UH Biocomputation Group - Neuromorphic hardwarehttp://biocomputation.herts.ac.uk/2022-11-16T18:28:39+00:00Neuromorphic Engineering Needs Closed-Loop Benchmarks2022-11-16T18:28:39+00:002022-11-16T18:28:39+00:00Nik Dennlertag:biocomputation.herts.ac.uk,2022-11-16:/2022/11/16/neuromorphic-engineering-needs-closed-loop-benchmarks.html<p class="first last">Nik Dennler's Journal Club session where he will talk about a paper "Neuromorphic Engineering Needs Closed-Loop Benchmarks"</p>
<p>This week on Journal Club session Nik Dennler will talk about a paper "Neuromorphic Engineering Needs Closed-Loop Benchmarks".</p>
<hr class="docutils" />
<p>Neuromorphic engineering aims to build (autonomous) systems by mimicking
biological systems. It is motivated by the observation that biological
organisms- from algae to primates- excel in sensing their
environment, reacting promptly to their perils and opportunities. Furthermore,
they do so more resiliently than our most advanced machines, at a fraction of
the power consumption. It follows that the performance of neuromorphic systems
should be evaluated in terms of real-time operation, power consumption, and
resiliency to real-world perturbations and noise using task-relevant evaluation
metrics. Yet, following in the footsteps of conventional machine learning, most
neuromorphic benchmarks rely on recorded datasets that foster sensing accuracy
as the primary measure for performance. Sensing accuracy is but an arbitrary
proxy for the actual system's goal- taking a good decision in a
timely manner. Moreover, static datasets hinder our ability to study and
compare closed-loop sensing and control strategies that are central to survival
for biological organisms. This article makes the case for a renewed focus on
closed-loop benchmarks involving real-world tasks. Such benchmarks will be
crucial in developing and progressing neuromorphic Intelligence. The shift
towards dynamic real-world benchmarking tasks should usher in richer, more
resilient, and robust artificially intelligent systems in the future.</p>
<div class="line-block">
<div class="line"><br /></div>
</div>
<p>Papers:</p>
<ul class="simple">
<li>M. Milde, S. Afshar, Y. Xu, A. Marcireau, D. Joubert, B. Ramesh, Y. Bethi, N.
Ralph, S. El, Arja, N. Dennler, A. van Schaik, G. Cohen, <a class="reference external" href="https://doi.org/10.3389/fnins.2022.813555">"Neuromorphic
Engineering Needs Closed-Loop Benchmarks"</a>, 2022, Frontiers in
Neuroscience, 16, 813555</li>
</ul>
<p><strong>Date:</strong> 2022/11/16 <br />
<strong>Time:</strong> 14:00 <br />
<strong>Location</strong>: online</p>
Rapid online learning and robust recall in a neuromorphic olfactory circuit2021-02-04T18:51:50+00:002021-02-04T18:51:50+00:00Damien Drixtag:biocomputation.herts.ac.uk,2021-02-04:/2021/02/04/rapid-online-learning-and-robust-recall-in-a-neuromorphic-olfactory-circuit.html<p class="first last">Damien Drix's Journal Club session where he will talk about a paper "Rapid online learning and robust recall in a neuromorphic olfactory circuit"</p>
<p>This week on Journal Club session Damien Drix will talk about a paper "Rapid online learning and robust recall in a neuromorphic olfactory circuit".</p>
<hr class="docutils" />
<p>We present a neural algorithm for the rapid online learning and identification
of odourant samples under noise, based on the architecture of the mammalian
olfactory bulb and implemented on the Intel Loihi neuromorphic system. As with
biological olfaction, the spike timing-based algorithm utilizes distributed,
event-driven computations and rapid (one shot) online learning. Spike
timing-dependent plasticity rules operate iteratively over sequential
gamma-frequency packets to construct odour representations from the activity of
chemosensor arrays mounted in a wind tunnel. Learned odourants then are
reliably identified despite strong destructive interference. Noise resistance
is further enhanced by neuromodulation and contextual priming. Lifelong
learning capabilities are enabled by adult neurogenesis. The algorithm is
applicable to any signal identification problem in which high-dimensional
