UH Biocomputation Group

Computational study of the effects of transcranial stimulation on the cerebellum

2025-06-11T12:10:52+01:00

On this week's Journal Club session, Eleonora Bernasconi will talk about her work in the presentation entitled "Computational study of the effects of transcranial stimulation on the cerebellum".

Transcranial magnetic stimulation (TMS) is a promising technique that could provide treatment for patients with cerebellar disorders. However, the underlying mechanisms of the action of TMS on the cerebellum are still unknown. Our goal is to study the effects of TMS on the cerebellum. We developed a model of the cerebellum. We stimulated all compartments of all neurons with a square wave, whose amplitude is solely dependent on the distance between the neuron compartment and the source of the stimulus. The stimulus was applied as a voltage using the extracellular mechanism in NEURON. We modelled the stimulus with frequencies of 1, 5, 10, 20 and 50 Hz. The higher the stimulus frequency, the bigger the amplitude of the modulation. For stimulus frequencies above 50 Hz, the Purkinje cell’s instantaneous firing rate no longer oscillates in accordance with the stimulus. We show that stimulus frequency can significantly impact the cell’s behaviour, highlighting the importance of carefully selecting this parameter in clinical settings. Future work will employ a more realistic simulation model.

Papers:

S. Sudhakar, S. Hong, I. Raikov, R. Publio, C. Lang, T. Close, D. Guo, M. Negrello, E. Schutter, "Spatiotemporal network coding of physiological mossy fiber inputs by the cerebellar granular layer", 2017, PLOS Computational Biology, 13, e1005754
Y. Zang, S. Dieudonné, E. De, Schutter, "Voltage- and Branch-Specific Climbing Fiber Responses in Purkinje Cells", 2018, Cell Reports, 24, 1536--1549
S. Diwakar, J. Magistretti, M. Goldfarb, G. Naldi, E. D'Angelo, "Axonal Na+ channels ensure fast spike activation and back-propagation in cerebellar granule cells", 2009, Journal of Neurophysiology, 101, 519--532

Date: 2025/06/13
Time: 14:00
Location: SP3011 & online

When the Robotic Maths Tutor is Wrong - Can Children Identify Mistakes Generated by ChatGPT?

2025-05-12T13:42:30+01:00

On this week's Journal Club session, Manal Helal will talk about her work in the presentation entitled "When the Robotic Maths Tutor is Wrong - Can Children Identify Mistakes Generated by ChatGPT?".

This study delves into integrating Large Language Models (LLMs), particularly ChatGPT- powered robots, as educational tools in primary school mathematics. Against the backdrop of Artificial Intelligence (AI) increasingly permeating educational settings, our investigation focuses on the response of young learners to errors made by these LLM- powered robots. Employing a user study approach, we conducted an experiment using the Pepper robot in a primary school classroom environment, where 77 primary school students from multiple grades (Year 3 to 5) took part in interacting with the robot. Our statistically significant findings highlight that most students, regardless of the year group, could discern between correct and incorrect responses generated by the robots, demonstrating a promising level of understanding and engagement with the AI-driven educational tool. Additionally, we observed that students' correctness in answering the Maths questions significantly influenced their ability to identify errors, underscoring the importance of prior knowledge in verifying LLM responses and detecting errors. Additionally, we examined potential confounding factors such as age and gender. Our findings underscore the importance of gradually integrating AI-powered educational tools under the guidance of domain experts following thorough verification processes. Moreover, our study calls for further research to establish best practices for implementing AI- driven pedagogical approaches in educational settings.

Papers:

M. Helal, P. Holthaus, L. Wood, V. Velmurugan, G. Lakatos, S. Moros, F. Amirabdollahian, "When the Robotic Maths Tutor is Wrong - Can Children Identify Mistakes Generated by ChatGPT?", 2024, 83--90
Helal M., "Beyond Accuracy: Primary Students’ Critical Engagement with LLM-Generated Mathematics - A Study of Error Detection, Trust Calibration, and Educational Implications", under review

Date: 2025/05/16
Time: 14:00
Location: SP3011 & online

Understanding structure-function mapping in small spiking neural networks.

2025-04-25T15:28:13+01:00

On this week's Journal Club session, Muhammad Yaqoob will talk about his work in the presentation entitled "Understanding structure-function mapping in small spiking neural networks.".

We know that the precise timing of spikes represents information in spiking neuronal networks. However, the information processing in spiking networks is not yet fully understood. One way to understand the working mechanism of a spiking network is to associate the structural connectivity of the network with the corresponding functional behaviour. In this journal club, I will begin with an introduction to different spiking neuron models and their properties. Then I’ll talk about the evolution of small spiking neural networks for temporal pattern recognition tasks. In the second half, we will discuss the structure-function mapping in spiking neural networks evolved or created to perform a computational task.

Papers:

M. Yaqoob, V. Steuber, B. Wróbel, "Autapses enable temporal pattern recognition in spiking neural networks", 2023, bioRxiv,
M. Yaqoob, V. Steuber, B. Wróbel, "The Importance of Self-excitation in Spiking Neural Networks Evolved to Recognize Temporal Patterns", 2019, 758--771
M. Yaqoob, V. Steuber, B. Wróbel, "Spiking Neural Networks as Finite State Transducers for Temporal Pattern Recognition", 2023, 51, S96--S97

Date: 2025/04/25
Time: 14:00
Location: SP3011 & online

Exploring the Use of Online Reviews in the Development of Video Game

2025-04-08T17:53:06+01:00

On this week's Journal Club session, Xinge Tong will talk about her PhD work in the presentation entitled "Exploring the Use of Online Reviews in the Development of Video Game".

It is now well established that online game reviews can play an important role in both reflecting game-player experiences and influencing business decisions regarding future video game development. Some studies have analysed different approaches to processing game reviews in the context of video game data development to assess their practical efficacy. However, most works in the field of game studies seem to rely on qualitative analysis approaches and software-related tools. These not only consume time and human resources but might also raise concerns about the difference in the focus of entertainment between software and games. The study described in this presentation sets out to explore the possibility of introducing natural language processing (NLP) into game review studies. It was noted that game reviews on Steam are a rich and valuable source. One of the main challenges of using NLP to process game reviews is how to extract effectively the diversity of meaning and accurately summarises the views of players. This study focuses on the accurate filtering of useful feedback information that is based on different contexts, the evaluation of user feedback, and the formulation of appropriate and effective game development strategies that incorporate this feedback. The main contribution of this study is the exploration and development of an NLP-based game review analysis system that: a) is able to process both topic and sentiment classification, b) makes highly accurate predictions and classifications to reviews by adopting advanced machine learning models with training on a newly produced game specific dataset, c) can be adopted and incorporated to the practical video game development lifecycle to process various tasks and meet developers’ diverse needs.

Date: 2025/04/11
Time: 14:00
Location: SP3011 & online

Detecting Anxiety via Machine Learning Algorithms: A Literature Review

2025-03-25T12:28:32+00:00

On this week's Journal Club session, Shamim Ibne Shahid will talk about his review paper "Detecting Anxiety via Machine Learning Algorithms: A Literature Review".

Recent machine learning (ML) advances have opened up new possibilities for addressing various challenges. Given their ability to tackle complex problems, the use of ML algorithms in diagnosing mental health disorders has seen substantial growth in both the number and scope of studies. Anxiety, a major health concern in today's world, affects a significant portion of the population. Individuals with anxiety often exhibit distinct characteristics compared to those without the disorder. These differences can be observed in their outward appearance—such as voice, facial expressions, gestures, and movements—and in less visible factors like heart rate, blood test results, and brain imaging data. In this context, numerous studies have utilized ML algorithms to extract a diverse range of features from individuals with anxiety, aiming to build predictive models capable of accurately identifying those affected by the disorder. This paper performs a comprehensive literature review on the state-of-the-art studies that employ machine learning algorithms to identify anxiety. This paper aims to cover a wide range of studies and categorize them based on their methodologies and data types used.

Papers:

M. Tayarani-N., S. Shahid, "Detecting Anxiety via Machine Learning Algorithms: A Literature Review", 2025, IEEE Transactions on Emerging Topics in Computational Intelligence, 1--24

Date: 2025/03/28
Time: 14:00
Location: SP3011 & online

Combinatorial threshold linear networks and how dynamics can be ascertained via graph motifs in the connectivity matrix.

2025-03-12T22:00:58+00:00

On this week's Journal Club session, Shabnam Kadir will talk about "Combinatorial threshold linear networks and how dynamics can be ascertained via graph motifs in the connectivity matrix.".

I shall talk about combinatorial threshold linear networks and how dynamics can be ascertained via graph motifs in the connectivity matrix. I shall explore ways in which it could be connected to Yaqoob’s work:

M. Yaqoob, V. Steuber, B. Wróbel, "Autapses enable temporal pattern recognition in spiking neural networks", 2023, bioRxiv.

Papers:

C. Curto, K. Morrison, "Graph rules for recurrent neural network dynamics: extended version", 2023, arXiv,
M. Yaqoob, V. Steuber, B. Wróbel, "Autapses enable temporal pattern recognition in spiking neural networks", 2023, bioRxiv,
C. Curto, J. Geneson, K. Morrison, "Stable fixed points of combinatorial threshold-linear networks", 2023, arXiv,

Date: 2025/03/14
Time: 14:00
Location: SP3011 & online

Exploring Convergence in Wearable Computing Development

2025-02-26T11:37:04+00:00

On this week's Journal Club session, Caroline McMillan will talk about her work in the presentation entitled "Exploring Convergence in Wearable Computing Development".

Wearable technologies draw on a range of disciplines. Due to methodological differences, wearables researchers can experience gaps or breakdowns in values, goals, and vocabulary when collaborating. This situation makes wearables development challenging, even more so when technologies are in the early stages of development, and their technological and cultural potential is not fully understood. Investigating a common ground to enhance convergent spaces in research allows researchers and developers to share information across domains, encouraging divergent perspectives, creativity, and inclusiveness. By presenting an example of an online search interface that allows users to explore wearable technologies beyond their discipline, the authors show how users with different mindsets and skills can engage with information and expand and share knowledge when developing wearables.

Papers:

L. Paredes, C. McMillan, W. Chan, S. Chandrasegaran, R. Singh, K. Ramani, D. Wilde, "CHIMERA: Supporting Wearables Development across Multidisciplinary Perspectives", 2022, Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., 5, 174:1--174:24

Date: 2025/02/28
Time: 14:00
Location: online

Optimising burst based deep brain stimulation in a population model of Parkinson's disease and Tremor

2024-12-09T12:44:07+00:00

On this week's Journal Club session, Nada Yousif will talk about her work in the presentation entitled "Optimising burst based deep brain stimulation in a population model of Parkinson's disease and Tremor".

Deep brain stimulation (DBS) is a therapy used to treat several neurological conditions such as Parkinson’s disease (PD) and essential tremor (ET), based on the implantation of electrodes into specific brain targets. Despite its success, the electrical stimulus has remained as a regular frequency square pulse for decades. Recently, there have been proposals that phase-locking, coordinated reset or irregular stimulation patterns may be more effective at desynchronising the pathological neural activity. Here we consider our population level model of the thalamocortical-basal ganglia network, which generates pathological oscillatory activity in the beta band (textasciitilde20 Hz) associated with PD and the tremor band (textasciitilde4 Hz) associated with ET. We stimulate the model with regular, irregular and phase-dependent bursts of DBS, and use an optimisation technique to find the best stimulation parameters in each case. Our results show that bursts can be as or more effective at suppressing pathological oscillations compared to continuous DBS, allowing exploration of stimulation mechanisms to formulate testable predictions regarding DBS.

Date: 2024/12/13
Time: 14:00
Location: Online

An ensemble learning algorithm for optimization of spark ignition engine performance fuelled with methane/hydrogen blends

2024-11-19T12:41:37+00:00

On this week's Journal Club session, Mohammad Tayaraninajaran will talk about his paper "An ensemble learning algorithm for optimization of spark ignition engine performance fuelled with methane/hydrogen blends".

The increasing global demand for sustainable and cleaner transportation has led to extensive research on alternative fuels for Internal Combustion (IC) engines. One promising option is the utilization of methane/hydrogen blends in Spark-Ignition (SI) engines due to their potential to reduce Green House Gas (GHG) emissions and improve engine performance. However, the optimal operation of such an engine is challenging due to the interdependence of multiple conflicting objectives, including Brake Mean Effective Pressure (BMEP), Brake Specific Fuel Consumption (BSFC), and nitrogen oxide (NOx) emissions. This paper proposes an evolutionary optimization algorithm that employs a surrogate model as a fitness function to optimize methane/hydrogen SI engine performance and emissions. To create the surrogate model, we propose a novel ensemble learning algorithm that consists of several base learners. This paper employs ten different learning algorithms diversified via the Wagging method to create a pool of base-learner algorithms. This paper proposes a combinatorial evolutionary pruning algorithm to select an optimal subset of learning algorithms from a pool of base learners for the final ensemble algorithm. Once the base learners are designed, they are incorporated into an ensemble, where their outputs are aggregated using a weighted voting scheme. The weights of these base learners are optimized through a gradient descent algorithm. However, when optimizing a problem using surrogate models, the fitness function is subject to approximation uncertainty. To address this issue, this paper introduces an uncertainty reduction algorithm that performs averaging within a sphere around each solution. Experiments are performed to compare the proposed ensemble learning algorithm to the classical learning algorithms and state-of-the-art ensemble algorithms. Also, the proposed smoothing algorithm is compared with the state-of-the-art evolutionary algorithms. Experimental studies suggest that the proposed algorithms outperform the existing algorithms.

Papers:

M. Tayarani-N., A. Paykani, "An ensemble learning algorithm for optimization of spark ignition engine performance fuelled with methane/hydrogen blends", 2024, Applied Soft Computing, 112468

Date: 2024/11/22
Time: 14:00
Location: SP4024A & online

A mechanism for ATP homeostasis based on activity-dependent synthesis and release of neurotransmitter

2024-10-16T14:44:02+01:00

On this week's Journal Club session, Reinoud Maex will talk about his work in the presentation entitled "A mechanism for ATP homeostasis based on activity-dependent synthesis and release of neurotransmitter".

It has long been recognised that many neurotransmitters bear a close relationship, for their synthesis or degradation, to compounds of the mitochondrial Krebs cycle, the central pathway for the production of the energy intermediary adenosine triphosphate (ATP). How neurons maintain an almost constant level of ATP in their cytosol, in spite of varying workloads, is still an outstanding question. Taking the relationship between glutamate (Glu) and alpha-ketoglutarate as an example, a simple metabolic model was built of a Glu- releasing axon terminal. The model included the interconversion between Glu and alpha- ketoglutarate, the accumulation of Glu in synaptic vesicles, and the recycling of Glu through an activity-induced supply of its precursor glutamine by the perisynaptic astrocyte. Assuming near-equilibrium conditions, the model was reduced to two differential equations of the cytosolic concentration of Glu and ATP. This model showed remarkable characteristics: at steady state, irrespective of the workload, the ATP level measured 1.5 mM, and only 4.7 % of the ATP was spent on Glu release and recycling. The speed of ATP homeostasis was determined by the (absolute) amount of ATP the axon spent on vesicular Glu accumulation. Overall, homeostasis came at little cost as it prevented futile ATP production when the workload decreased. These analytical and numerical results indicate that neurotransmitter release may confer to neurons a metabolic advantage in the form of a more accurate ATP homeostasis.

