968) BOXES: An experiment in adaptive control. Machine Intelligence 2, Dale and Michie, editors, pp137–152, Oliver and Boyd.
Minsky ML (1961) Steps towards artificial intelligence. Proc. of the Inst. of Radio Eng. 49:8–30
Richter J, Ullman S (1982) A Model for the temporal organization of X-and Y-type receptive fields in the primate retina. Biol Cybern 43:127–145
Sutton RS (1988) Learning to predict by the methods of temporal difference. Machine Learning 3:9–44
Sutton RS (1996) Reinforcement learning with replacing eligibility traces. Machine Learning 22:123–158
Sutton RS (1997) On the significance of Markov decision processes. Proceedings of ICANN 97
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1997 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Becker, M., Eckmiller, R. (1997). Spatio-temporal filter adjustment from evaluative feedback for a retina implant. In: Gerstner, W., Germond, A., Hasler, M., Nicoud, JD. (eds) Artificial Neural Networks — ICANN'97. ICANN 1997. Lecture Notes in Computer Science, vol 1327. Springer, Berlin, Heidelberg. https://doi.org/10.1007/BFb0020311
Download citation
DOI: https://doi.org/10.1007/BFb0020311
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-63631-1
Online ISBN: 978-3-540-69620-9
eBook Packages: Springer Book Archive