UH Biocomputation Group - Rebecca Miko

Decoding the amplitude and slope of continuous signals into spikes with a spiking point neuron model

2023-11-10T10:49:07+00:00

On this week's Journal Club session, Rebecca Miko will talk about a paper she will soon submit entitled "Decoding the amplitude and slope of continuous signals into spikes with a spiking point neuron model".

In this study, we harness the signal processing potential of neurons, utilizing the Izhikevich point neuron model to efficiently decode the slope or amplitude of fluctuating continuous input signals. Using biophysically detailed compartmental neurons often requires significant computational resources. We present a novel approach to create behaviours and simulate these interactions in a lower-dimensional space, thereby reducing computational requirements. We began by conducting an extensive search of the Izhikevich parameter space, leading to the first significant outcome of our study: i) the identification of optimal parameter sets for generating slope or amplitude detectors, thereby achieving signal processing goals using neurons. Next, we compared the performance of the slope detector we discovered with a biophysically detailed two-compartmental pyramidal neuron model. Our findings revealed several key observations: ii) bursts primarily occurred on the rising edges of similar input signals, iii) our slope detector exhibited bidirectional slope detection capabilities, iv) variations in burst duration encoded the magnitude of input slopes in a graded manner. Overall, our study demonstrates the efficient and accurate simulation of dendrosomatic behaviours. Real-time applications in robotics or neuromorphic hardware can utilize our approach. While biophysically detailed compartmental neurons are compatible with such hardware, Izhikevich point neurons are more efficient. This work has the potential to facilitate the simulation of such interactions on a larger scale, encompassing a greater number of neurons and neuronal connections for the same computational power.

Date: 2023/11/10
Time: 14:00
Location: online

Magnetic-inspired optimization algorithms: Operators and structures

2019-05-16T15:02:58+01:00

Mohammad Tayarani-Najaran's journal club session, where he will present the paper "Magnetic-inspired optimization algorithms Operators and structures (M.-H. Tayarani-N., M.-R. Akbarzadeh-T, 2014)"

A novel optimization algorithm, called the Magnetic Optimization Algorithms (MOAs), is proposed in this paper which is inspired by the principles of magnetic field theory. In MOA, the possible solutions are some magnetic particles scattered in the search space. In this respect, each magnetic particle has a measure of mass and magnetic field according to its fitness. In this scheme, the fitter magnetic particles are more massive, with stronger magnetic field. In terms of interaction, these particles are located in a structured population and apply a long range force of attraction to their neighbors. Ten different structures are proposed for the algorithm and the structure that offers the best performance is found. Also, to improve the exploration ability of the algorithm, several operators are proposed: a repulsive short-range force, an explosion operator, a combination of short-range force and explosion operator and a crossover interaction between the neighboring particles. In order to test the proposed algorithm and the proposed operators, the algorithm is compared with a variety of existing algorithms on 21 numerical benchmark functions. The experimental results suggest that the proposed algorithm outperforms some of the existing algorithms.

Date: 17/05/2019
Time: 16:00
Location: E251

Introduction to Functional Data Analysis

2019-05-16T14:53:41+01:00

Yi Sun's journal club session, where she will present the paper "Applied Functional Data Analysis Methods and Case Studies (Ramsay, J. O. and Silverman B.W, 2002)"

The field of Functional Data Analysis (FDA) has seen rapid development over the last two decades. FDA refers to a collection of methods for analysing data over a curve, surface or continuum. It is very much involved with computational statistics. FDA has been applied to quite broadly in medicine, business and engineering.

In this talk, Yi will introduce the basic idea of FDA using a case study presented in the paper: zooming in on human growth.

“Human growth is not at all the simple process that one might imagine at first sight… Collecting records is time-consuming and expensive, because children have to be measured accurately and tracked for a long period of their lives.

[We] consider how to make this sort of record into a useful functional datum to incorporate into further analyses. A smooth curve drawn through the points is commonly called a growth curve, but growth is actually the rate of increase of the height of the child. In children this is necessarily positive… [We] develop a monotone smoothing method that takes this sort of consideration into account and yields a functional datum that picks out important stages in a child’s growth."

Date: 10/05/2019
Time: 16:00
Location: E251

An EMG-CT method using multiple surface electrodes in the forearm

2019-04-10T11:08:44+01:00

Emil Dmitruk's journal club session, where he will present an overview of an EMG-CT method, while referencing various papers.

