id: 00221025
dt: j
an: 00221025
au: Amit, Daniel J.; Fusi, Stefano
ti: Constraints on learning in dynamic synapses.
so: Network 3, No.4, 443-464 (1992).
py: 1992
pu: IOP Publishing Ltd., Bristol
la: EN
cc: 
ut: analogue dynamic synapses; constraints; brain function; unsupervised
    learning; attractor neural networks; Hebbian-type learning; exponential
    decay; rate of presentation of patterns; refresh time interval;
    retrieval
ci: 
li: doi:10.1088/0954-898X/3/4/008
ab: Summary: Hebbian-type learning is discussed in a network whose synapses are
    analogue, dynamic variables, whose values have to be periodically
    refreshed due to possible exponential decay, or other instability of
    continuous synaptic efficacies. It is shown that the end product of
    learning in such networks is very sensitive to the relation between the
    rate of presentation of patterns and the size of the refresh time
    interval. It is shown that in the limit of slow presentation, the
    network can learn at most $O(\ln N)$ patterns in $N$ neurons, and must
    learn each one in one shot, thus learning all errors present in a
    corrupt stimulus presented for retrieval. It is then shown that as the
    rate of presentation is increased the performance is increased rapidly.
    Another option we investigate is that in which the refresh mechanism is
    acting stochastically. In this case the rate of learning can be slowed
    down very significantly, but the number of stored patterns cannot
    surpass $\sqrt{N}$.
rv: