Showing papers on "Artificial immune system published in 1992"

Computation and the immune system

Abstract: 1 1ntr0duct10n Natura1 1mmune 5y5tem5 pr0v1de an exce11ent examp1e 0f emer9ent c0mputat10n [5], 1n wh1ch adaptat10n 0perate5 at the 10ca1 1eve1 and u5efu1 6ehav10r (e.9., rec09n121n9 ant19en5) emer9e5 at the 9106a1 1eve1:. We are deve10p1n9 5evera1 a190r1thm5 6a5ed 0n the 1mmune 5y5tem • 5 a6111ty t0 rec09n12e and remem6er an en0rm0u5 num6er 0f pattern5. Fr0m an 1nf0rmat10n-pr0ce551n9 per5pect1ve, the 1mmune 5y5tem mu5t rec09n12e an a1m05t 11m1t1e55 num6er 0f m01ecu1ar pattern5 0n f0re19n ce115 and m01ecu1e5, 1.e., ant19en5, w1th extreme1y 11m1ted re50urce5. A m0u5e 15 th0u9ht t0 6e a61e t0 make 0n the 0rder 0f 1011 d1fferent recept0r m01ecu1e5, even th0u9h 1t5 ent1re 9en0me pr06a61y c0nta1n5 fewer than 105 9ene5 [2, 1]. 7he 1mmune 5y5tem • 5 pattern-rec09n1t10n a6111t1e5 are even m0re 1mpre551ve when 0ne c0n51der5 that the 5y5tem 15 d15tr16uted thr0u9h0ut 0ur 60d1e5; there 15 n0 centra1 • • 1mmune 0r9an • • that c0ntr015 wh1ch ant160d1e5 are pr0duced 0r when. 7hu5, the 1nd1v1dua1 1ymph0cyte5 and ant160d1e5 that c0mpr15e the 1mmune 5y5tem enc0de and 0perate the c0ntr01 mechan15m 1n para11e1. 1n 0rder t0 pr0v1de 1t5 defen5e funct10n5 eff1c1ent1y, the 1mmune 5y5tem mu5t perf0rm pattern rec09n1t10n ta5k5 t0 d15t1n9u15h 5e1[ m01ecu1e5 and ce115 fr0m f0re19n 0ne5 (ant19en5). 7he num6er 0f f0re19n m01ecu1e5 that the 1mmune 5y5tem can rec09n12e 15 unkn0wn 6ut 1t ha5 6een e5t1mated t0 6e 9reater than 10 7M [12]. 1n pract1ca1 term5, e55ent1a11y any f0re19n m01ecu1e pre5ented t0 the 1mmune 5y5tem, even th05e created 1n the 1a60rat0ry and thu5 never hav1n9 appeared 6ef0re 1n a11 0f ev01ut10nary t1me, are rec09n12ed a5 6e1n9 f0re19n. 8e51de5 th15 1mmen5e rec09n1t10n capac1ty, the 0ther feature that d15t1n9u15he5 the verte6rate 1mmune 5y5tem fr0m the defen5e 5y5tem5 0f 10wer 0r9an15m5 15 that the 1mmune 5y5tem 1earn5 and exh161t5 mem0ry. 7hu5 the re5p0n5e t0 the 5ec0nd exp05ure 0f the 5ame ant19en 0ccur5 m0re 4u1ck1y and v190r0u51y than the f1r5t re5p0n5e. We have deve10ped a m0de1 d1rected at under5tand1n9 the pattern rec09n1t10n pr0ce55e5 0f tw0 a5pect5 0f the 1mmune 5y5tem, c10na1 5e1ect10n and 10n9-term ev01ut10n 0f 9ene5. 7he m0de1 15 6a5ed 0n a un1ver5e f1r5t 1ntr0duced 6y Farmer et a1.

Randomness and pattern scale in the immune network

Abstract: The immune system is a beautiful example of a complex information processing system The complexity of the immune system is comparable to that of the nervous system Both systems are comprised of a large number of different cell types communicating via the production of stimulatory or inhibitory molecules Several of these cell types form connected networks of billions of different nodes In the neural network the nodes are fixed and communicate via electrical signals; in the immune system the nodes recirculate and communicate via molecules (e g , antibody) or cell-to-cell contacts An important property of both systems is "learning" During its early life the immune system learns to discriminate between self and nonself Additionally, the immune system has a form memory which is known as "immunity": secondary immune responses are usually different from primary responses

The evolution of cooperation in immune system libraries

Abstract: The immune system is a distributed and highly parallel information processing system. We introduce an abstract model of the immune system and evaluate its ability to perform a pattern recognition task. The experiments here show that simulated evolution is a sufficient mechanism for organizing the genetic material of the immune system. It is also shown that the organization of the immune system can be induced even when the fitness function is stochastic and the sampling noise is very high. The components of the immune system must work cooperatively to recognize foreign elements in the body, a task that we equate with performing distributed computation. This suggests that software for distributed computing systems could be produced through a process of adaptive evolution rather than direct programming.