Date: 2024/10/25
Time: 14:00
Location: SP4024A & online

Topological Data Analysis Reveals Brain Connectivity Differences between Schizophrenia Subjects and Healthy Controls

2024-07-02T15:38:38+01:00

On this week's Journal Club session, Emil Dmitruk will talk about his work in the presentation entitled "Topological Data Analysis Reveals Brain Connectivity Differences between Schizophrenia Subjects and Healthy Controls".

The aim of this study is to reveal structural connectivity patterns prevalent across a large population and whether any of these patterns differ between schizophrenia subjects (SCH) and healthy controls (HC). We use topological data analysis methods, namely persistent homology using the weight rank clique filtration on brain connectivity matrices obtained from a probabilistic fibre-tracking algorithm run on the Centre for Biomedical Research Excellence (COBRE) dataset (N = 44 SCH, N = 44 HC) to investigate group differences. We show that some of the connectivity structures differ in strength relative to connectivity structures affecting other brain regions. These differences would not have been apparent using traditional methods that explore only the absolute strength of connectivity between regions. We show that many connectivity structures (cycles) are shared among numerous subjects and some of the cycles are significantly different between the two studied populations- both on a whole brain level (all cycles combined), and also at the level of individual cycle classes. Brain regions involved in cycles that were found were mentioned in studies of brain alteration in schizophrenia, e.g. precuneus, motor, visual, and parietal cortices (the last being tied to experiences of psychosis). We show that HC and SCH subjects share a lot in their structural connectome and the effect of schizophrenia is an alteration in the relative strength, as measured by topological persistence, of various white matter connectivity structures. We believe that further studies could lead to prognostication for individual patients and be the first step to developing personalised treatment.

Date: 2024/07/05
Time: 14:00
Location: C258 & online

Visual experience, topology, and the perception of art

2024-06-11T23:12:40+01:00

On this week's Journal Club session, Shabnam Kadir will talk about her work in the presentation entitled "Visual experience, topology, and the perception of art".

The brain has access to visual information via a variety of neural codes, e.g. there are cells tuned to edge orientation, spatial frequency, edges, corners, contrasts and colour. Higher cortical areas of the brain are thought to then further process visual information as well as integrate a visual scene with other senses and the conscious mind. How this visual information is combined with experience to form a holistic experience that elicits an emotional response of pleasure, distress and/or other forms of meaning is an active research question. The human perception of art and the question of what constitutes art touch upon all these elements of perception and integration. We shall show how topology provides a vital new lens in examining these questions.

We discuss recent work where methods from applied topology, namely persistent homology of certain cubical complexes, are applied to images produced by both a human artist and a neural network shown in two exhibitions in Torun, Poland. The images shown in the two exhibitions were matched for certain information-theoretic pixel-based characteristics. Experimental measurements of EEG, tracking of eye movement, as well as conscious perception/appreciation of these paintings were collected from a selection of visitors to these exhibitions, namely second-year art students.

Date: 2024/06/14
Time: 14:00
Location: C258 & online

Transforming Data into Knowledge: How to Build Knowledge Graphs

2024-05-28T20:00:35+01:00

On this week's Journal Club session, Yi Sun will talk about her work in the presentation entitled "Transforming Data into Knowledge: How to Build Knowledge Graphs".

In our recent KTP project, we need to build a knowledge graph. In my presentation, I will outline a general procedure for constructing a knowledge graph based on the following references:

Papers:

P. Chandak, K. Huang, M. Zitnik, "Building a knowledge graph to enable precision medicine", 2023, Scientific Data, 10, 67
M. Bravo, H. Luis, J. Reyes, "Methodology for ontology design and construction", 2019, Contaduría y Administración, 64, 134
G. Tamašauskaitė, P. Groth, "Defining a Knowledge Graph Development Process Through a Systematic Review", 2023, ACM Transactions on Software Engineering and Methodology, 32, 27:1--27:40

Date: 2024/05/31
Time: 14:00
Location: C258 & online

A 3D atlas-based model for visualising brain nuclei targeted by high-intensity focused ultrasound in the treatment of tremor

2024-04-20T19:23:50+01:00

On this week's Journal Club session, Nada Yousif will talk about her paper under review in the presentation entitled "A 3D atlas-based model for visualising brain nuclei targeted by high-intensity focused ultrasound in the treatment of tremor".

Essential tremor is a common movement disorder which impacts significantly on activities of daily living. One treatment is high intensity focused ultrasound (FUS) is a non- invasive technique which thermally ablates the ventral intermediate (VIM) nucleus of the thalamus. However, the VIM is a small nucleus, impossible to visualise on clinical MRI. Therefore, we constructed a three-dimensional computational model based on data from a brain atlas to improve the visualisation of VIM with the aim of optimising targeting and post-treatment analysis. We show that the clinical targeting approach and actual lesion locations can be plotted in the model to understand their relation to the neuroanatomy. Future directions will also be discussed with respect to imaging-based models and modelling the impact of heating and ablation on thalamic neurons and networks.

Date: 2024/04/26
Time: 14:00
Location: online

Complexity of Reachability and Mortality for Low-dimensional Dynamical Systems

2024-04-14T17:42:53+01:00

On this week's Journal Club session, Olga Tveretina will talk about her work in the presentation entitled "Complexity of Reachability and Mortality for Low-dimensional Dynamical Systems".

Theory of dynamical systems provides a powerful framework for understanding complex dynamics, and its applications span a wide range of fields, including biological systems.

The reachability problem involves determining whether a given state or configuration of a system can be reached from another state through a sequence of transitions or actions. It is a fundamental question in computer science and has numerous applications across various domains. Thus, reachability analysis applied in systems biology helps to model and analyze biological networks such as gene regulatory networks, protein interaction networks, and metabolic pathways.

The mortality problem can be stated as follows: given a dynamical system, is it the case that all trajectories of the system are mortal? The mortality problem is relevant to the field of program termination, and it has been studied in different contexts and in different variants.

In this talk, I will present my current work on the computational complexity of reachability and mortality for specific classes of low- dimensional dynamical systems. Areas where variations of such systems arise include, among others, biological systems (gene regulatory networks), robotics (the configuration space of a robotic arm), and learning algorithms (finding a low-dimensional parameterization of high-dimensional data).

Papers:

M. Oliveira, Oliveira, O. Tveretina, "Mortality and Edge-to-Edge Reachability are Decidable on Surfaces", 2022, Hybrid Systems: Computation and Control, 1--10
P. Bell, S. Chen, L. Jackson, "On the decidability and complexity of problems for restricted hierarchical hybrid systems", 2016, Theoretical Computer Science, 652, 47--63

Date: 2024/04/19
Time: 14:00
Location: C258 & online

Attention in Neural Networks

2024-04-08T12:20:04+01:00

On this week's Journal Club session, Na Helian will talk about "Attention in Neural Networks".

Based on the following papers, I will introduce different types of attention techniques for natural language and image processing applications.

Papers:

A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. Gomez, Ł. Kaiser, I. Polosukhin, "Attention is All you Need", 2017, Advances in Neural Information Processing Systems, 30,
A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, N. Houlsby, "An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale", 2021, International Conference on Learning Representations,
Q. Wang, B. Wu, P. Zhu, P. Li, W. Zuo, Q. Hu, "ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks", 2020, IEEE/CVF Conference on Computer Vision and Pattern Recognition,
S. Woo, J. Park, J. Lee, I. Kweon, "CBAM: Convolutional Block Attention Module", 2018, European Conference on Computer Vision,

Date: 2024/04/12
Time: 14:00
Location: C258 & online

Metabolic activity organizes olfactory representations

2024-04-02T23:21:03+01:00

On this week's Journal Club session, Jim Bower will talk about the paper "Metabolic activity organizes olfactory representations".

Hearing and vision sensory systems are tuned to the natural statistics of acoustic and electromagnetic energy on earth and are evolved to be sensitive in ethologically relevant ranges. But what are the natural statistics of odors, and how do olfactory systems exploit them? Dissecting an accurate machine learning model (Lee et al., 2022) for human odor perception, we find a computable representation for odor at the molecular level that can predict the odor-evoked receptor, neural, and behavioral responses of nearly all terrestrial organisms studied in olfactory neuroscience. Using this olfactory representation (principal odor map [POM]), we find that odorous compounds with similar POM representations are more likely to co-occur within a substance and be metabolically closely related; metabolic reaction sequences (Caspi et al., 2014) also follow smooth paths in POM despite large jumps in molecular structure. Just as the brain’s visual representations have evolved around the natural statistics of light and shapes, the natural statistics of metabolism appear to shape the brain’s representation of the olfactory world.

Papers:

P. Wang, Y. Sun, R. Axel, L. Abbott, G. Yang, "Evolving the olfactory system with machine learning", 2021, Neuron, 109, 3879--3892.e5
W. Qian, J. Wei, B. Sanchez-Lengeling, B. Lee, Y. Luo, M. Vlot, K. Dechering, J. Peng, R. Gerkin, A. Wiltschko, "Metabolic activity organizes olfactory representations", 2023, eLife, 12, e82502
B. K. Lee et al., "A Principal Odor Map Unifies Diverse Tasks in Human Olfactory Perception", 2022, bioRxiv, 504602

Date: 2024/04/05
Time: 14:00
Location: online

CEPS: An Open Access MATLAB Graphical User Interface (GUI) for the Analysis of Complexity and Entropy in Physiological Signals

2024-03-02T19:04:26+00:00

On this week's Journal Club session, Deepak Panday will talk about his paper in the presentation entitled "CEPS: An Open Access MATLAB Graphical User Interface (GUI) for the Analysis of Complexity and Entropy in Physiological Signals".

Background: We developed CEPS as an open access MATLAB® GUI (graphical user interface) for the analysis of Complexity and Entropy in Physiological Signals (CEPS), and demonstrate its use with an example data set that shows the effects of paced breathing (PB) on variability of heart, pulse and respiration rates. CEPS is also sufficiently adaptable to be used for other time series physiological data such as EEG (electroencephalography), postural sway or temperature measurements. Methods: Data were collected from a convenience sample of nine healthy adults in a pilot for a larger study investigating the effects on vagal tone of breathing paced at various different rates, part of a development programme for a home training stress reduction system. Results: The current version of CEPS focuses on those complexity and entropy measures that appear most frequently in the literature, together with some recently introduced entropy measures which may have advantages over those that are more established. Ten methods of estimating data complexity are currently included, and some 28 entropy measures. The GUI also includes a section for data pre- processing and standard ancillary methods to enable parameter estimation of embedding dimension m and time delay τ (‘tau’) where required. The software is freely available under version 3 of the GNU Lesser General Public License (LGPLv3) for non-commercial users. CEPS can be downloaded from Bitbucket. In our illustration on PB, most complexity and entropy measures decreased significantly in response to breathing at 7 breaths per minute, differentiating more clearly than conventional linear, time- and frequency-domain measures between breathing states. In contrast, Higuchi fractal dimension increased during paced breathing. Conclusions: We have developed CEPS software as a physiological data visualiser able to integrate state of the art techniques. The interface is designed for clinical research and has a structure designed for integrating new tools. The aim is to strengthen collaboration between clinicians and the biomedical community, as demonstrated here by using CEPS to analyse various physiological responses to paced breathing.

Papers:

D. Mayor, D. Panday, H. Kandel, T. Steffert, D. Banks, "CEPS: An Open Access MATLAB Graphical User Interface (GUI) for the Analysis of Complexity and Entropy in Physiological Signals", 2021, Entropy, 23, 321

Date: 2024/03/08
Time: 14:00
Location: C258 & online

An analytical model of the energy of ion mixing in the brain

2024-02-28T21:27:37+00:00

On this week's Journal Club session, Reinoud Maex will talk about his work in the presentation entitled "An analytical model of the energy of ion mixing in the brain".

Neurons store energy in the ionic concentration gradients they build between their cell interior and the extracellular space. They can subsequently use this energy to do work, such as charging their membrane capacitor, transporting metabolites across their cell membrane, etc. During hyperactivity or energy deprivation the ions mix and the neurons swell. In this presentation I will show how a single principle, namely that neurons try to maximize the energy stored, can explain three unrelated phenomena: a) why extracellular space occupies a volume fraction of about 20%, b) why neurons swell during energy deprivation, and c) why calcium ions are so heavily buffered.

Papers:

R. Maex, "Effect of extracellular volume on the energy stored in transmembrane concentration gradients", 2021, Physical Review E, 104, 044409
R. Maex, "An Isotonic Model of Neuron Swelling Based on Co-Transport of Salt and Water", 2023, Membranes, 13, 206
R. Maex, "Energy optimisation predicts the capacity of ion buffering in the brain", 2023, Biological Cybernetics, 117, 467--484

Date: 2024/03/01
Time: 14:00
Location: C258 & online

How the cerebellum recognises learnt and novel patterns: computational approach with a biologically detailed network model

2024-02-19T17:14:00+00:00

On this week's Journal Club session, Ohki Katakura will talk about his work in the talk entitled "How the cerebellum recognises learnt and novel patterns: computational approach with a biologically detailed network model".

The cerebellum is essential for motor control, timing and cognition. Although its anatomy and physiology have been investigated, recent experimental studies raise new open questions. These include how the granule cells process mossy fibre signals and how the Purkinje cells code learnt and novel patterns. In this research, a detailed cerebellar cortex network model was constructed by incorporating existing models [1,2] and introducing long-term depression at granule cell-Purkinje cell synapses. The network connectivity switches between the oscillatory and non-oscillatory activity states, affecting the sparsity of activated granule cells. Pattern recognition criteria differed across these states: in the oscillatory network, novel prompted longer pauses in Purkinje cell spikes, while the non-oscillatory network responded with longer bursts. The number of storable patterns in the network corresponds to the sparsity of activation.

Papers:

E. De, Schutter, J. Bower, "An active membrane model of the cerebellar Purkinje cell. I. Simulation of current clamps in slice", 1994, Journal of Neurophysiology, 71, 375--400
S. Sudhakar, S. Hong, I. Raikov, R. Publio, C. Lang, T. Close, D. Guo, M. Negrello, E. Schutter, "Spatiotemporal network coding of physiological mossy fiber inputs by the cerebellar granular layer", 2017, PLOS Computational Biology, 13, e1005754

Date: 2024/02/23
Time: 14:00
Location: C258 & online

Exploitation of footballer analytics

2024-02-10T17:50:05+00:00

On this week's Journal Club session, Edward Wakelam will give a talk entitled "Exploitation of footballer analytics".

There is growing on-going research into how footballer attributes, collected prior to, during and post-match, may address the demands of clubs, media pundits and gaming developers. These attributes include static, such as age and height, or dynamic, such as pass completions and shots on target. Many are objectively measured, such as goals scored or pass interceptions, but in some cases they are subjectively measured by observers, such as quality of passing or athleticism. A handful of databases are available to clubs to be used to assess transfer targets or to assess their current players, however even elite clubs are failing to exploit this data. Additionally, there is limited application of AI and Machine Learning methods to exploit the wealth of data available. Critically, very few of the attributes address character traits, despite their use by the commercial world for 130 years and some well publicised problem player recruitment. I will present an analysis of the data available collected from available research over the past 25 years, identifying the mix of static vs dynamic and objective vs subjective data, and draw attention to the limited character trait content of the data. I will discuss the likely reason for this and how clubs may develop this dimension of their player analytics.

Papers:

E. Wakelam, V. Steuber, J. Wakelam, "The collection, analysis and exploitation of footballer attributes: A systematic review", 2022, Journal of Sports Analytics, 8, 31--67

Date: 2024/02/16
Time: 14:00
Location: C258 & online

Does combinatorial coding hinder the understanding of structure odour relationships in human olfaction?

