Spiking Neural P Systems - Advancing Artificial Intelligence Through Biological Process Applications - page 72

as pattern recognition, training/learning, etc.;
develop dedicated hardware for simulating SN P
systems, both with didactical goals (e.g., check-
ing examples or proofs) and perhaps useful in
applications; after having an enhanced model,
faithful to neuro-biological reality, and efficient
software tools, look for biological or bio-medical
applications; try to learn computer science relevant
ideas from the computing by spiking. All these
research directions were already preliminarily
addresses, hence we believe that they will prove
to be solid trends along which the study of SN P
systems will develop.

tionally hard problems; investigate further the
feature used in (Leporati et al., 2007a), (Leporati
et al., 2007b), i.e., the fact that checking the ap-
plicability of a spiking rule is done in one clock
tick, irrespective of the complexity of the regular
expression which control the rule; define systems
with a dynamical synaptic structure; compare the
SN P systems as generator/acceptor/transducers
of infinite sequences with other devices handling
such sequences; investigate further systems with
exhaustive and other parallel ways of using the
rules, as well as systems working in a non-syn-
chronized way; find classes of (accepting) SN P
systems for which there is a difference between
deterministic and non-deterministic systems; find
classes which characterize levels of computabil-
ity different from those corresponding to finite
automata (semilinear sets of numbers or regular
languages) or to Turing machines (recursively
enumerable sets of numbers or languages). Several
of these research topics were preliminarily investi-
gated - see the additional reading provided below
- but serious research efforts are still needed.

CONCLUSION

We have only briefly introduced the reader to a
fast developing and much promising branch of
membrane computing, inspired from the way
the neurons communicate by means of electrical
impulses. Although the model is rather reduction-
istic (we may say, minimalistic), it proves to be
both powerful and efficient from a computational
point of view.

REFERENCES

FUTURE RESEARCH TOPICS

Cavaliere, M., Egecioglu, O., Ibarra, O.H., Io-
nescu, M., Păun, Gh., &. Woodworth, S. (2007).
Unsynchronized spiking neural P systems: decid-
ability and undecidability. Proceedings of 13rd
DNA Based Computers Conference, Memphis,
USA.

Many problems were already mentioned above,
many others can be found in the papers listed
below, and further problems are given in (Păun,
2007). Large research directions (trends) were also
mentioned in a previous section. That is why we list
here only a few more technical questions. Investi-
gate the way the axon not only transmit impulses,
but also amplifies them; consider not only “posi-
tive” spikes, but also inhibitory impulses; define
a notion of memory in this framework, which can
be read without being destroyed; provide ways
for generating an exponential working space (by
splitting neurons? by enlarging the number of
synapses?), in such a way to trade space for time
and provide polynomial solutions to computa-

Chen, H., Freund, R., Ionescu, M., Păun, Gh., &
Perez-Jimenez, M.J. (2007). On string languages
generated by spiking neural P systems. Funda-
menta Informaticae , 75 (1-4), 141-162.

Chen, H., Ishdorj, T.-O., Păun, Gh., & Perez-
Jimenez, M.J. (2006a). Spiking neural P systems
with extended rules. In (Gutierrez-Naranjo et al.,
eds., (2006)), vol. I, 241-265.

Chen, H., Ishdorj, T.-O., Păun, Gh., Perez-
Jimenez, M.J. (2006b). Handling languages with