Write Code will travel*....... *(Draw)

I presented my Morphogenetics project yesterday at the AAC. A soft starfish grips  an abstract 
vase like form beneath. With each grip, a measurement is taken of its dimensions, the data
being incrementally gathered from a data set imported into processing. 

I’m going to start working with a Hebbian type of Neural network in order that the starfish will
be able to recognise the form from only partial data gathered. The network should be able to
complete the ‘memory’ of the form by updating the weight values between the Neurons at
each time step. 

Thanks to my Tutors Martha Tsigkari and Angelos Chronis for their help with this. 

A quick description of Hebbian learning:
Hebbian theory is a scientific theory in biological neuroscience which explains the adaptation
of neurons in the brain during the learning process. It describes a basic mechanism for 
synaptic plasticity wherein an increase in synaptic efficacy arises from the presynaptic cell’s 
repeated and persistent stimulation of the post synaptic cell. Introduced by Donald Hebb in
1949, it is also called Hebb’s rule, Hebb’s postulate, and cell assembly theory.

http://en.wikipedia.org/wiki/Hebbian_theory 

Ive been working on some prelim studies for the Morphogonetic Programming module on
the Bartlett AAC. 

These are based on the Soft Robotics project that I have been working on with Zoi Karagkiozi
in the previous module. 
 

Im essentially programming a Bezier function to define the surfaces for the inflatable
actuators here. In particular I am also trying to build some of the elements that are native
to Non Uniform B splines such as the Knot Vector. Knot Vectors give a greater degree of
control than is normally associated with the standard Bezier function. On the other hand,
Bezier functions come with the benefit of being able to define the beginning and end
points of the surface more explicitly. Therefore I am experimenting with adding
different weightings to the Control points to give at least the ‘feel’ of non uniformity.
 

This has thrown up further questions about the distributed surface geometry however for
which I will be looking into a Dynamic Relaxtion algorithm to solve.
 

(Source: ben-zoi, via ben-zoi)

(Source: ben-zoi, via ben-zoi)

(Source: ben-zoi)