arXiv subjects

Agnieszka Pregowska

Publications and source records attributed to Agnieszka Pregowska.

16 recordsLinked to original sources

arXiv2026

Implicit neural representations as a coordinate-based framework for continuous environmental field reconstruction from sparse ecological observations

Reconstructing continuous environmental fields from sparse and irregular observations remains a central challenge in environmental modelling and biodiversity informatics. Many ecological datasets are heterogeneous in space and time, making grid-based approaches difficult to scale or generalise across domains. Here, we evaluate implicit neural representations (INRs) as a coordinate-based modelling framework for learning continuous spatial and spatio-temporal fields directly from coordinate inputs. We analyse their behaviour across three representative modelling scenarios: species distribution reconstruction, phenological dynamics, and morphological segmentation derived from open biodiversity data. Beyond predictive performance, we examine interpolation behaviour, spatial coherence, and computational characteristics relevant for environmental modelling workflows, including scalability, resolution-independent querying, and architectural inductive bias. Results show that neural fields provide stable continuous representations with predictable computational cost, complementing classical smoothers and tree-based approaches. These findings position coordinate-based neural fields as a flexible representation layer that can be integrated into environmental modelling pipelines and exploratory analysis frameworks for large, irregularly sampled datasets.

cs.LG↗

arXiv2026

Continuous ageing trajectory representations for knee-aware lifetime prediction of lithium-ion batteries across heterogeneous dataset

Accurate assessment of lithium-ion battery ageing is challenged by cell-to-cell variability, heterogeneous cycling protocols, and limited transferability of data-driven models across datasets. In particular, robust identification of degradation transitions, such as the knee point, and reliable early-life prediction of remaining useful life (RUL) remain open problems. This study proposes a unified framework for battery ageing analysis based on continuous representations of voltage-capacity and capacity-cycle trajectories learned from heterogeneous public datasets (NASA, CALCE, ISU-ILCC). The continuous formulation enables consistent extraction of degradation descriptors, including curvature, plateau length and knee-related metrics, while reducing sensitivity to dataset-specific discretisation. Across more than 250 cells, statistically significant correlations between knee onset and end-of-life (Pearson 0.75-0.84) are observed. Additional early-life analysis confirms that knee-related features retain predictive value when estimated from partial trajectories. Early-life models provide increasingly stable RUL predictions as the number of observed cycles increases, with meaningful predictive performance emerging within the first 5-20 cycles and remain robust under cross-dataset domain shift. The framework integrates continuous modelling, feature extraction and uncertainty-aware prediction, providing an interpretable and dataset-consistent approach demonstrating robustness across heterogeneous dataset types. Compared with conventional discrete or feature-based methods, the proposed representation reduces sensitivity to sampling resolution and improves cross-dataset consistency. The study is limited to laboratory-scale datasets and capacity-based end-of-life definitions.

cs.LG↗

arXiv2026

Improving Liver Disease Diagnosis with SNNDeep: A Custom Spiking Neural Network Using Diverse Learning Algorithms

Purpose: Spiking neural networks (SNNs) have recently gained attention as energy-efficient, biologically plausible alternatives to conventional deep learning models. Their application in high-stakes biomedical imaging remains almost entirely unexplored. Methods: This study introduces SNNDeep, the first tailored SNN specifically optimized for binary classification of liver health status from computed tomography (CT) features. To ensure clinical relevance and broad generalizability, the model was developed and evaluated using the Task03\Liver dataset from the Medical Segmentation Decathlon (MSD), a standardized benchmark widely used for assessing performance across diverse medical imaging tasks. We benchmark three fundamentally different learning algorithms, namely Surrogate Gradient Learning, the Tempotron rule, and Bio-Inspired Active Learning across three architectural variants: a fully customized low-level model built from scratch, and two implementations using leading SNN frameworks, i.e., snnTorch and SpikingJelly. Hyperparameter optimization was performed using Optuna. Results: Our results demonstrate that the custom-built SNNDeep consistently outperforms framework-based implementations, achieving a maximum validation accuracy of 98.35%, superior adaptability across learning rules, and significantly reduced training overhead. Conclusion:This study provides the first empirical evidence that low-level, highly tunable SNNs can surpass standard frameworks in medical imaging, especially in data-limited, temporally constrained diagnostic settings, thereby opening a new pathway for neuro-inspired AI in precision medicine.

