<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body>
<p>Hi all,</p>
<p>We are offering one postdoc position on self-organized network
modelling at the Basque Center for Applied Mathematics in Bilbao
(Basque Country, Spain) under the supervision of MIguel Aguilera.</p>
<p>DETAILS: **Postdoctoral Fellow on open-ended, self-organized,
bio-inspired networks**<br>
<br>
Topics: One of the outstanding challenges in modelling living and
cognitive systems is to capture their ability to continuously
adapt and develop. Our behavioural responses are not fixed but
driven instead by a complex ecology, composed of myriads of fluid
and inconspicuous neurodynamical patterns that have slowly grown
on us. AI models like neural networks can capture complex
behaviours, but often this open-ended, liquid and ongoing
reconfiguration is incredibly challenging to reproduce with static
topologies. In contrast, liquid neural networks (or 'liquid
brains') are a widespread class of networks with a particular
feature: nodes (which may represent 'neurons' or 'agents' and
typically all share identical rules) not only process information
from neighbour nodes, but they also dynamically modify their
network connections, e.g. by moving in space. These networks show
how neural-like processing typically associated with static
physiological networks can also emerge from fluid collective
interaction, dynamically exploring configuration spaces beyond
standard connection weight changes. Examples of this kind of model
include collective decision-making and fluctuations in ant
colonies or idiotypic cascades in immune networks. <br>
<br>
Objective: The aim of this project is to develop a theory of
learning in liquid brains, focused on two aspects: 1) How do
liquid brains learn? How is this process different from static
neural networks?, and 2) what is the adaptive potential of liquid
brains when they are embodied as an agent in interaction with a
changing external environment? Answering these questions has the
potential to extend the idea of liquid brains from a theoretically
deep and intriguing concept to a useful tool available to the
machine learning community. Specifically, liquid brains could
afford more open-ended, self-improving systems, exploiting fluid
reconfiguration of nodes as an adaptive dimension which is
generally unexplored. This could also allow modes of learning that
avoid catastrophic forgetting, as reconfigurations in the network
are based on reversible movement patterns. In terms of technology
transfer, this advances can also have important implications for
new paradigms like edge computing. <br>
<br>
PI in charge: <a moz-do-not-send="true"
href="https://maguilera.net/">Miguel Aguilera </a><br>
<br>
Salary and conditions: The gross annual salary of the Fellowship
will be 29.120€ - 35.360€ according to experience. Additionally,
we offer a moving allowance up to 2.000€. Should the researcher
have a family at the time of recruitment: 1. 2.000€ gross in a
single payment will be offered (you must be married-official
register or with children and the certificate to prove it must be
sent). 2. 1.200€ gross per year/per child (up to 2 children) will
be offered (the certificate to prove it must be sent). <br>
<br>
Contract and offer: 1 + 1 years Deadline: 8TH September 2023,
14:00 CET <br>
<br>
How to apply: Applications must be submitted on-line at: <a
moz-do-not-send="true"
href="https://joboffers.bcamath.org/apply/ic2023-08-postdoctoral-fellow-in-open-ended-self-organized-bio-inspired-networks"
class="moz-txt-link-freetext">https://joboffers.bcamath.org/apply/ic2023-08-postdoctoral-fellow-in-open-ended-self-organized-bio-inspired-networks</a></p>
<p>**Scientific Profile Requested**<br>
<br>
Requirements: PhD in Physics, Engineering, Computer Science,
Maths, Artificial Intelligence, Computational Neuroscience and
related fields. <br>
<br>
Skills and track-record: <br>
* Modelling experience with complex or neural network models, e.g.
Hopfield networks, Ising models, Boltzmann machines, sandpile
models, flock/swarm/insect colony models. <br>
* Solid programming skills <br>
* Analytical study of stochastic processes <br>
* Demonstrated ability to work independently and as part of a
collaborative research team. <br>
* Ability to present and publish research outcomes in spoken
(talks) and written (papers) form. <br>
* Fluency in spoken and written English. <br>
<br>
Scientific Profile:<br>
The preferred candidate will have: <br>
* Strong background in complex systems, neural network modelling,
biophysics, information theory and/or statistical mechanics <br>
* Interest to work in interdisciplinary research projects</p>
<p>Application and Selection Process Formal Requirements: The
selected candidate must have applied before the application
deadline online at the webpage <a moz-do-not-send="true"
href="https://joboffers.bcamath.org/"
class="moz-txt-link-freetext">https://joboffers.bcamath.org/</a>
The candidates that do not fulfil the mandatory requirements will
not be evaluated with respect to their scientific profile.
Additional documents could be requested during the evaluation
process so as to check this fulfilment. <br>
<br>
Application: Required documents: <br>
* CV <br>
* Letter of interest <br>
* 2 recommendation letters <br>
* Statement of past and proposed future research (2-3 pages) <br>
<br>
Evaluation: Based on the provided application documents of each
candidate, the evaluation committee will evaluate qualitatively:
the adaption of the previous training and career to the profile
offered, the recommendation letters, the main results achieved
(papers, proceedings, etc.), the statement of past and proposed
future research and other merits; taking in account the alignment
of these items to the topic offered. <br>
<br>
Incorporation: As soon as possible<br>
</p>
<pre class="moz-signature" cols="72">--
Miguel Aguilera | Ikerbasque Research Fellow | BCAM – Basque Center for Applied Mathematics | <a class="moz-txt-link-freetext" href="https://maguilera.net" moz-do-not-send="true">https://maguilera.net</a></pre>
</body>
</html>