2024-01-28T15:52:27+00:00

On this week's Journal Club session, Simon O'Connor will give the talk "Does combinatorial coding hinder the understanding of structure odour relationships in human olfaction?".

Since Malcolm Dyson first suggested a vibrational theory of smell in 1928 proponents of the competing theories of olfaction have struggled for nearly a century to make sense of how the chemical structure of odour molecules elicit the perception of odours. Here I will give a short introduction to the olfactory system. I will then discuss a dataset that I constructed from fragrance industry catalogues, Infrared Spectra simulations using ‘Gaussian’ modelling software and RDkit bit vectors based on the structure of the molecules. Next, I will describe KNN and Random Forrest studies in which I probe the dataset before moving on to hierarchical agglomerative cluster analysis. I will then talk about Hyperbolic Hierarchical Clustering (HypHC) software and whether an embeddings-based approach might be the right approach considering the combinatorial nature of the receptor output.

Papers:

B. Lee, E. Mayhew, B. Sanchez-Lengeling, J. Wei, W. Qian, K. Little, M. Andres, B. Nguyen, T. Moloy, J. Parker, R. Gerkin, J. Mainland, A. Wiltschko, "A Principal Odor Map Unifies Diverse Tasks in Human Olfactory Perception", 2022, bioRxiv,

Date: 2024/02/02
Time: 14:00
Location: C258 & online

A probabilistic meta-heuristic optimisation algorithm for image multi-level thresholding

2024-01-22T14:06:02+00:00

On this week's Journal Club session, Mohammad Tayaraninajaran will talk about his paper "A probabilistic meta-heuristic optimisation algorithm for image multi-level thresholding".

The spread of the Severe Acute Respiratory Syndrome CoronaVirus 2 (SARS-CoV-2) which causes CoronaVirus Disease 2019 (COVID-19) has challenged many countries. To curb the effect of the pandemic requires the development of low-cost and rapid tools for detecting and diagnosing the patients. In this regard, chest X-ray scan images provide a reliable way of detecting the patients. One limitation, however, is the need for experts to analyse the images and identify the cases which can be a burden, when a large number of images are to be processed. The aim of this paper is to propose a method to extract rapidly, from the X-ray images, the regions in which there exist indications of COVID-19 infection. To identify the regions, image segmentation is required which is performed in this paper with a novel optimization algorithm. The proposed optimization algorithm uses probabilistic representation for the solutions. To improve the optimization process, we propose a diversity preserving operator. For multi-level image thresholding via optimization algorithms, different fitness functions have been proposed in the literature. In the proposed method in this paper, we use three fitness functions to benefit from the advantages of all. A fitness swapping scheme is proposed which swaps between the fitness functions in the optimization process. Also, a diversity preserving operator is proposed in this paper which compares the individuals and reinitializes the similar ones to inject diversity in the population. The proposed algorithm is tested on a number of COVID-19 benchmark images and experimental analysis suggest better performance for the proposed algorithm.

Papers:

M. Najaran, "A probabilistic meta-heuristic optimisation algorithm for image multi-level thresholding", 2023, Genetic Programming and Evolvable Machines, 24, 14

Date: 2024/01/26
Time: 14:00
Location: C258 & online

Rehabilitation Robotics Using Cerebellar Controller

2023-12-11T10:53:57+00:00

On this week's Journal Club session, Mahsa Aliakbarzadeh will talk about her work in the talk entitled "Rehabilitation Robotics Using Cerebellar Controller".

Stroke is a major cause of disability in adults. More than 15 million strokes occur every year in the world, and more than 100,000 of these affect patients in the UK. Stroke patients often have an impaired ability to control their upper limbs and need assistance with every-day tasks. Relearning motor skills after stroke is like learning new motor skills, but a problem for stroke survivors is that their impaired movements often restrict the ability to use sensory feedback for re-learning. Rehabilitation robotics has shown promise to augment the rehabilitation process and to offer feedback on performance. However, the personalisation of the therapy to individual needs remains a major challenge to date. I will use a computational model of the cerebellum that is to optimise robotic rehabilitation for individual subjects. The cerebellum has been optimised throughout vertebrate evolution to become an adaptive controller of biological skeletomuscular structures that is unrivalled by any artificial adaptive motor control algorithm.

Date: 2023/12/15
Time: 14:00
Location: online

Computer Vision and Image Analysis for Industry 4.0

2023-12-04T16:02:38+00:00

On this week's Journal Club session, Shamim Ibne Shahid will talk about the book "Computer Vision and Image Analysis for Industry 4.0".

Eight years ago, since the Omniglot data was first released, very few papers have addressed the original Omniglot challenge, which is to carry out within-alphabet one-shot classification tasks as opposed to selecting the test samples between the al- phabets. Most researchers have made the task easier by introducing new splits in the dataset and have taken advantage of significant sample and class augmentation. Amongst the deep learning models that have adopted the Omniglot challenge as it is, the Recursive Cortical network has the highest performance of 92.75%. In this presentation , I will talk about a new similarity function to aid in the training procedure of matching network, which helps achieve 95.75% classification accuracy on the Omniglot challenge without requiring any data augmentation.

Reference:

N.Siddique, M. S. Arefin, M. A. R. Ahad, and M. A. A. Dewan, "Computer Vision and Image Analysis for Industry 4.0". CRC Press, 2023.

