P. Dittrich and W. Banzhaf. Self-evolution in a constructive binary string system. Artificial Life, 4(2):203--220, 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....work. For example, Dittrich and Banzhaf have shown that algorithmic reaction systems (sometimes also called artificial chemistries ) are capable of selfevolution in which the components responsible for the evolutionary behavior are (only) the individuals of the population system itself (Dittrich and Banzhaf, 1998, p. 204) The Dittrich and Banzhaf system exhibits complex evolutionary dynamics without hand seeding 1 Another recent related system is SeMar (Suzuki, 2000) but, like Tierra, is not designed to solve computational problems; indeed one of the goals of the project was to produce evolutionary ....

Dittrich, P., and W. Banzhaf. 1998. Self-Evolution in a Constructive Binary String System. Artificial Life, Vol. 4, pp. 203-220.

....succeeded in adapting to their environments and have evolved various forms of life. This has been a strong motivation in that, until recently, a number of life like computational systems have been proposed and studied. Genetic algorithms [1, 2, 3, 4] DNA computing [5, 6] chemical computation [7, 8, 9, 10, 11], membrane computing [12, 13, 14] and a number of artificial life systems [15] have been classified into this category, and they have proved that the mechanisms they borrowed from biological systems are in fact efficient and useful for the purpose of computation. Among them, the author focuses ....

....proved that the mechanisms they borrowed from biological systems are in fact efficient and useful for the purpose of computation. Among them, the author focuses on chemical computation and several core memory type artificial life systems, which are relevant to this paper. Chemical computation [7, 8, 9, 10, 11] is an approach of trying to devise a computer system using a comparison between information objects and chemical molecules. A set of objects (denoted by characters or strings) are prepared in a computer memory, and a set of operations are defined between them in the imitation of chemical ....

Dittrich, P., Banzhaf, W.: Self-evolution in a constructive binary string system. Artificial Life 4 (1998) 203--220

....between objects. A similar approach was recently taken by Szuba [51] who designed a chemical reaction like system which proceeds Prolog inference. Rasmussen et al. 36, 37] devised a core memory system in which core words react with one another and change their inner codes. Banzhaf et al. [8, 9, 10, 18, 19] introduced a kind of information object that catalyzes the change of another object. They expressed the objects by binary strings and made them work not only as operands but also operators of computation. These studies 3 succeeded in inducing so called catalytic networks, in which reaction ....

....if it were extended to an analysis of evolutionary programming techniques, it might provide valuable information for SeMar. The second approach to chemical computation is represented by experimental studies by Bagley et al. 6] Fontana [20] Szuba [51] Rasmussen et al. 36, 37] Banzhaf et al. [8, 9, 10, 18, 19], and Ikegami et al. 25] These researchers implemented computational systems which include chemical reaction like operations acting on informational objects, and conducted simulations. They succeeded in organizing (auto )catalytic networks between operations or objects. Among these experiments, ....

Dittrich, Peter and Wolfgang Banzhaf, "Self-evolution in a constructive binary string system", Artificial Life 4 (1998), 203--220.

....one unit can win and others die out. This is a selection process, and selection is usually considered a necessary aspect of evolution. The appearance of autocatalytic sets in artificially defined chemistries has also been investigated by Farmer, Kauffman Packard (1986) Bagley Farmer (1992) Dittrich Banzhaf (1998), and Huning (2000) Often the part of the reaction network where reactions keep taking place is much smaller than the whole reaction network, i.e. all reactions in the chemistry. How can this self organisation be used for association networks or other neural networks In a first approach ....

....a way to implement the rules, they need to be applied to a population. The population now contains a number of strings, see Figure 5 left. For the moment, the strings are hand coded to correspond to the symbols in the association task. The dynamics is implemented according to a reactor algorithm (Dittrich Banzhaf, 1998) that considers two partners in a collision (only for linking three letters to one word a substitution by rules with two collision partners has to be found) A pair of strings is drawn from a population at each time step, and then all rules are tried (Figure 5 middle) If a rule matches the pair ....

Dittrich, P. & Banzhaf, W. (1998). Self-Evolution in a Constructive Binary String System. Artificial Life 4(2), 203-220.