Electromyography computed tomography (EMG-CT) method is proposed for visualizing the individual muscle activities in the human forearm. An EMG conduction model was formulated for reverse-estimation of muscle activities using EMG signals obtained with multi surface electrodes. The optimization process was calculated using sequential quadratic programming by comparing the estimated EMG values from the model with the measured values. The individual muscle activities in the deep region were estimated and used to produce an EMG tomographic image.

Emil will reference the following papers:

"An EMG-CT method using multiple surface electrodes in the forearm (Y. Nakajima et al, 2014)"

"An Experimental Model on the Activity of Forearm Muscles Using Surface Electromyography (Y. Nakajima et al, 2009)"

"Surface Conduction Analysis of EMG Signal from Forearm Muscles (Y. Nakajima et al, 2009)"

Date: 12/04/2019
Time: 16:00
Location: D118

OPTICS: Ordering Points To Identify the Clustering Structure

2019-04-03T10:42:27+01:00

Na Helian's journal club session, where she will present the paper "OPTICS Ordering Points To Identify the Clustering Structure (Mihael Ankerst et al, 1999)".

Cluster analysis is a primary method for database mining. It is either used as a stand-alone tool to get insight into the distribution of a data set, e.g. to focus further analysis and data processing, or as a preprocessing step for other algorithms operating on the detected clusters. Almost all of the well-known clustering algorithms require input parameters which are hard to determine but have a significant influence on the clustering result. Furthermore, for many real-datasets there does not even exist a global parameter setting for which the result of the clustering algorithm describes the intrinsic clustering structure accurately. We introduce a new algorithm for the purpose of cluster analysis which does not produce a clustering of a data set explicitly; but instead creates an augmented ordering of the database representing its density-based clustering structure. This cluster-ordering contains information which is equivalent to thedensity-based clusterings corresponding to a broad range of parameter settings. It is a versatile basis for both automatic and interactive cluster analysis. We show how to automatically and efficiently extract not only ‘traditional’ clustering information (e.g. representative points, arbitrary shaped clusters), but also the intrinsic clustering structure. For medium sized data sets, the cluster-ordering can be represented graphically and for very large data sets, we introduce an appropriate visualization technique. Both are suitable for interactive exploration of the intrinsic clustering structure offering additional insights into the distribution and correlation of the data.

Date: 05/04/2019
Time: 16:00
Location: D118

The Potential for Student Performance Prediction in Small Cohorts with Minimal Available Attributes using Learning Analytics Techniques

2019-03-20T12:14:34+00:00

Edward Wakelam's journal club session, where he will present his work.

The measurement of student performance during their progress through university study provides academic leadership with critical information on each student’s likelihood of success. Academics have traditionally used their interactions with individual students through class activities and interim assessments to identify those at risk of failure/withdrawal. However, modern university environments, offering easy on-line availability of course material, may see reduced lecture/tutorial attendance, making such identification more challenging. Modern data mining and machine learning techniques provide increasingly accurate predictions of student examination assessment marks, though data mining and machine learning approaches have focussed upon large student populations and wide ranges of data attributes per student. However, many university modules comprise relatively small student cohorts, with institutional protocols limiting the student attributes available for analysis. It appears that very little research attention has been devoted to this area of analysis and prediction. Ed describes an experiment conducted on a final year university module student cohort of 23, where individual student data is limited to lecture/tutorial attendance, virtual learning environment accesses and intermediate assessments. Ed found potential for predicting individual student interim and final assessment marks in small student cohorts with very limited attributes and that these predictions could be useful to support module leaders in identifying students potentially at risk.

Date: 22/03/2019
Time: 16:00
Location: D118

What Is Decidable About Low Dimensional Hybrid Systems?

2019-03-13T19:01:21+00:00

Olga Tveretina's journal club session, where she will present herself and Andrei Sandler's work.

A hybrid system is a dynamic system that exhibits both continuous and discrete behaviour, and the number of its dimensions is determined by the number of the continuous variables. The hybrid system paradigm is a useful tool for describing a wide range of real-world applications. Examples come from robotics, avionics, biological networks, chemical processes, etc. Most of the hybrid systems are safety critical. Formally, verifying safety properties of hybrid systems consists of construction of a set of reachable states and checking whether this set intersects with a set of unsafe states. Therefore, one of the most fundamental problems in the analysis of hybrid systems is the reachability problem. The reachability problem is known for being difficult, and it is only decidable for special kinds of hybrid systems. Even though many attempts have been made to define the boundary between decidable and undecidable hybrid systems, it is far from being resolved. Asarin, Mysore, Pnueli and Schneider defined some classes of low dimensional hybrid systems lying on the boundary between decidable and undecidable systems in their seminal paper "Low dimensional hybrid systems – decidable, undecidable, don’t know (Asarin et al., 2012)", and for which decidability is unknown. In this talk, Olga will present an overview of the area and discuss the recent work on the decidability of reachability for a class of hybrid systems due to Asarin, Mysore, Pnueli and Schneider.

Date: 15/03/2019
Time: 16:00
Location: D118

ASP: Learning to Forget With Adaptive Synaptic Plasticity in Spiking Neural Networks

2019-02-27T15:41:47+00:00

Sam Sutton's journal club session, where he will present the paper "ASP, Learning to Forget With Adaptive Synaptic Plasticity in Spiking Neural Networks (Panda et al., 2018)"

A fundamental feature of learning in animals is the “ability to forget” that allows an organism to perceive, model, and make decisions from disparate streams of information and adapt to changing environments. Against this backdrop, we present a novel unsupervised learning mechanism adaptive synaptic plasticity (ASP) for improved recognition with spiking neural networks (SNNs) for real time online learning in a dynamic environment. We incorporate an adaptive weight decay mechanism with the traditional spike timing dependent plasticity (STDP) learning to model adaptivity in SNNs. The leak rate of the synaptic weights is modulated based on the temporal correlation between the spiking patterns of the pre- and post-synaptic neurons. This mechanism helps in gradual forgetting of insignificant data while retaining significant, yet old, information. ASP, thus, maintains a balance between forgetting and immediate learning to construct a stable-plastic self-adaptive SNN for continuously changing inputs. We demonstrate that the proposed learning methodology addresses catastrophic forgetting, while yielding significantly improved accuracy over the conventional STDP learning method for digit recognition applications. In addition, we observe that the proposed learning model automatically encodes selective attention toward relevant features in the input data, while eliminating the influence of background noise (or denoising) further improving the robustness of the ASP learning.

Date: 01/03/2019
Time: 16:00
Location: D118

Rank order decoding of temporal input patterns

2019-02-20T13:02:57+00:00

Volker Steuber's journal club session, where he will briefly discuss different forms of neural coding.

In this talk, Volker will briefly discuss different forms of neural coding. Volker will focus on temporal coding and summarise John Hopfield's (1995) suggestion how neurons with subthreshold oscillations in their membrane potential could perform scale-invariant temporal encoding of input patterns (and which conditions have to be met for the encoding to be scale invariant). However, temporal coding is sensitive to noise, and Volker will describe the rank order decoding scheme that was suggested by Simon Thorpe (1998) in order to provide robustness against noise. Furthermore, Volker will outline how Purkinje cells in cerebellar cortex could implement a form of rank-order decoding of temporal parallel fibre input patterns.

Volker will reference the following papers:

"Pattern recognition computation using action potential timing for stimulus representation (J. J. Hopefield, 1995)"

"Speed of processing in the human visual system (S. Thorpe, D. Fize and C. Marlot, 1996)"

"Face processing using one spike per neurone (R. Van Rullen et al., 1998)"

"Rank order decoding of temporal parallel fibre input patterns in a complex Purkinje cell model (V. Steuber and E. De Schutter, 2002)"

Date: 22/02/2019
Time: 16:00
Location: D118

Modelling nicotine addiction

2019-02-14T13:58:52+00:00

Reinoud Maex's journal club session, where he will summerise the work he did in Paris, which was sponsored by Targacept: a pharmaceutical company which specialised in nicotinic compounds.

Nicotine and nicotinic compounds bind to acetylcholine receptors, but, surprisingly, it is largely unknown whether they act by activating or by desensitising (= inactivating) these receptors.

We simulated a simple circuit of dopaminergic and gaba-ergic neurons, each of them expressing different subtypes of nicotinic receptors.

Our main conclusion was that many nicotinic compounds act by desensitising their receptors. We also formulated a new hypothesis on nicotine addiction. We propose that smoking in addicted people my be a form of self-medication: by desensitising the nicotinic receptors on GABAergic neurons, nicotine disinhibits the dopamine-neurons and restores their physiological response to acetylcholine.

Date: 15/02/2019
Time: 16:00
Location: D118

Learning in Cephalopod Brains

2019-02-06T16:11:06+00:00

Damien Drix's journal club session, where he will present an overview of learning in Caphalopod brains, while referencing various papers.

Cephalopods are very much unlike the familiar model organisms of mammalian neuroscience; yet they are emerging as a promising model organism in comparative neurobiology and in bio-inspired robotics. This talk will start with a general overview of cephalopod brains and behaviour. We will then focus on learning in the vertical lobe and compare it with the mushroom body of insects.

Damien will reference the following papers:

"Computation in the learning system of cephalopods (J. Z. Young, 1991)"

"Functional and comparative assessments of the octopus learning and memory system (Binyamin Hochner, 2010)"

"Alternative sites of synaptic plasticity in two homologous "fan-out fan-in" learning and memory networks (Tal Shomrat et al., 2011)"

"An Embodied View of Octopus Neurobiology (Binyamin Hochner, 2012)"

Date: 08/02/2019
Time: 16:00
Location: D118

The power of deep networks and learning

2019-01-30T14:59:18+00:00

Shabnam Kadir's journal club session, where she will present the papers "The power of deeper networks for expressing natural functions (David Rolnick and Max Tegmark, 2018)" and "Why does deep and cheap learning work so well? (Henry W. Lin, Max Tegmark and David Rolnick, 2017)".

Abstracts:

"The power of deeper networks for expressing natural functions (David Rolnick and Max Tegmark, 2018)"

It is well-known that neural networks are universal approximators, but that deeper networks tend in practice to be more powerful than shallower ones. We shed light on this by proving that the total number of neurons m required to approximate natural classes of multivariate polynomials of n variables grows only linearly with n for deep neural networks, but grows exponentially when merely a single hidden layer is allowed. We also provide evidence that when the number of hidden layers is increased from 1 to k, the neuron requirement grows exponentially not with n but with n1/k, suggesting that the minimum number of layers required for practical expressibility grows only logarithmically with n."

"Why does deep and cheap learning work so well? (Henry W. Lin, Max Tegmark and David Rolnick, 2017)"

We show how the success of deep learning could depend not only on mathematics but also on physics: although well-known mathematical theorems guarantee that neural networks can approximate arbitrary functions well, the class of functions of practical interest can frequently be approximated through "cheap learning" with exponentially fewer parameters than generic ones. We explore how properties frequently encountered in physics such as symmetry, locality, compositionality, and polynomial log-probability translate into exceptionally simple neural networks. We further argue that when the statistical process generating the data is of a certain hierarchical form prevalent in physics and machine-learning, a deep neural network can be more efficient than a shallow one. We formalize these claims using information theory and discuss the relation to the renormalization group. We prove various "no-flattening theorems" showing when efficient linear deep networks cannot be accurately approximated by shallow ones without efficiency loss, for example, we show that n variables cannot be multiplied using fewer than 2^n neurons in a single hidden layer."

Date: 01/02/2019
Time: 16:00
Location: D118

Microcircuits and their interactions in epilepsy: is the focus out of focus?

2019-01-22T21:18:35+00:00

Julia Goncharenko's journal club session, where she will present the paper "Microcircuits and their interactions in epilepsy, is the focus out of focus? (Jeanne Paz and John Huguenard, 2015)".

Epileptic seizures represent dysfunctional neural networks dominated by excessive and/or hypersynchronous activity. Recent progress in the field has outlined two concepts regarding mechanisms of seizure generation, or ictogenesis. First, all seizures, even those associated with what have historically been thought of as ‘primary generalized’ epilepsies, appear to originate in local microcircuits and then propagate from that initial ictogenic zone. Second, seizures propagate through cerebral networks and engage microcircuits in distal nodes, a process that can be weakened or even interrupted by suppressing activity in such nodes. We describe various microcircuit motifs, with a special emphasis on one that has been broadly implicated in several epilepsies: feed-forward inhibition. Furthermore, we discuss how, in the dynamic network in which seizures propagate, focusing on circuit ‘choke points’ remote from the initiation site might be as important as that of the initial dysfunction, the seizure ‘focus’.