Date: 2023/12/08
Time: 14:00
Location: online

Enhancing Deep Learning Models through Tensorization: A Comprehensive Survey and Framework

2023-11-27T13:36:16+00:00

On this week's Journal Club session, Manal Helal will talk about her under-review draft paper "Enhancing Deep Learning Models through Tensorization: A Comprehensive Survey and Framework".

The burgeoning growth of public domain data and the increasing complexity of deep learning model architectures have underscored the need for more efficient data representation and analysis techniques. This paper is motivated by the work of (Helal, 2023) and aims to present a comprehensive overview of tensorization. This transformative approach bridges the gap between the inherently multidimensional nature of data and the simplified 2-dimensional matrices commonly used in linear algebra-based machine learning algorithms. This paper explores the steps involved in tensorization, multidimensional data sources, various multiway analysis methods employed, and the benefits of these approaches. A small example of Blind Source Separation (BSS) is presented comparing 2-dimensional algorithms and a multiway algorithm in Python. Results indicate that multiway analysis is more expressive. Contrary to the intuition of the dimensionality curse, utilising multidimensional datasets in their native form and applying multiway analysis methods grounded in multilinear algebra reveal a profound capacity to capture intricate interrelationships among various dimensions while, surprisingly, reducing the number of model parameters and accelerating processing. A survey of the multi-away analysis methods and integration with various Deep Neural Networks models is presented using case studies in different application domains.

References:

M. Helal, "Enhancing Deep Learning Models through Tensorization: A Comprehensive Survey and Framework", 2023, arXiv,
M. Helal, "Introduction to Tensor Computing in Python: From First Principles to Deep Learning" Book published by Amazon Publishing PROS, 2023. IBSN:978-1-916626-33-1

Date: 2023/12/01
Time: 14:00
Location: 2J124 & online

Olfaction prediction using 3DCNN and molecular electron density.

2023-11-20T16:09:59+00:00

On this week's Journal Club session, Pinaki Saha will talk about his work in the talk entitled "Olfaction prediction using 3DCNN and molecular electron density.".

Olfaction, or the sense of smell, is a complex and poorly understood phenomenon that involves the interaction of odorant molecules with olfactory receptors. The molecular structure and properties of odorants determine their affinity and specificity for different olfactory receptors, and thus their perceived odor quality and intensity. In this presentation, I talk about a novel approach for olfaction prediction based on 3D convolutional neural networks (3D CNNs) and promolecule electron density of odorant molecules.

Date: 2023/11/24
Time: 14:00
Location: online

On Instance Weighted Clustering Ensembles

2023-11-13T14:40:44+00:00

On this week's Journal Club session, Paul Moggridge will talk about his conference paper entitled "On Instance Weighted Clustering Ensembles".

Ensemble clustering is a technique which combines multiple clustering results, and instance weighting is a technique which highlights important instances in a dataset. Both techniques are known to enhance clustering performance and robustness. In this research, ensembles and instance weighting are integrated with the spectral clustering algorithm. We believe this is the first attempt at creating diversity in the generative mechanism using density-based instance weighting for a spectral ensemble. The proposed approach is empirically validated using synthetic datasets comparing against spectral and a spectral ensemble with random instance weighting. Results show that using the instance weighted sub-sampling approach as the generative mechanism for an ensemble of spectral clustering leads to improved clustering performance on datasets with imbalanced clusters.

Papers:

P. Moggridge, N. Helian, Y. Sun, M. Lilley, "On Instance Weighted Clustering Ensembles", 2023, ESANN 2023 - proceedings

Date: 2023/11/17
Time: 14:00
Location: LJ115 & online

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

Exploring Computer Science Students’ Perception of ChatGPT in Higher Education: A Descriptive and Correlation Study

2023-10-30T13:10:44+00:00

On this week's Journal Club session, Harpreet Singh will talk about the paper entitled "Exploring Computer Science Students’ Perception of ChatGPT in Higher Education: A Descriptive and Correlation Study".

ChatGPT is an emerging tool that can be employed in many activities including in learning/teaching in universities. Like many other tools, it has its benefits and its drawbacks. If used properly, it can improve learning, and if used irresponsibly, it can have a negative impact on learning. The aim of this research is to study how ChatGPT can be used in academia to improve teaching/learning activities. In this paper, we study students’ opinions about how the tool can be used positively in learning activities. A survey is conducted among 430 students of an MSc degree in computer science at the University of Hertfordshire, UK, and their opinions about the tool are studied. The survey tries to capture different aspects in which the tool can be employed in academia and the ways in which it can harm or help students in learning activities. The findings suggest that many students are familiar with the tool but do not regularly use it for academic purposes. Moreover, students are skeptical of its positive impacts on learning and think that universities should provide more vivid guidelines and better education on how and where the tool can be used for learning activities. The students’ feedback responses are analyzed and discussed and the authors’ opinions regarding the subject are presented. This study shows that ChatGPT can be helpful in learning/teaching activities, but better guidelines should be provided for the students in using the tool.

Papers:

H. Singh, M. Tayarani-Najaran, M. Yaqoob, "Exploring Computer Science Students’ Perception of ChatGPT in Higher Education: A Descriptive and Correlation Study", 2023, Education Sciences, 13, 924

Date: 2023/11/03
Time: 14:00
Location: online