....uses these segment boundaries as xed crossover points, and an algorithm is developed for further changes of the segmentations to achieve global convergence. In contrast to standard genetic algorithms which progress from generation to generation, we adopt the viewpoint of the reactor algorithm (Dittrich Banzhaf, 1998) which permits an analysis similar to Eigen s (1971) molecular evolution model. From this viewpoint, we consider employing cross over at every time step, and present xed point analysis and phase portraits of the competitive dynamics. A problem is that second order interactions like cross over ....

....string. In contrast to cross over, point mutations lead to rst order replication of strings, since only one string is the source of what may be put back into the population. Second order interactions also play a role in the generation of new species in arti cial chemical reaction systems (Dittrich Banzhaf, 1998). The evolution of this system exhibits long phases of exploration as well as convergence. In the reactor algorithm, strings are sampled at random (like parents for cross over) and t results are put back into the populations at every time step. Using the reactor metaphor, we can draw on the ....

Dittrich, P. & Banzhaf, W. (1998). Self-Evolution in a Constructive Binary String System. Arti cial Life 4(2), 203-220.

....was an essential part of the organization he was interesting in observing. It is also possible to track macroscopic measures of the whole system which might be expected to give some indication of interesting behaviour at the microscopic level. Dittrich and Banzhaf use this approach in their work [13]. Both of these approaches will be employed in Nidus to search for the most useful measures. 9 Relation to Other Work A growing number of people have started looking at artificial chemistries over the last five years or so, for a number of different reasons. Nidus bears some similarity to many ....

....emerge. The system can therefore not be used, at least in its present form, to model individual organisms in an evolving population. Banzhaf, Dittrich et al. BinSys etc. Wolfgang Banzhaf, Peter Dittrich and colleagues are working on a catalytic self organising reaction system of binary strings [3, 13]. Their most recent work involves the decoding of these strings as programs which determine how one string reacts with another, which is a very similar concept to the idea of E OPs in Nidus. However, the decoded binary string performs operations that determine the product of the reaction by ....

[Article contains additional citation context not shown here]

Peter Dittrich and Wolfgang Banzhaf. Self-evolution in a constructive binary string system. Preprint, University of Dortmund, Department of Computer Science, April 1998.

....n , where n might be infinite, describes all valid molecules which may appear in an AC. A vast variety of molecule definitions can be found in di#erent approaches. For example, molecules may be abstract symbols [47] character sequences [9, 48, 74, 92] lambda expressions [49] binary strings [13, 40, 129], numbers [11] hierarchical tree data structures [21, 22, 96] combinators [117] or proofs [52] A molecule s representation is often referred to as its structure and is set in contrast to its function which is given by the reaction rules R. The description of valid molecules and their structure ....

....algorithm which can be interpreted as explicit (passive) mutation. 3.4 Abstract Automata The concept of an abstract automaton or machine commonly used in computer science provides a variety of approaches to define structure function relationships. Interesting candidates are finite state machines [40] or Turing machines [66, 129] which are operating on binary data. So, it is quite natural to represent the molecules as collections of bits organized as binary strings. The state transition function of the machines can be represented as a lookup table or by a program (a sequence of commands) A ....

[Article contains additional citation context not shown here]

P. Dittrich and W. Banzhaf. Self-evolution in a constructive binary string system. Artificial Life, 4(2):203--220, 1998.

....variation operators. To this end, several schemes of how adaptive operators may be adapted are analyzed empirically. The adaptation of the adaptation operators can be achieved by adding additional evolutionary levels or by recursively applying the variation operators onto themselves [7, 14]. Here, this is operationalized by expressing genetic operators as graph programs that may undergo their own evolution, using the same methods in a hierarchical and recursive fashion. Before describing the seven variants of meta evolution that we examine the following section introduces briefly ....

Peter Dittrich and Wolfgang Banzhaf. Self-evolution in a constructive binary string system. Artificial Life, 4(2):203--220, 1998.

