Marko Puljic; Interests
A system is a collection of many components. Global states show the general mood of the components: mostly active or mostly inactive, mostly synchronized or mostly de-synchronized, and so on. In dynamical systems states jump from one to another quickly through the sudden changes or state transitions. One wonders what happens during and before the moments of sudden changes. Brains of animals jump from one thought to another and make decisions in complex environment more successfully then machines. But machines can be successful as well, if they use brain-like computational principles (Freeman, 1975). Humans and animals can solve difficult identification tasks fast and with high accuracy, even in the noisy environments. Odor identification is based on a few sniffs; visual filtering on scanning a given image a few times, etc. In the universe as the punctuated equilibrium system, many similar parts of system interact in time, affecting the behavior of those parts and the behavior of their interactions, and thus creating the emergent system's dynamics. Examples can be seen in a collective behavior of mass particles in a gravitational field, rapid climate transitions on earth in the past few million years, short term weather patterns, geological landscape formation, crystallization, magnetic and ferroelectric domains in solids, biochemical synthesis from cell culture, traffic flow, technological evolution in the marketplace, and so on. Out of all systems, the brain seems to be the most amazing. Are there and if so what are the similarities between the brain state transitions and the state transitions in other systems?
Brain-like computing operates through chaotic dynamical system or chaotic dynamical memories. Its principles are different from successful digital and analog computers, which adopted the 1940s Turing machine (TM) model of computation. TM incorporates the information on a tape into a machine by means of symbols. Operations are interpreted as the manipulation of the symbols according to certain semantic rules. At present, many are accustomed to the Turing machine workings as a metaphor for brain functioning. To them, the brain's job is to incorporate features of the outside world and to make internal syntactical representations of these data, which constitute a world model that serves to control the output. View of chaotic dynamical memories assumes that memories do not emerge only from stored information, but from perception and cognition as well.
Discrete model based on chaotic dynamical memories and cellular automaton is Neuropercolation model. Neuropercolation model is governed by probabilistic processes rather than by differential equations. With appropriate habituation and Hebbian learning rules, they are pattern recognition devices. An added advantage of the probabilistic approach is inherent noise, so no external noise source is required to stabilize the system. To compute by principles of chaotic itinerancy Neuropercolation needs to manifest following properties of brain spatio-temporal dynamics: state transition of an excitatory population from a point attractor with zero activity to a non-zero point attractor with steady-state activity by positive feedback; emergence of oscillations through negative feedback between excitatory and inhibitory neural populations; state transitions from a point attractor to a limit cycle attractor that regulates steady-state oscillation of a mixed excitatory-inhibitory cortical population; genesis of chaos as background activity by combined negative and positive feedback among three or more mixed excitatory-inhibitory populations; distributed wave of chaotic dendritic activity that carries a spatial pattern of amplitude modulation made by the local heights of the wave; the increase of non-linear feedback gain that is driven by input to mixed population, which results in construction of an amplitude-modulation pattern as the first step in perception; the embodiment of meaning in amplitude-modulation patterns of neural activity, which are shaped by synaptic interactions that have been modified through learning; attenuation of microscopic sensory-driven activity and enhancement of macroscopic amplitude modulation patterns by divergent convergent cortical projections underlying solipsism; the divergence of corollary discharges in pre-afference followed by multi-sensory convergence into entorhinal cortex as the basis of Gestalt formation; the formation of a sequence of global amplitude-modulation patterns of chaotic activity that integrates and directs the intentional state of an entire hemisphere.
Lab at University of Memphis
People we work with
Department of Computer Science
807 Walker Avenue, GOH 107D
Memphis, Tennessee 38126
phone: (901) 435-1399
Doctor of Philosophy in Computer Science, University of Memphis, 2005
Master of Science in Finance, Fogelman College of Business and Economics, 1998
Bachelor of Science in Mathematics/Economics, University of Zagreb, 1996
Computer Science Professor; 2005-present
Teaching and organizing computer science courses such as computer networks, computer organization and administration, data security and structures, microcomputers, AI, and UNIX OS.
University of Tennessee
Data Analyst; 2005-2007
Mining data obtained from the researchers of theoretical and computational methods in genetics and molecular biology to find relationships between identified genes.
University of Memphis
Assistant Researcher; 2002-2005
Simulating a neuropil to study brain dynamics and describing complex systems in order to understand the emergence caused by the interactions of many components.
Analyst, U.S. Department of Energy DOE PN 02-13 2005-2007
Principal Investigator: Dr. Derek Lovley
Analysis of the genetic potential and gene expression of microbial communities involved in the in situ bioremediation of uranium and
harvesting electrical energy from organic matter.
Researcher, NSF Grant EIA-01-30352 2002-2005
Principal Investigator: Dr. Robert Kozma
A grant for a study on dynamical behavior in percolation models related to phase transitions in the cortex during sensory information processing.
J. Krushkal, M. Puljic, L. DiDonato, R. Adkins, K. Nevin, B. Methe, Y. Bin, Genome-wide Expression Profiling in Geobacter sulfurreducens: Identification of Fur and RpoS Transcription Regulatory Sites in a relGsu Mutant, Functional & Integrative Genomics, (2007)
M. Puljic, R. Kozma, Activation Clustering in Neural and Social Networks, Complexity, vol. 10, Issue 4, pp. 42-50 (2005)
R. Kozma, M. Puljic, P. Balister, B. Bollobas, W.J. Freeman, Phase Transitions in the Neuropercolation Model of Neural Populations with Mixed Local and Non-Local Interactions, Biological Cybernetics, vol. 92, num. 6, pp. 367-379 (2005)R. Kozma, M. Puljic, P. Balister, B.
Bollobas, W.J. Freeman, Neuropercolation: A Random Cellular Automata Approach to Spatio-Temporal Neurodynamics,Lecture Notes in Computer Science, vol. 3305, pp. 435-443 (2004)