Showing 1–6 of 6 results for author: Bak, P

Intelligent systems in the context of surrounding environment

Authors: Joseph Wakeling, Per Bak

Abstract: We investigate the behavioral patterns of a population of agents, each controlled by a simple biologically motivated neural network model, when they are set in competition against each other in the Minority Model of Challet and Zhang. We explore the effects of changing agent characteristics, demonstrating that crowding behavior takes place among agents of similar memory, and show how this allows… ▽ More We investigate the behavioral patterns of a population of agents, each controlled by a simple biologically motivated neural network model, when they are set in competition against each other in the Minority Model of Challet and Zhang. We explore the effects of changing agent characteristics, demonstrating that crowding behavior takes place among agents of similar memory, and show how this allows unique `rogue' agents with higher memory values to take advantage of a majority population. We also show that agents' analytic capability is largely determined by the size of the intermediary layer of neurons. In the context of these results, we discuss the general nature of natural and artificial intelligence systems, and suggest intelligence only exists in the context of the surrounding environment (embodiment). Source code for the programs used can be found at http://neuro.webdrake.net . △ Less

Submitted 24 January, 2002; originally announced January 2002.

Comments: 9 pages, 7 figures; should be listed as (c) American Physical Society 2001 if possible

Journal ref: Physical Review E 64 (2001) 051920

Adaptive learning by extremal dynamics and negative feedback

Authors: Per Bak, Dante R Chialvo

Abstract: We describe a mechanism for biological learning and adaptation based on two simple principles: (I) Neuronal activity propagates only through the network's strongest synaptic connections (extremal dynamics), and (II) The strengths of active synapses are reduced if mistakes are made, otherwise no changes occur (negative feedback). The balancing of those two tendencies typically shapes a synaptic l… ▽ More We describe a mechanism for biological learning and adaptation based on two simple principles: (I) Neuronal activity propagates only through the network's strongest synaptic connections (extremal dynamics), and (II) The strengths of active synapses are reduced if mistakes are made, otherwise no changes occur (negative feedback). The balancing of those two tendencies typically shapes a synaptic landscape with configurations which are barely stable, and therefore highly flexible. This allows for swift adaptation to new situations. Recollection of past successes is achieved by punishing synapses which have once participated in activity associated with successful outputs much less than neurons that have never been successful. Despite its simplicity, the model can readily learn to solve complicated nonlinear tasks, even in the presence of noise. In particular, the learning time for the benchmark parity problem scales algebraically with the problem size N, with an exponent $k\sim 1.4$. △ Less

Submitted 14 September, 2000; originally announced September 2000.

Self-Organization of Complex Systems

Authors: Maya Paczuski, Per Bak

Abstract: The basic laws of physics are simple, so why is the world complex? The theory of self-organized criticality posits that complex behavior in nature emerges from the dynamics of extended, dissipative systems that evolve through a sequence of meta-stable states into a critical state, with long range spatial and temporal correlations. Minor disturbances lead to intermittent events of all sizes. Thes… ▽ More The basic laws of physics are simple, so why is the world complex? The theory of self-organized criticality posits that complex behavior in nature emerges from the dynamics of extended, dissipative systems that evolve through a sequence of meta-stable states into a critical state, with long range spatial and temporal correlations. Minor disturbances lead to intermittent events of all sizes. These events organize the system into a complex state that cannot be reduced to a few degrees of freedom. This type of ``punctuated equilibrium'' dynamics has been observed in astrophysical, geophysical, and biological processes, as well as in human social activity. △ Less

Submitted 5 June, 1999; originally announced June 1999.

Comments: 6 figures, Proceedings of 12th Chris Engelbrecht Summer School, Figs. 1 and 2 missing

Learning from Mistakes

Authors: Dante R. Chialvo, Per Bak

Abstract: A simple model of self-organised learning with no classical (Hebbian) reinforcement is presented. Synaptic connections involved in mistakes are depressed. The model operates at a highly adaptive, probably critical, state reached by extremal dynamics similar to that of recent evolution models. Thus, one might think of the mechanism as synaptic Darwinism. A simple model of self-organised learning with no classical (Hebbian) reinforcement is presented. Synaptic connections involved in mistakes are depressed. The model operates at a highly adaptive, probably critical, state reached by extremal dynamics similar to that of recent evolution models. Thus, one might think of the mechanism as synaptic Darwinism. △ Less

Submitted 28 July, 1997; originally announced July 1997.

Comments: 12 pages, Latex2e

Mass Extinctions vs. Uniformitarianism in Biological Evolution

Authors: Per Bak, Maya Paczuski

Abstract: It is usually believed that Darwin's theory leads to a smooth gradual evolution, so that mass extinctions must be caused by external shocks. However, it has recently been argued that mass extinctions arise from the intrinsic dynamics of Darwinian evolution. Species become extinct when swept by intermittent avalanches propagating through the global ecology. These ideas are made concrete through s… ▽ More It is usually believed that Darwin's theory leads to a smooth gradual evolution, so that mass extinctions must be caused by external shocks. However, it has recently been argued that mass extinctions arise from the intrinsic dynamics of Darwinian evolution. Species become extinct when swept by intermittent avalanches propagating through the global ecology. These ideas are made concrete through studies of simple mathematical models of coevolving species. The models exhibit self-organized criticality and describe some general features of the extinction pattern in the fossil record. △ Less

Submitted 2 February, 1996; originally announced February 1996.

Comments: 17 pages uuencoded with style file lamuphys.sty. 9 figures not included but can be obtained via [email protected]. to appear in ``Physics of Biological Systems'' Lecture Notes in Physics (Springer-Verlag, Heidelberg , 1996)

Democratic Reinforcement: Learning via Self-Organization

Authors: Dimitris Stassinopoulos, Per Bak

Abstract: The problem of learning in the absence of external intelligence is discussed in the context of a simple model. The model consists of a set of randomly connected, or layered integrate-and fire neurons. Inputs to and outputs from the environment are connected randomly to subsets of neurons. The connections between firing neurons are strengthened or weakened according to whether the action is succe… ▽ More The problem of learning in the absence of external intelligence is discussed in the context of a simple model. The model consists of a set of randomly connected, or layered integrate-and fire neurons. Inputs to and outputs from the environment are connected randomly to subsets of neurons. The connections between firing neurons are strengthened or weakened according to whether the action is successful or not. The model departs from the traditional gradient-descent based approaches to learning by operating at a highly susceptible ``critical'' state, with low activity and sparse connections between firing neurons. Quantitative studies on the performance of our model in a simple association task show that by tuning our system close to this critical state we can obtain dramatic gains in performance. △ Less

Submitted 23 January, 1996; originally announced January 1996.

Comments: 9 pages (TeX), 3 figures supllied on request