....independent from their instantiating structure. In order to investigate these phenomena artificial chemistries have been proven to be important constructive and analytical tools. McCaskill (1988) Banzhaf (1993b) Ikegami and Hashimoto (1995) Ehricht, Ellinger, and McCascill (1997) Dittrich and Banzhaf (1998), Breyer, Ackermann, and McCaskill (1999) and Ono and Ikegami (1999) for instance, studied different artificial chemistries, ranging from abstract automata to rewriting systems; whereas Hjelmfelt, Weinberger, and Ross (1991) Adleman (1994) Rambidi and Maximychev (1997) Segr e, Lancet, Kedem, ....

Dittrich, P. and W. Banzhaf (1998). Self-evolution in a constructive binary string system. Artificial Life 4 (2), 203--220.

....therefore replacing b 0 . Instead of going into more details here we point the reader to a precise formal specification of the automata reaction as source code, available from (Dittrich, 1997) and a discussion of self evolution in artificial chemistries based on the automata reaction in (Dittrich and Banzhaf, 1998). As a short summary the following basic properties of the automata reaction should be noted: 1 The instruction set used here is ID, MOV, SETP, TMM, TDIR, UNSETP, CPON, CPOFF, STOP, OR, EQ, EXOR, NOP, ID, AND. The resulting reaction is usually called a2 reaction 0 0.2 0.4 0.6 0.8 1 0 500 ....

....type: catalytic second order, filter condition: active replication disabled (elastic) initialization: random strings (full system seeding) The observed behaviors in the series were very diverse ranging from early stabilization with short transients to complex, oscillating dynamics (Fig. 10) In (Dittrich and Banzhaf, 1998) we have identified two different evolutionary phases. The first one is characterized by high constructive activity which creates many new strings per generation. The second evolutionary phase begins with the emergence of a very stable, self replicating core set of cooperating strings dominating ....

Dittrich, P. and Banzhaf, W. (1998). Self-evolution in a constructive binary string system. Artificial Life, in press.

....therefore replacing b 0 . Instead of going into more details here we point the reader to a precise formal specification of the automata reaction as source code, available from (Dittrich, 1997) and a discussion of self evolution in artificial chemistries based on the automata reaction in (Dittrich and Banzhaf, 1998). As a short summary the following basic properties of the automata reaction should be noted: 2 The instruction set used here is ID, MOV, SETP, TMM, TDIR, UNSETP, CPON, CPOFF, STOP, OR, EQ, EXOR, NOP, ID, AND. The resulting reaction is usually called a2 reaction 0 0.2 0.4 0.6 0.8 1 0 500 ....

....type: catalytic second order, filter condition: active replication disabled (elastic) initialization: random strings (full system seeding) The observed behaviors in the series were very diverse ranging from early stabilization with short transients to complex, oscillating dynamics (Fig. 10) In (Dittrich and Banzhaf, 1998) we have identified two different evolutionary phases. The first one is characterized by high constructive activity which creates many new strings per generation. The second evolutionary phase begins with the emergence of a very stable, self replicating core set of cooperating strings dominating ....

Dittrich, P. and Banzhaf, W. (1998). Self-evolution in a constructive binary string system. Artificial Life, in press.

....groups disperse. This contribution investigates the group formation in bounded space. 1.1 Introduction The seceder model [5] shows how the local tendency to be di erent gives rise to the formation and stable coexistence of groups. The mechanism is simple compared to other individual based models [1, 4, 6] for the formation of species or hierarchical organizations. Despite of its simplicity it shows comparably complex behavior. The seceder model does not require global energy functions [3, 6] spatially separated populations [9, 11] or sexual recombination [3, 8, 10] The model consists of a ....

Dittrich, P., and W. Banzhaf, \Self-evolution in a constructive binary string system", Articial Life 4, 2 (1998), 203-220.

....Artificial Chemistries An artificial chemistry is an artificial system, which is similar to a chemical system. Usually, an artificial chemistry consists of: 1. a set of objects S : These objects may be abstract symbols [16] character sequences [1, 12, 14] lambda expressions [8] binary strings [3, 6, 15], numbers [4] or proofs [10] 2. a set of rules R, describing the interaction among objects: The rules can be defined explicitly [16, 7] or implicitly by using string matching string concatenation [2, 13, 14] lambda calculus [8, 9] Turing machines [15] finite state machines or machine code ....

