Review

From: Annamalai Muthu (muthu@cs.washington.edu)
Date: Wed Dec 01 2004 - 02:16:42 PST

  • Next message: Craig M Prince: "Reading Review 12-01-2004"

    Paper Title: The evolutionary origin of complex features

    One-line summary of the paper: The paper tries to explain the role of
    various components of evolution in the building of complex features, with
    the help of .digital organisms..

    Main contributions of the paper:

    -It has always puzzled me as to how humans had managed to attain such a
    level of complexity in their structure. This paper uses a model
    constituting of digital organisms to explain how certain complex features,
    which provide immense benefit to the organism, evolve over time.

    -The digital organisms (which are nothing but a sequence of instructions),
    thrive by capturing energy from the environment to execute instructions
    and replicate. Every time the sequence of instructions produces
    meaningful functions, it is rewarded appropriately. This gives rise to a
    form of competition between digital organisms. The good ones manage to
    survive, while the bad ones die out. This competitive nature directs
    evolution (mutation and replication) in the right direction.

    -The paper shows how the EQU function (the most complex of them all)
    evolves incrementally. First, the simpler functions are realized, then
    they go on to build the more complex form.

    Flaws:

    -The paper presents a picture which claims that these digital organisms
    can represent the natural evolution process very well. But I do not
    agree. All organisms in the digital world live in an ideal world where
    everyone is rewarded for being good, all organisms reside in the same
    environment, the number of possible mutations is low and every organism is
    very similar to the other organisms since it comes from the same set of 26
    instructions. In the real world, the environment is not ideal, since not
    all good organisms are rewarded, some good traits can go unrewarded. The
    number of possible mutations is of the order of at least a few billion.
    There are a variety of organisms in the real world, each type being
    extremely diverse.

    -Most complex organisms reproduce sexually, hence the applicability of
    this idea might be greatly restricted.

    Is this search???!!!

    This can be thought of as search. The states constitute of the various
    organisms put together, while the evaluation function used to guide the
    search can be the fitness of the individuals in the population. The
    operations used to get from one state to another are replication and
    mutation. The goal state can be a variety of things, depending on the
    required solution. It could be defined as one in which at least one
    organism has reached a state of optimality. Alternatively, it could be
    defined as one in which at least a certain portion of the organisms have
    reached optimality. This search requires you to have great foresight,
    because an immediate deleterious effect can have a good impact a few
    generations down the line.


  • Next message: Craig M Prince: "Reading Review 12-01-2004"

    This archive was generated by hypermail 2.1.6 : Wed Dec 01 2004 - 02:16:43 PST