prof. christoph lischka

Wir haben als erste Form der Datenabstraction "Strukturen" kennengelernt - "positions", "students" etc.

Nicht jedem scheint dies so geläufig; also würde ich vorschlagen, daß der nächste Dienstag ganz im Zeichen der Vertiefung stehen sollte. Sektionen 6.5 - 6.7, sowie Sektion 7 (komplett) von HtDP wären dazu angesagt. Also: bitte bis nächsten Dienstag (incl.) durcharbeiten; am 27. 11. gibt's dann Übung 7.4 als Test

Außerdem trägt Marcel am 27. über L-Systems vor ... 

Nagehan wrote a nice summary of her talk last week on Bruno Latour's Introduction to "Reassembling the Social":

Introduction Summary of the Book ‘Reassembling the Social’

French sociologist Bruno Latour intends to devise an alternative approach to social theory.
Latour’s goal is to redefine the notion of the social while going back to the traditional
meaning of the word ‘social’ itself and explaining the Latin root of the word as ‘to follow’.
He firstly emphasizes the metamorphosis of the two terms; science and social, by doubting the
existence of specific relations and orders that are combined together to form and function as a
society (p.2). According to Latour since both the social and the science is stable over years
there can not be a strong designation for the term ‘socio-logy’. Therefore he devises an
alternative definition of sociology by the inquiry of such questions as what society is and what
the word social means mainly and in order to reveal these inquiries he comments on two
different views on social theory where one of them is the common sense whereas the other is
the alternative approach taken into analysis throughout the book.

Beginning with the first approach, Latour defines it as the ‘sociology of the social’ in
which the actors of a predefined social ‘context’ are limited to the degree of taking actions
which can only be observed and that’s effects can only be explained by the trained eyes of
social scientists (p.4). In this approach ‘social’ is considered as a specific domain of reality
such as mathematics, physics etc. in where the scientist can observe the phenomenon on
material basis and establish some rules. By then, the limitation of the actor’s activity by social
scientists is inevitable when the observed one is already a defined pattern of actions. This so-
called specific domain of reality also deals mainly with the issues that can not be explained by
other domains such as psychology, law, economics etc. but rather it can enlighten the
remaining aspects of these domains. By this view it is generally believed that there are some
hidden forces in actor’s motivation while acting so and these forces can only be revealed by
certain existing rules of ‘sociology of science’.

The second approach, on which Bruno Latour establishes his alternative definition as
actor-network-theory, refuses to accept the specification of social context, social order,
distinct domain of reality and the stability of the social. Rather than accepting the social as the
predefined order, it embraces the idea that the social is the connecting elements of the actors’
interaction within themselves or their environment. Since ‘the actor’ is the determinant factor
for both the innovations he establishes and the method he uses, the actor is the one to be
followed in order to reach a consensus on social theory. In this context, Latour defines this
second approach as the “sociology of associations” or the “sociology of collective” (p. 9).

For actor-network theory Latour comes up with an acronym A.N.T as a definition to this
alternative approach and he adopts the idea of himself mimicking an ANT between the other
ANTs -actors- who must be followed and observed within the social. Rather than an absolutist
application, this approach brings along relativity as a means of analyzing accelerated
innovations of actors. The actors within this understanding, according to Latour, should not
also be limited to humans but also should be extended to non-humans. The actor can not be
reduced to objects; however, it should be understood within the context of different elements
influencing each other to create new associations. Relative to that context, he defines the
social as the “associations between heterogeneous elements” (p.5). He also grants the
differences of the actors as the root sources of new innovations within the social. The social
also can not be considered as a materialized object since it is established by “the connections
of things that are not themselves social” (p.5). From there arises the actor network defined in
Latour’s theory as the action connected together with all its affects of the surrounding and
then forming a network.

Bruno Latour also embraces French sociologist, criminologist and social psychologist
Gabriel Tarde as an early ancestor of ANT and mentions his definition of society as an
opposition to younger generation sociologist Durkheim (p.15). According to Tarde the
definition of the society is built up by the movements of the actors that act in a “law of
evolution” which presupposes series of irregular changes and dynamism within itself (p.14).
The substantial part of this evolution is not human beings but innovations within social
connection, created by aggregations over time.

In the first part of the book Bruno Latour addresses the reason of abstention to limit the
actors that ‘sort of beings’ living in the network.  He aims to explain new innovations by
setting up a mind that centers relativity. In the second part of the book Latour emphasizes the
ways to trace these new innovations by observing social connections. Finally he concludes by
defending the crucial fact of assembling the collective after assimilating the understanding of
society and definition of social.

References:
Latour, B. (2005) Reassembling the Social (pp.1-17). Oxford University Press.

Actually, in our last meeting in the "Evolutionary Computation" class we established three working groups, which are

• Embedded Erlang. Main focus of this group is to get Erlang running on our gumstix. Some starting points are: Trapexit: Erlang on embedded linux (no permalink available), edtk, the Dryverl Toolkit, and - as a general resource - the SWIG stuff. Nice introductions are Scott Lystig Fritchie's paper, and Romain Lenglet's description of Dryverl.
• Spiking Neuron Networks. There are several advantages of Spiking Neuron Networks (SNNs) over traditional connectionist models, in particular for encoding temporal patterns. Seminal work has been done by Dario Floreano et al. at the EPFL, which sets the basis for our own implementation. Still a classic is the introductory reader by Maass/Bishop: Pulsed Neural Networks. The following links will give us a good overview of the research done at EPFL: Evolutionary Bits 'n' Spikes, Evolution of Spiking Neural Controllers for Autonomous Vision-Based Robots, and From Wheels to Wings with Evolutionary Spiking Circuits.
• Genetic Algorithms. We want the controller to get evolved, and Genetic Algorithms turned out to be an appropriate tool. Here are some good entries: the slides on "Evolutionary Computation" by C.J. Burgess, and the tutorial by Darrell Whitley. In order to get the stuff running under Erlang, the following links might be of some help: the goevo tool from EPFL (single processor), the parallel PIKAIA implementation (in FORTRAN), and the (general) GAUL toolchain. To get some inspiration for a functional implementation, have a look at Brown et al., A genetic algorithm framework using Haskell.

You can find a nice and short summary/reference of Erlang programming language here:

Erlang Handbook. 

Tim-Christian gave a nice overview on the first chapter of Eduardo Kac's new reader "Signs of Life", which you can download here:

Robert Baruch wrote a nice introductory paper on Erlang – recommended to people who have already programming experience. Get it here.

• Most recent version of Erlang
• Latest release of JRE (Java Runtime)
• Latest release of ErlyBird