cs.NE↗

arXiv2026

Implicit neural representations for larval zebrafish brain microscopy: a reproducible benchmark on the MapZebrain atlas

Implicit neural representations (INRs) offer continuous coordinate-based encodings for atlas registration, cross-modality resampling, sparse-view completion, and compact sharing of neuroanatomical data. Yet reproducible evaluation is lacking for high-resolution larval zebrafish microscopy, where preserving neuropil boundaries and fine neuronal processes is critical. We present a reproducible INR benchmark for the MapZebrain larval zebrafish brain atlas. Using a unified, seed-controlled protocol, we compare SIREN, Fourier features, Haar positional encoding, and a multi-resolution grid on 950 grayscale microscopy images, including atlas slices and single-neuron projections. Images are normalized with per-image (1,99) percentiles estimated from 10% of pixels in non-held-out columns, and spatial generalization is tested with a deterministic 40% column-wise hold-out along the X-axis. Haar and Fourier achieve the strongest macro-averaged reconstruction fidelity on held-out columns (about 26 dB), while the grid is moderately behind. SIREN performs worse in macro averages but remains competitive on area-weighted micro averages in the all-in-one regime. SSIM and edge-focused error further show that Haar and Fourier preserve boundaries more accurately. These results indicate that explicit spectral and multiscale encodings better capture high-frequency neuroanatomical detail than smoother-bias alternatives. For MapZebrain workflows, Haar and Fourier are best suited to boundary-sensitive tasks such as atlas registration, label transfer, and morphology-preserving sharing, while SIREN remains a lightweight baseline for background modelling or denoising.

cs.CV↗

arXiv2026

Learning Internal Biological Neuron Parameters and Complexity-Based Encoding for Improved Spiking Neural Networks Performance

This study proposes a novel learning paradigm for spiking neural networks (SNNs) that replaces the perceptron-inspired abstraction with biologically grounded neuron models, jointly optimizing synaptic weights and intrinsic neuronal parameters. We evaluate two architectures, leaky integrate-and-fire (LIF) and a meta-neuron model, under fixed and learnable intrinsic dynamics. Additionally, we introduce a biologically inspired classification framework that combines SNN dynamics with Lempel-Ziv complexity (LZC), enabling efficient and interpretable classification of spatiotemporal spike data. Training is conducted using surrogate-gradient backpropagation, spike-timing-dependent plasticity (STDP), and the Tempotron rule on spike trains generated from Poisson processes, widely adopted in computational neuroscience as a standard stochastic model of neuronal spike generation due to their analytical tractability and empirical relevance. Learning intrinsic parameters improves classification accuracy by up to 13.50 percentage points for LIF networks and 8.50 for meta-neuron models compared to baselines tuning only network size and learning rate. The proposed SNN-LZC classifier achieves up to 99.50% accuracy with sub-millisecond inference latency and competitive energy consumption. We further provide theoretical justification by formalizing how optimizing intrinsic dynamics enlarges the hypothesis class and proving descent guarantees for intrinsic-parameter updates under standard smoothness assumptions, linking intrinsic optimization to provable improvements in the surrogate objective.

cs.NE↗

arXiv2026

Accuracy-Efficiency Trade-Offs in Spiking Neural Networks: A Lempel-Ziv Complexity Perspective on Learning Rules

Training spiking neural networks (SNNs) remains challenging due to temporal dynamics, non-differentiability of spike events, and sparse event-driven activations. This paper studies how the choice of learning paradigm (unsupervised, supervised, and hybrid) affects classification performance and computational cost in temporal pattern recognition. Building on our earlier study [Rudnicka et al., 2026], we use Lempel-Ziv complexity (LZC) as a compact, decision-relevant descriptor of spike-train temporal organization to quantify how different learning rules reshape class-conditional temporal structure. The pipeline combines a leaky integrate-and-fire (LIF) SNN with an LZC-based decision rule. We evaluate learning rules on synthetic sources with controlled temporal statistics (Bernoulli, two-state Markov, and Poisson spike processes) and on two-class subsets of MNIST and N-MNIST. Across datasets, gradient-based learning achieves the highest accuracy but at high computational cost, whereas bio-inspired rules (e.g., Tempotron and SpikeProp) offer favorable accuracy--efficiency trade-offs. These results highlight that selecting a learning rule should be guided by application constraints and the desired balance between separability and computational overhead.

