[my-paper]
"This paper presents the guidelines of an ongoing project of the 'Movement Dynamics' team in the Movement and perception' Lab, UMR6152, Marseille." => done

"In order to give an intuitive view of the learning mechanisms we plan to implement, we present here a simple sequence classification experiment (see figure 1). The network is made of I&F neurons. The delays are inhomogeneous, and the mean transmission time is 10 ms. The temporal resolution is 0.5 ms. The network is composed of 3 layers : the first layer is composed of 4 neurons that receive a discontinuous input signal. The internal layer is composed of 200 neurons, whose interconnection pattern is a centered Normal law. The output layer is composed of 2 neurons. The output links are initially excitatory and homogeneous (while the delays are inhomogeneous), and the two output neurons are mutually inhibitive."
=>
In order to have an intuitive view of ... we plan to implement, we present here 
a couple of ... 

aiming a recopgnition of moving object from 2-D video scenario, the task here is ...
for the sake of simplisity

The first one is the model proposed by ..., which motivate us to this ...

So the second is 

A realistic model of the saccade adaptation is for instance currently under development

If the current from receptor Nr(x, y) is stable, i.e. current SNr(x, y, t) is
equal to current SNr(x, y, t-��t), the excitatory input and the inhibitory input of neuron
N1(x, y) can be balanced by adjusting the parameters of synapses, and then neuron
N1(x, y) is silent. If the current of receptor Nr(x, y) becomes stronger, i.e. the current
SNr(x, y, t) is larger than current SNr(x, y, t-��t), the balance is broken, and then neuron
N1(x, y) will generate spikes if SNr(x, y, t) is larger enough than SNr(x, y, t-��t). If the
current of receptor Nr(x, y) becomes weaker, neuron N1(x, y) does not fire.

In this case, input neuron N2(x, y) through inhibitory synapse becomes weaker, but input
through excitatory synapse is still strong. Neuron N2(x, y) will fire if SNr(x, y, t) is
smaller enough than SNr(x, y, t-��t). Therefore, the gray scale changes of pixels in the
image are reflected in the output neuron layer, i.e. Neuron N(x�f, y�f) will fire if Neuron
N1 or Neuron N2 fires.


=> let's check with single pair 

to simplify the authors claim let's ....


then far much lager system, say, even 1000x1000 input from pixcel -> not realistic to adjust w in this way

so ...


=> lot's of future