Prof. Meir's research interests include:
- Dynamics and Computation in Biological Networks
- Reinforcement Learning in Natural and Artificial Systems
- Neural Encoding and Decoding and Multisensory Integration
- Computational Motor Control
web site: http://www.ee.technion.ac.il/~rmeir
Ron Meir - Research Statement
Organisms in the natural world, evolving over many millions of years, display remarkable abilities of adaptation, decision making and action selection in uncertain, and often hostile, environments. The means by which they achieve the impressive performance levels they do is poorly understood, although there is little doubt that much of this is achieved through evolution and learning, occurring at the cellular, individual and cultural levels. In contrast to the natural world, the performance of currently existing engineered systems in such tasks is brittle and inflexible. Over fifty years of research in artificial intelligence have produced little in the way of creative, robust, and flexible solutions, hallmarks of the biological world. While biological systems differ significantly from engineered systems, they share many common goals. Think of building a robot to clean an unknown building effectively, and the difficult perceptual, motor, and decision making tasks required. These problems are very much inline with those facing a human attempting to face this task.
While engineering practice cannot at this stage match the performance of biological systems, it does provide a solid theoretical foundation in domains such as learning, decision making, signal processing and control. Clearly, these theories were developed with an eye to the 'artificial world', but my basic premise is that they have important things to tell us about the biological realm, even if they require significant extensions and modifications. Since our goal here is to reverse-engineer a non-engineered system, we should expect the unexpected!
The long term goal driving my research is the creation of a conceptual and mathematical framework which will aid us in understanding perception, decision making, action selection and motor performance in the biological world. This framework will be based on ideas from Physics and Engineering, and will be constrained by relevant biological data. This type of work involves serious cooperation between experimentalists and theoreticians, in order to guarantee that theories developed are relevant to real life, and do not exist only in the theoretician's dream world. As such, we envisage studying experimental systems, and at the same time synthesizing artificial and/or hybrid systems, which aim to mimic the real world.
For a description of four current research directions see http://www.ee.technion.ac.il/~rmeir/ResearchInterests.html
A list of selected publications can be found at http://www.ee.technion.ac.il/~rmeir/rmeir_publications.html