signals are embedded in unknown backgrounds.</p>
<div class="line-block">
<div class="line"><br /></div>
</div>
<p>Papers:</p>
<ul class="simple">
<li>Imam, N., Cleland, T.A. <a class="reference external" href="https://www.nature.com/articles/s42256-020-0159-4">"Rapid online learning and robust recall in a neuromorphic olfactory circuit"</a>, Nat Mach Intell 2, 181–191 (2020)</li>
</ul>
<p><strong>Date:</strong> 2021/02/05 <br />
<strong>Time:</strong> 16:00 <br />
<strong>Location</strong>: online</p>
Prediction of Electrode Position inside rat brain using NeuromorphicHardware2020-11-18T10:58:33+00:002020-11-18T10:58:33+00:00Emil Dmitruktag:biocomputation.herts.ac.uk,2020-11-18:/2020/11/18/prediction-of-electrode-position-inside-rat-brain-using-neuromorphichardware.html<p class="first last">Shavika Rastogi's Journal Club session where she will talk about her master thesis entitled "Prediction of Electrode Position inside rat brain using Neuromorphic Hardware".</p>
<p>This week on Journal Club session Shavika Rastogi will talk about her master thesis entitled "Prediction of Electrode Position inside rat brain using Neuromorphic Hardware".</p>
<hr class="docutils" />
<p>Neural probes with large number of close packed recording sites each comprising
of 32 electrodes are being developed for large scale neuronal recordings from
multiple brain areas simultaneously to understand complex brain activity in vivo.
By precisely mapping the position of each site inside rat brain, they help us to
characterize neural activity on the basis of cortical depth from where it is
obtained. Their application lies in neurosurgery, where it is important to locate
the target of surgical interest inside the brain in real time. In this work, we
have first compared various methods from literature to analyze extracellular
activity recorded using CMOS neural probes from different cortical depths and
from different locations along same lateral axis of rat brain to find a criterion
on the basis of which recordings can be classified. After finding out the most
promising criterion, we have tried to implement it on neuromorphic hardware
SpiNNaker. We tested single neuron and spiking excitatory-inhibitory network
for implementation and found that excitatory-inhibitory network is more robust to
noise present in signal and its output can be improved by introducing lateral
inhibition. Our results show that SpiNNaker can be used for rough indication of
cortical depth.</p>
<div class="line-block">
<div class="line"><br /></div>
</div>
<p><strong>Date:</strong> 20/11/2020 <br />
<strong>Time:</strong> 16:00 <br />
<strong>Location</strong>: online</p>
Open Position: Postdoc in neuromorphic pattern recognition2017-04-25T17:08:31+01:002017-04-25T17:08:31+01:00Michael Schmukertag:biocomputation.herts.ac.uk,2017-04-25:/2017/04/25/open-position-postdoc-in-neuromorphic-pattern-recognition.html<p class="first last">Applications are invited for a postdoc position in the Biocomputation group, at the University of Hertfordshire, UK. The appointment is due to start as soon as possible. Please apply online via <a class="reference external" href="http://jobs.herts.ac.uk">http://jobs.herts.ac.uk</a>, vacancy reference 014592, until 22nd May, 2017 at the latest. More details within.</p>
<p>Applications are invited for a postdoc position in the Biocomputation group at the <a class="reference external" href="http://www.herts.ac.uk/">University of Hertfordshire</a>, UK. The appointment is due to start as soon as possible, with an initial fixed duration until 31st March 2018, with a possibility to extend, pending funding decisions.</p>
<p>The successful candidate will join our efforts to advance neuromorphic pattern recognition on the <a class="reference external" href="http://apt.cs.manchester.ac.uk/projects/SpiNNaker/">SpiNNaker</a> and <a class="reference external" href="http://brainscales.kip.uni-heidelberg.de/public/">BrainScaleS</a> hardware systems that are built within the <a class="reference external" href="https://www.humanbrainproject.eu/en/">Human Brain Project (HBP)</a>. We are looking for a candidate with a keen tenacity in pushing forward the boundaries of future computing off the well-trodden path.</p>
<p>The Biocomputation group provides a rich and inspiring interdisciplinary research environment that connects Computer Science with Neuroscience and branches out into Machine Learning and Robotics.</p>
<p>HBP membership provides excellent opportunities to connect to world-leading scientists in all aspects of neuroscience, high-performance computing, neurorobotics and neuromorphic engineering. We enjoy first-class access to the latest neuromorphic technologies developed in HBP, and we work in tight interaction with the groups developing the hardware systems, <a class="reference external" href="http://apt.cs.manchester.ac.uk/projects/SpiNNaker/">SpiNNaker</a> and <a class="reference external" href="http://brainscales.kip.uni-heidelberg.de/public/">BrainScaleS</a>.</p>
<p>Research in Computer Science at the University of Hertfordshire has been recognized as excellent in the REF 2014, with 50% of the research submitted rated as internationally excellent or world leading. The University is situated in Hatfield, in the green belt just north of London.</p>
<p>Candidates should have a PhD in neuromorphic computing, machine learning, computational neuroscience, computer science, physics or another relevant subject area. In addition, the successful candidate should have demonstrable experience in either:</p>
<ul class="simple">
<li>Machine learning and pattern recognition, ideally deep and recurrent neural networks, self-organisation, or</li>
<li>Neuromorphic computing, ideally with hands-on experience with <a class="reference external" href="http://apt.cs.manchester.ac.uk/projects/SpiNNaker/">SpiNNaker</a>, <a class="reference external" href="http://brainscales.kip.uni-heidelberg.de/public/">BrainScaleS</a>, or other neuromorphic hardware systems, or GPU-accelerated simulations.</li>
</ul>
<p>Further desired skills:</p>
<ul class="simple">
<li>Ability to conduct original research, as evidenced by high-quality, peer-reviewed papers,</li>
<li>Excellent programming skills, as evidenced e.g. by a link to a github account that shows work on relevant projects,</li>
<li>Ability to work and communicate in a multidisciplinary team.</li>
</ul>
<p>Informal inquiries about this post are warmly welcome and should be directed to Dr Michael Schmuker, <code>m.schmuker AT herts DOT ac DOT uk</code>.</p>
<p>Please apply online via <a class="reference external" href="http://jobs.herts.ac.uk">http://jobs.herts.ac.uk</a>, vacancy reference 014592, until 22nd May, 2017 at the latest.</p>
Neuromorphic hardware - what is it, why is everybody talking about it, and does it live up to the hype?2017-03-09T17:05:06+00:002017-03-09T17:05:06+00:00Michael Schmukertag:biocomputation.herts.ac.uk,2017-03-09:/2017/03/09/neuromorphic-hardware-what-is-it-why-is-everybody-talking-about-it-and-does-it-live-up-to-the-hype-.html<p class="first last">Michael Schmuker's journal club session where he discusses neuromorphic hardware</p>
<p>During the last decade, a variety of platforms for neuromorphic computing, so-called neuromorphic hardware systems, have been introduced by different players in academia and industry. They are often praised as promising systems for future computing, especially now that conventional processors don’t get faster anymore. But what is this “neuromorphic hardware” after all, and what are the promises it holds? In my presentation I will give an overview on the neuromorphic technologies that we can access through the Human Brain Project. I will also present results from a recent study where we compared the performance of neuromorphic systems to conventional technologies. This study sheds light on constraints of neuromorphic systems, and opportunities to overcome them. Finally, I will provide an outlook on how neuromorphic sensory computation might look like in the future.</p>
<p><strong>Date:</strong> 10/03/2017 <br />
<strong>Time:</strong> 16:00 <br />
<strong>Location</strong>: LB252</p>