Date: 25/01/2019
Time: 16:00
Location: D118

Finding K-Means Clustering

2018-12-12T10:40:53+00:00

Deepak Panday's journal club session, where he will present the papers "Recovering the number of clusters in data sets with noise features using feature rescaling factors (Renato Cordeiro de Amorima and Christian Hennig, 2015)" and "Intelligent Choice of the Number of Clusters in K-Means Clustering An Experimental Study with Different Cluster Spreads (Mark Ming-Tso Chiang and Boris Mirkin, 2010)".

"Recovering the number of clusters in data sets with noise features using feature rescaling factors (Renato Cordeiro de Amorima and Christian Hennig, 2015)" abstract:

In this paper we introduce three methods for re-scaling data sets aiming at improving the likelihood of clustering validity indexes to return the true number of spherical Gaussian clusters with additional noise features. Our method obtains feature re-scaling factors taking into account the structure of a given data set and the intuitive idea that different features may have different degrees of relevance at different clusters. We experiment with the Silhouette (using squared Euclidean, Manhattan, and the pth power of the Minkowski distance), Dunn’s, Calinski–Harabasz and Hartigan indexes on data sets with spherical Gaussian clusters with and without noise features. We conclude that our methods indeed increase the chances of estimating the true number of clusters in a data set.

"Intelligent Choice of the Number of Clusters in K-Means Clustering An Experimental Study with Different Cluster Spreads (Mark Ming-Tso Chiang and Boris Mirkin, 2010)" abstract:

The issue of determining “the right number of clusters” in K-Means has attracted considerable interest, especially in the recent years. Cluster intermix appears to be a factor most affecting the clustering results. This paper proposes an experimental setting for comparison of different approaches at data generated from Gaussian clusters with the controlled parameters of between- and within-cluster spread to model cluster intermix. The setting allows for evaluating the centroid recovery on par with conventional evaluation of the cluster recovery. The subjects of our interest are two versions of the “intelligent” K-Means method, ik-Means, that find the “right” number of clusters by extracting “anomalous patterns” from the data one-by-one. We compare them with seven other methods, including Hartigan’s rule, averaged Silhouette width and Gap statistic, under different between- and within-cluster spread-shape conditions. There are several consistent patterns in the results of our experiments, such as that the right K is reproduced best by Hartigan’s rule – but not clusters or their centroids. This leads us to propose an adjusted version of iK-Means, which performs well in the current experiment setting.

Date: 14/12/2018
Time: 16:00
Location: LB250

On the Nernst-Planck equation

2018-11-13T17:21:23+00:00

Reinoud Maex's journal club session on his review: "On the Nernst-Planck equation (Reinoud Maex, 2017)"

Curious to know why the electro-diffusion equation is called the Nernst-Planck equation, I read the original papers by these authors a few years ago. This reading formed the basis of a little review paper of mine with the above title (published in Journal of Integrative Neuroscience 2017).

This review first discussed Nernst's and Planck's early papers on electro-diffusion, the brief priority conflict that followed, and the role these papers played in shaping the emerging concept of membrane excitability. The second part of this review discussed in greater detail the constraints of the Nernst-Planck theory, and showed more recent examples of its applicability to neuronal modelling.

Date: 16/11/2018
Time: 16:00
Location: D120

Computational Model of the Cerebellum and the Basal Ganglia for Interval Timing Learning

2018-11-05T14:49:43+00:00

Ohki Katakura's journal club session on his master's work "Computational Model of the Cerebellum and the Basal Ganglia for Interval Timing Learning (Ohki Katakura; Tadashi Yamazaki, 2016)"

In temporal information processing, both the cerebellum and the basal ganglia play essential roles. In particular, for interval timing learning, the cerebellum exhibits temporally localized activity around the onset of the unconditioned stimulus, whereas the basal ganglia represents the passage of time by their ramping-up activity from the onset of the conditioned stimulus to that of the unconditioned stimulus. We present a unified computational model of the cerebellum and the basal ganglia for the interval timing learning task. We report that our model reproduces the localized activity in the cerebellum and the gradual increase of the activity in the basal ganglia. These results suggest that the cerebellum and the basal ganglia play different roles in temporal information processing.