....[4] or proofs [10] 2. a set of rules R, describing the interaction among objects: The rules can be defined explicitly [16, 7] or implicitly by using string matching string concatenation [2, 13, 14] lambda calculus [8, 9] Turing machines [15] finite state machines or machine code language [6], proof theory [9] matrix multiplication [3] or simple arithmetic operations [4] 3. an algorithm A driving the system: The algorithm describes how the rules are applied to a collection of objects (soup population) The algorithm may simulate a wellstirred reaction vessel with no topology [1, 6, ....

[Article contains additional citation context not shown here]

P. Dittrich and W. Banzhaf. Self-evolution in a constructive binary string system. submitted to Artificial Life, revision in preparation, 1998.

....which catalyzed reactions occur and with what strength. Here, we describe an artificial chemistry as a system which consists of the following three parts: 1. a set of objects S : These objects may be abstract symbols [2] character sequences [1, 3, 4] lambda expressions [5] binary strings [6, 7, 8], numbers [9] or proofs [10] 2. a set of rules R, describing the interaction among objects: The rules can be defined explicitly [2, 11] or implicitly by using string matching string concatenation [1, 4, 12] lambda calculus [5, 13] Turing machines [8] finite state machines or machine code ....

....[9] or proofs [10] 2. a set of rules R, describing the interaction among objects: The rules can be defined explicitly [2, 11] or implicitly by using string matching string concatenation [1, 4, 12] lambda calculus [5, 13] Turing machines [8] finite state machines or machine code language [7], proof theory [13] matrix multiplication [6] or simple arithmetic operations [9] 3. an algorithm A driving the system: The algorithm describes how the rules are applied to a collection of objects (soup population) The algorithm may simulate a well stirred reaction vessel with no topology [1, ....

[Article contains additional citation context not shown here]

P. Dittrich and W. Banzhaf. Self-evolution in a constructive binary string system. To appear in: Artificial Life, 4(2), 1998.

....structure. Syntactical and semantic closure. But hard to find for variable length systems. Abstract Automata 1. Turing machine like, RNA motivated [McCaskill, 1988, Thurk, 1993] 2. Machine Tape Interaction [Ikegami and Hashimoto, 1995] 3. Finite state machine, abstract [Dittrich, 1995, Dittrich and Banzhaf, 1998] Machine Tape Interaction [Ikegami and Hashimoto, 1995] 1. Objects Tapes: 7 bit circular binary strings S T = f0; 1g 7 Machines: 16 bit binary strings ( 4 bit head, 4 bit tail, 8 bit transition table) 2. Reactions M i T j X Gamma M i T j M k T l ; X implicit, X ] const: 3. ....

....= M k T l ; M k T l = M i T j Machine Tape Interaction Observations (III) III. High noise [Ikegami and Hashimoto, 1995] 1. Only small, degenerated core networks emerge. 2. Low diversity and very low (even no) active mutation. Self evolution in an Automaton Chemistry [Dittrich, 1995, Dittrich and Banzhaf, 1998] 1. Objects 32 bit binary strings S = f0; 1g 32 2. Reactions s 1 s 2 X = s 1 s 2 s 3 X implicit, X ] const: s 1 folded to an automaton A. A is applied to s 2 to produce product s 3 . 3. Dynamics explicit population, single collision simulation Automata Reaction ALU 0 1 1 0 ....

[Article contains additional citation context not shown here]

Dittrich, P. and Banzhaf, W. (1998). Self-evolution in a constructive binary string system. Artificial Life, 4(2):203--220.

....the automaton will halt after a finite number of steps. Because A s 1 is a deterministic finite automaton, the automata reaction defines a function r a2 : f0; 1g 32 Theta f0; 1g 32 Gamma f0; 1g 32 . Due to space limitations, the automaton is not shown here. It is explained in detail in [14, 15]. The collision rules are all secondorder catalytic reactions of the form s 1 s 2 X Gamma s 1 s 2 s 3 , shortly s 1 s 2 = s 3 . All collisions of two objects s 1 ; s 2 will have a unique outcome s 3 . 7 Self organizing Topology Based on Hashing A topology restricts the possible ....

P. Dittrich and W. Banzhaf. Self-evolution in a constructive binary string system. To appear in: Artificial Life, 4(2), 1998.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC