🚀 go-pugleaf

In <1993Jan21.185014.25414@msi.com>
David Christopher Rogers <drogers@riacs.edu> writes:

  When talking about either genetic algorithms or neural networks to people,
  I usually say that these fields were inspired by the physical models, but 
  are certainly not a serious attempt to model either genetics or nervous 
  systems.

  I believe that we algorithmicists need to be quite careful in how we 
  use the vocabulary of other fields, especially if the algorithm is 
  using that field as a neat analogy rather than as a model ...

  ... Science is fighting a terrible tower-of-babel effect
  as it is without needless redefinition, often for what seems to be just
  the ability to have snappy words for the titles of papers.


There is another twist to this that is more bothersome to me. It seems
that if a certain field becomes 'hot' (e.g. neural nets, GAs, fuzzy
etc.)  people may want to describe their algorithms in the new
terminology (why?). For example, some statistical classifier or
clustering algorithms that have been around before the current NN wave
have been reformulated in neural network terminology.  Sometimes I wonder
if some use the terminology in describing their algorithms because of
the true biological inspiration or because of political/economic
reasons.

If biological terminology helps in understanding or explaining an
algorithm, that is good. I think the biological analogy in the case of
GAs is generally useful in explaining this class of
optimization algorithms.  But sometimes it just makes things
confusing.

Philip Thrift