Date: 09/11/2018
Time: 16:00
Location: D120

Biocomputation Robots and ArchaeaBot Project

2018-10-31T13:25:13+00:00

Alex May and Anna Dumitriu's journal club session on their latest collaborative robotics artwork projects “ArchaeaBot” and “BioCompuation Bots”.

Anna Dumitriu and Alex May (Visiting Research Fellows: Artists in Residence in Computer Science at the University of Hertfordshire) will discuss their latest projects “ArchaeaBot” and “BioCompuation Bots” and demonstrate some of these new robotic artworks.

“ArchaeaBot: A Post Singularity and Post Climate Change Life-form” takes the form of an underwater robotic installation that explores what ‘life’ might mean in a post singularity, post climate change future. The project is based on new research about archaea (the oldest life forms on Earth) combined with machine learning & artificial intelligence to create the ‘ultimate’ species for the end of the world as we know it. The project has made in collaboration with researcher/cryomicroscopist Amanda Wilson as part of the EU FET Open H2020 funded MARA project based in the Beeby Lab at Imperial College London, and with Professor Daniel Polani from the School of Computer Science at the University of Hertfordshire. The project is supported through an EMAP/EMARE artists’ residency at LABoral Centro de Arte y Creación Industrial in Spain via funding from Creative Europe and with generous support from Arts Council England.

“BioCompuation Bots” are a brand-new series of artworks made collaboration with Professor Volker Steuber from the School of Computer Science at the University of Hertfordshire and artistically respond directly to research projects being undertaken within the university and with international collaborators in order to engage the public and international research community:

Two mouse-like quadruped robots explore research into controlling Petit Mal epilepsy using LEDs embedded in the brain that can ‘reset’ genetically modified cells before a fit occurs. In the research a complex data set was analysed to work out the perfect moment to turn on the fit stopping LED, in the artwork audiences use a blue torch to reset the robot’s ‘brains’ and ‘unfreeze’ them from their virtual fit. Also in collaboration with Dr Freek Hoebeek at Erasmus University in Rotterdam.

Another robot on tracked wheels roams around searching for smells that appeal to it and focusses on a kind of perfume designed to appeal to it. Senses such as smell are widely considered to be uniquely related to biological life and closely related to our understanding of consciousness, an assumption that this artwork throws into question. The robot explores artificial nose research being undertaken at the university with Dr Michael Schmucker.

Date: 30/11/2018
Time: 16:00
Location: D120

Olfactory coding in the turbulent realm

2018-02-12T14:36:04+00:00

Long-distance olfactory search behaviors depend on odor detection dynamics. Due to turbulence, olfactory signals travel as bursts of variable concentration and spacing and are characterized by long-tail distributions of odor/no-odor events, challenging the computing capacities of olfactory systems. How animals encode complex olfactory scenes to track the plume far from the source remains unclear. Here we focus on the coding of the plume temporal dynamics in moths. We compare responses of olfactory receptor neurons (ORNs) and antennal lobe projection neurons (PNs) to sequences of pheromone stimuli either with white-noise patterns or with realistic turbulent temporal structures simulating a large range of distances (8 to 64 m) from the odor source. For the first time, we analyze what information is extracted by the olfactory system at large distances from the source. Neuronal responses are analyzed using linear–nonlinear models fitted with white-noise stimuli and used for predicting responses to turbulent stimuli.

Date: 16/02/2018
Time: 16:00
Location: LB252

How the Olfactory Bulb processes naturalistic time-varying inputs

2017-10-11T18:31:43+01:00

Rebecca will be presenting the work she conducted for her master's thesis titled 'How the Olfactory Bulb processes naturalistic time-varying inputs'.

Abstract is below:

The olfactory bulb in mammals is responsible for receiving, processing and relaying olfactory information (odours). This project investigates how naturalistic temporally fluctuating odour signals are processed and which neurons or neural mechanisms are able to extract information from these signals. Multiple computation models were created to represent different OB circuits between periglomerular cells and mitral cells using NEURON (Hines and Carnevale, 2006, 2001). The results show that the strength and frequency of these odour signals can be determined by looking at a combination of the latency and the firing rates of the output from the mitral cells.

Date: 20/10/2017
Time: 16:00
Location: LB252