What problem does this paper attempt to address?

The problem that this paper attempts to solve is to understand the influence of synaptic plasticity on the neuron dynamics in neural networks and propose a mathematical framework to analyze the statistical properties of spike trains generated by synaptic plasticity. Specifically, the paper focuses on the following aspects: 1. **Mathematical Modeling of Synaptic Plasticity**: The paper introduces a mathematical framework for analyzing the statistical properties of spike trains generated by neural networks under the effect of synaptic plasticity. This includes defining and analyzing synaptic update rules or learning rules, which describe how synaptic weights are adjusted according to the activity patterns of neurons. 2. **Statistical Model of Spike Trains**: The paper explores the use of Gibbs distribution as an effective model for the statistical properties of spike trains. Gibbs distribution is widely used in statistical mechanics to describe the equilibrium state of a system, and the paper believes that this distribution can be naturally derived from the dynamics of neural networks. 3. **Variational Principle of Synaptic Plasticity**: The paper proposes that some synaptic plasticity rules can be related to a certain variational principle, that is, the change of synaptic weights is described by maximizing or minimizing a certain function (such as topological pressure). This variational principle can explain the optimization properties of synaptic plasticity, such as the optimization of information transmission. 4. **Dynamical Effects**: The paper studies the influence of synaptic plasticity on the dynamics of neural networks, especially how to change the periodic orbits in the network and the statistical properties of spike trains. Through specific examples of adaptation rules, the paper shows how the change of synaptic weights leads to gradual or abrupt changes (bifurcations) in network dynamics. 5. **Equilibrium Distribution**: The paper defines the "equilibrium point", that is, the state where synaptic weights no longer change, and derives that in this state, the distribution of spike trains can be described by Gibbs distribution. This provides a theoretical basis for understanding the long - term behavior of neural networks under the effect of synaptic plasticity. In summary, this paper aims to deeply understand how synaptic plasticity affects the dynamics and functions of neural networks through mathematical and statistical methods, thereby providing new perspectives and tools for related problems in neuroscience.

Statistics of spikes trains, synaptic plasticity and Gibbs distributions

A framework for studying synaptic plasticity with neural spike train data

Large Deviations Properties of Maximum Entropy Markov Chains from Spike Trains

Estimation of Spike Train Statistics in Spontaneously Active Biological Neural Networks

Entropy and information in neural spike trains: Progress on the sampling problem

Self-consistent determination of the spike-train power spectrum in a neural network with sparse connectivity

The Computational Structure of Spike Trains

Moment Neuronal Networks: Stochastic Computation in Neuronal Systems

Gibbs distribution analysis of temporal correlations structure in retina ganglion cells

A generative spike train model with time-structured higher order correlations

Neural spikes as rare events

Spike Pattern Structure Influences Efficacy Variability under STDP and Synaptic Homeostasis

Entropy of Neuronal Spike Patterns

Natural-gradient learning for spiking neurons

Applications of information geometry to spiking neural network activity

On stationary distributions of stochastic neural networks

Estimation of neuronal interaction graph from spike train data

A stochastic-field description of finite-size spiking neural networks

Effects of Firing Variability on Network Structures with Spike-Timing-Dependent Plasticity

Synaptic Modifications Driven by Spike-Timing-dependent Plasticity in Weakly Coupled Bursting Neurons

Exact computation of the Maximum Entropy Potential of spiking neural networks models