cs.AI↗

arXiv2025

Impact of Neuron Models on Spiking Neural Networks performance. A Complexity Based Classification Approach

This study explores how the selection of neuron models and learning rules impacts the classification performance of Spiking Neural Networks (SNNs), with a focus on applications in bio-signal processing. We compare biologically inspired neuron models, including Leaky Integrate-and-Fire (LIF), metaneurons, and probabilistic Levy-Baxter (LB) neurons, across multiple learning rules, including spike-timing-dependent plasticity (STDP), tempotron, and reward-modulated updates. A novel element of this work is the integration of a complexity-based decision mechanism into the evaluation pipeline. Using Lempel-Ziv Complexity (LZC), a measure related to entropy rate, we quantify the structural regularity of spike trains and assess classification outcomes in a consistent and interpretable manner across different SNN configurations. To investigate neural dynamics and assess algorithm performance, we employed synthetic datasets with varying temporal dependencies and stochasticity levels. These included Markov and Poisson processes, well-established models to simulate neuronal spike trains and capture the stochastic firing behavior of biological neurons.Validation of synthetic Poisson and Markov-modeled data reveals clear performance trends: classification accuracy depends on the interaction between neuron model, network size, and learning rule, with the LZC-based evaluation highlighting configurations that remain robust to weak or noisy signals. This work delivers a systematic analysis of how neuron model selection interacts with network parameters and learning strategies, supported by a novel complexity-based evaluation approach that offers a consistent benchmark for SNN performance.

q-bio.NC↗

arXiv2025

Integrating Complexity and Biological Realism: High-Performance Spiking Neural Networks for Breast Cancer Detection

Spiking Neural Networks (SNNs) event-driven nature enables efficient encoding of spatial and temporal features, making them suitable for dynamic time-dependent data processing. Despite their biological relevance, SNNs have seen limited application in medical image recognition due to difficulties in matching the performance of conventional deep learning models. To address this, we propose a novel breast cancer classification approach that combines SNNs with Lempel-Ziv Complexity (LZC) a computationally efficient measure of sequence complexity. LZC enhances the interpretability and accuracy of spike-based models by capturing structural patterns in neural activity. Our study explores both biophysical Leaky Integrate-and-Fire (LIF) and probabilistic Levy-Baxter (LB) neuron models under supervised, unsupervised, and hybrid learning regimes. Experiments were conducted on the Breast Cancer Wisconsin dataset using numerical features derived from medical imaging. LB-based models consistently exceeded 90.00% accuracy, while LIF-based models reached over 85.00%. The highest accuracy of 98.25% was achieved using an ANN-to-SNN conversion method applied to both neuron models comparable to traditional deep learning with back-propagation, but at up to 100 times lower computational cost. This hybrid approach merges deep learning performance with the efficiency and plausibility of SNNs, yielding top results at lower computational cost. We hypothesize that the synergy between temporal-coding, spike-sparsity, and LZC-driven complexity analysis enables more-efficient feature extraction. Our findings demonstrate that SNNs combined with LZC offer promising, biologically plausible alternative to conventional neural networks in medical diagnostics, particularly for resource-constrained or real-time systems.

cs.NE↗

arXiv2024

Efficient Model-Stealing Attacks Against Inductive Graph Neural Networks

Graph Neural Networks (GNNs) are recognized as potent tools for processing real-world data organized in graph structures. Especially inductive GNNs, which allow for the processing of graph-structured data without relying on predefined graph structures, are becoming increasingly important in a wide range of applications. As such these networks become attractive targets for model-stealing attacks where an adversary seeks to replicate the functionality of the targeted network. Significant efforts have been devoted to developing model-stealing attacks that extract models trained on images and texts. However, little attention has been given to stealing GNNs trained on graph data. This paper identifies a new method of performing unsupervised model-stealing attacks against inductive GNNs, utilizing graph contrastive learning and spectral graph augmentations to efficiently extract information from the targeted model. The new type of attack is thoroughly evaluated on six datasets and the results show that our approach outperforms the current state-of-the-art by Shen et al. (2021). In particular, our attack surpasses the baseline across all benchmarks, attaining superior fidelity and downstream accuracy of the stolen model while necessitating fewer queries directed toward the target model.

cs.LG↗

arXiv2024

How scanning probe microscopy can be supported by Artificial Intelligence and quantum computing

We focus on the potential possibilities for supporting Scanning Probe Microscopy measurements, emphasizing the application of Artificial Intelligence, especially Machine Learning as well as quantum computing. It turned out that Artificial Intelligence can be helpful in the experimental processes automation in routine operations, the algorithmic search for good sample regions, and shed light on the structure property relationships. Thus, it contributes to increasing the efficiency and accuracy of optical nanoscopy scanning probes. Moreover, the combination of Artificial Intelligence based algorithms and quantum computing may have a huge potential to increase the practical application of Scanning Probe Microscopy. The limitations were also discussed. Finally, we outline a research path for the improvement of the proposed approach.

q-bio.NC↗

arXiv2024

Health Digital Twins Supported by Artificial Intelligence-based Algorithms and Extended Reality in Cardiology

Recently, significant efforts have been made to create Health Digital Twins (HDTs), digital twins for clinical applications. Heart modeling is one of the fastest-growing fields, which favors the effective application of HDTs. The clinical application of HDTs will be increasingly widespread in the future of healthcare services and has a huge potential to form part of the mainstream in medicine. However, it requires the development of both models and algorithms for the analysis of medical data, and advances in Artificial Intelligence (AI) based algorithms have already revolutionized image segmentation processes. Precise segmentation of lesions may contribute to an efficient diagnostics process and a more effective selection of targeted therapy. In this paper, a brief overview of recent achievements in HDT technologies in the field of cardiology, including interventional cardiology was conducted. HDTs were studied taking into account the application of Extended Reality (XR) and AI, as well as data security, technical risks, and ethics-related issues. Special emphasis was put on automatic segmentation issues. It appears that improvements in data processing will focus on automatic segmentation of medical imaging in addition to three-dimensional (3D) pictures to reconstruct the anatomy of the heart and torso that can be displayed in XR-based devices. This will contribute to the development of effective heart diagnostics. The combination of AI, XR, and an HDT-based solution will help to avoid technical errors and serve as a universal methodology in the development of personalized cardiology. Additionally, we describe potential applications, limitations, and further research directions.

q-bio.NC↗

arXiv2024

Artificial Intelligence-based algorithms in medical image scan seg-mentation and intelligent visual-content generation -- a concise overview

Recently, Artificial Intelligence (AI)-based algorithms have revolutionized the medical image segmentation processes. Thus, the precise segmentation of organs and their lesions may contribute to an efficient diagnostics process and a more effective selection of targeted therapies as well as increasing the effectiveness of the training process. In this context, AI may contribute to the automatization of the image scan segmentation process and increase the quality of the resulting 3D objects, which may lead to the generation of more realistic virtual objects. In this paper, we focus on the AI-based solutions applied in the medical image scan segmentation, and intelligent visual-content generation, i.e. computer-generated three-dimensional (3D) images in the context of Extended Reality (XR). We consider different types of neural networks used with a special emphasis on the learning rules applied, taking into account algorithm accuracy and performance, as well as open data availability. This paper attempts to summarize the current development of AI-based segmentation methods in medical imaging and intelligent visual content generation that are applied in XR. It concludes also with possible developments and open challenges in AI application in Extended Reality-based solutions. Finally, the future lines of research and development directions of Artificial Intelligence applications both in medical image segmentation and Extended Reality-based medical solutions are discussed

q-bio.NC↗

arXiv2023

Does a larger neural network mean greater information transmission efficiency?

Realistic modeling of brain involves large number of neurons. The important question is how this size affects transmission efficiency? Here, these issue is studied in terms of Shannon's Theory. Mutual Information between input and output signals for simple class of networks with an increasing number of neurons is analyzed theoretically and numerically. Levy-Baxter neural model is applied. It turned out, that for these networks the Mutual Information converges, with increasing size, asymptotically very slowly to saturation level. This suggests that from certain level, the increase of neurons number does not imply significant increase in transmission efficiency, contributes rather to reliability.

q-bio.NC↗

arXiv2020

Signal Fluctuations and the Information Transmission Rates in Binary Communication Channels

In nervous system information is conveyed by sequence of action potentials (spikes-trains). As MacKay and McCulloch proposed, spike-trains can be represented as bits sequences coming from Information Sources. Previously, we studied relations between Information Transmission Rates (ITR) carried out by the spikes, their correlations, and frequencies. Here, we concentrate on the problem of how spikes fluctuations affect ITR. The Information Theory Method developed by Shannon is applied. Information Sources are modeled as stationary stochastic processes. We assume such sources as two states Markov processes. As a spike-trains' fluctuation measure, we consider the Standard Deviation SD, which, in fact, measures average fluctuation of spikes around the average spike frequency. We found that character of ITR and signal fluctuations relation strongly depends on parameter s which is a sum of transitions probabilities from no spike state to spike state and vice versa. It turned out that for smaller s (s<1) the quotient ITR/SD has a maximum and can tend to zero depending on transition probabilities. While for s large enough 1<s the ITR/SD is separated from 0 for each s. Similar behavior was observed when we replaced Shannon entropy terms in Markov entropy formula by their approximation with polynomials. We also show that the ITR quotient by Variance behaves in a completely different way. We show that for large transition parameter s the Information Transmission Rate by SD will never decrease to 0. Specifically, for 1<s<1.7 the ITR will be always, independently on transition probabilities which form this s, above the level of fluctuations, i.e. we have SD<ITR. We conclude that in a more noisy environment, to get appropriate reliability and efficiency of transmission, Information Sources with higher tendency of transition from the state no spike to spike state and vice versa should be applied.

q-bio.QM↗

arXiv2018

How far can neural correlations reduce uncertainty? Comparison of Information Transmission Rates for Markov and Bernoulli processes

The nature of neural codes is central to neuroscience. Do neurons encode information through relatively slow changes in the emission rates of individual spikes (rate code), or by the precise timing of every spike (temporal codes)? Here we compare the loss of information due to correlations for these two possible neural codes. The essence of Shannon's definition of information is to combine information with uncertainty: the higher the uncertainty of a given event, the more information is conveyed by that event. Correlations can reduce uncertainty or the amount of information, but by how much? In this paper we address this question by a direct comparison of the information per symbol conveyed by the words coming from a binary Markov source (temporal codes) with the information per symbol coming from the corresponding Bernoulli source (uncorrelated, rate code source). In a previous paper we found that a crucial role in the relation between Information Transmission Rates (ITR) and Firing Rates is played by a parameter s, which is the sum of transitions probabilities from the no-spike-state to the spike-state and vice versa. It turned out that also in this case a crucial role is played by the same parameter s. We found bounds of the quotient of ITRs for these sources, i.e. this quotient's minimal and maximal values. Next, making use of the entropy grouping axiom, we determined the loss of information in a Markov source in relation to its corresponding Bernoulli source for a given length of word. Our results show that in practical situations in the case of correlated signals the loss of information is relatively small, thus temporal codes, which are more energetically efficient, can replace the rate code effectively. These phenomena were confirmed by experiments.

q-bio.QM↗

arXiv2016

Temporal code versus rate code for binary Information Sources

Neuroscientists formulate very different hypotheses about the nature of neural code. At one extreme, it has been argued that neurons encode information in relatively slow changes of individual spikes arriving "rates codes" and the irregularity in the spike trains reflects noise in the system, while in the other extreme this irregularity is the temporal codes thus the precise timing of every spike carries additional information about the input. It is known that in the estimation of Shannon information the patterns and temporal structures are taken into account, while the rate code is determined by firing rate. We compare these types of codes for binary Information Sources which model encoded spike-trains. Assuming that the information transmitted by a neuron is governed by uncorrelated stochastic process or by process with a memory we compare the information transmission rates carried by such spike-trains with their firing rates. We showed that the crucial role in the relation between information and firing rates is played by a quantity which we call "jumping" parameter. It corresponds to the probabilities of transitions from no-spike-state to the spike-state and vice versa. For low values of jumping parameter the quotient of information and firing rates is monotonically decreasing function of firing rate, thus there is straightforward, one-to-one, relation between temporal and rate codes. On the contrary, it turns out that for large enough jumping parameter this quotient is non-monotonic function of firing rate and it exhibits a global maximum, in this case optimal firing rate exists. Moreover, there is no one-to-one relation between information and firing rates, so the temporal and rate codes differ qualitatively. This leads to the observation that the behavior of the quotient of information and firing rates for large jumping parameter is important in the context of bursting phenomena.

q-bio.NC↗