AI Article Synopsis
- The paper assesses the sensitivity of spiking neural networks (SNNs) to input perturbations, specifically focusing on leaky integrate-and-fire (LIF) neurons, which display complex neuronal dynamics.
- It proposes a new sensitivity definition based on the differences between perturbed and unperturbed outputs and develops an algorithm to calculate the overall sensitivity of the SNN from individual neuron sensitivities, using desired firing times for improved accuracy.
- The findings indicate that while sensitivity correlates with various parameter changes in the network, it shows a piecewise decreasing trend concerning time steps, revealing nuances in how different neuron parameters impact network behavior.
Article Abstract
To investigate the behavior of spiking neural networks (SNNs), the sensitivity of input perturbation serves as an effective metric for assessing the influence on the network output. However, existing methods fall short in evaluating the sensitivity of SNNs featuring biologically plausible leaky integrate-and-fire (LIF) neurons due to the intricate neuronal dynamics during the feedforward process. This paper first defines the sensitivity of a temporal-coded spiking neuron (SN) as the deviation between the perturbed and unperturbed output under a given input perturbation with respect to overall inputs. Then, the sensitivity algorithm of an entire SNN is derived iteratively from the sensitivity of each individual neuron. Instead of using the actual firing time, the desired firing time is employed to derive a more precise analytical expression of the sensitivity. Moreover, the expectation of the membrane potential difference is utilized to quantify the magnitude of the input deviation. The theoretical results achieved with the proposed algorithm are in reasonable agreement with the simulation results obtained with extensive input data. The sensitivity also varies monotonically with changes in other parameters, except for the number of time steps, providing valuable insights for choosing appropriate values to construct the network. Nevertheless, the sensitivity exhibits a piecewise decreasing tendency with respect to the number of time steps, with the length and starting point of each piece contingent upon the specific parameter values of the neuron.
Download full-text PDF |
Source |
|---|---|
| http://www.ncbi.nlm.nih.gov/pmc/articles/PMC11592311 | PMC |
| http://dx.doi.org/10.3390/brainsci14111149 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Peripheral contributions to resting state brain dynamics.
Nat Commun
December 2024
Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, US.
The correlational structure of brain activity dynamics in the absence of stimuli or behavior is often taken to reveal intrinsic properties of neural function. To test the limits of this assumption, we analyzed peripheral contributions to resting state activity measured by fMRI in unanesthetized, chemically immobilized male rats that emulate human neuroimaging conditions. We find that perturbation of somatosensory input channels modifies correlation strengths that relate somatosensory areas both to one another and to higher-order brain regions, despite the absence of ostensible stimuli or movements.
View Article and Find Full Text PDFOn machine learnability of local contributions to interatomic potentials from density functional theory calculations.
Sci Rep
December 2024
Department of Physics, Shahid Beheshti University, Tehran, 1983969411, Iran.
Machine learning interatomic potentials, as a modern generation of classical force fields, take atomic environments as input and predict the corresponding atomic energies and forces. We challenge the commonly accepted assumption that the contribution of an atom can be learned from the short-range local environment of that atom. We employ density functional theory calculations to quantify the decay of the induced electron density and electrostatic potential in response to local perturbations throughout insulating, semiconducting and metallic samples of different dimensionalities.
View Article and Find Full Text PDFTopology synthesis of coupled inductor based four port DC-DC converter for multi-input PV-battery application with autonomous mode selection.
Sci Rep
December 2024
Department of Electrical and Electronics Engineering, Dr.N.G.P. Institute of Technology, Coimbatore, India.
Multiport converters are the most reliable and integral component for latest renewable source integration with multiple inputs. This article is one among the kind, which proposes a novel Coupled Inductor based Four Port topology Multiport Converter (CI-FP-MPC) for integrating multiple PV sources with different voltages. The adoption of coupled inductor contributes an increased voltage gain with reduced stress on the switches and diodes.
View Article and Find Full Text PDFSetpoint weighted PI-FOPD cascade controllers synthesis for unstable time-delayed processes satisfying prespecified safety margins.
ISA Trans
December 2024
Department of Electrical Engineering, Tungnan University, No. 152, Section 3, Peishen Rd., Shenkeng Dist., New Taipei 222, Taiwan. Electronic address:
This article introduces a novel setpoint weighted PI-FOPD (SWPI-FOPD) cascade controller with a prefilter. It further describes a four-stage design strategy that sequentially enhances tracking responses, reduces overshoot, and ensures robustness for unstable time-delayed (UTD) processes. The controller applies to integrating and non-integrating UTD processes of any order and does not necessitate model order reduction or delay approximation.
View Article and Find Full Text PDFEvaluations of the Perturbation Resistance of the Deep-Learning-Based Ligand Conformation Optimization Algorithm.
J Chem Inf Model
December 2024
School of Physics, Shandong University, Jinan 250100, China.
In recent years, the deep learning (DL) technique has rapidly developed and shown great success in scoring the protein-ligand binding affinities. The protein-ligand conformation optimization based on DL-derived scoring functions holds broad application prospects, for instance, drug design and enzyme engineering. In this study, we evaluated the robustness of a DL-based ligand conformation optimization protocol (DeepRMSD+Vina) for optimizing structures with input perturbations by examining the predicted ligand binding poses and scoring.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!