📘 Modeling Memristors with Machine Learning

🌟 Overview

This project focuses on modeling the behavior of memristors using simulated datasets and applying machine learning (ML) to:

• 🔌 Predict current (I) from voltage (V) and device parameters
• 🟩 Classify High / Low resistance states
• ⏳ Analyze time-dependent behavior with LSTM networks
• 📊 Visualize I-V characteristics, resistance hysteresis, and parameter sensitivity

Memristors are non-linear resistive devices with memory, crucial for neuromorphic computing and next-generation memory systems.
Accurate modeling supports the design of robust circuits and intelligent systems.

---

🎯 Objectives

• 🔎 Compare memristor models: Yakopcic, MMS, stat, VTEAM
• 📈 Perform regression to predict current (I) from voltage (V)
• 🟩 Perform classification of resistance states (High / Low)
• ⏳ Model time-dependent dynamics with LSTM
• 📊 Visualize I-V curves, R-V hysteresis, and parameter sensitivity

---

📂 Dataset

Dataset available on Kaggle in .mat format.

Included Models

• Yakopcic
• MMS
• stat
• VTEAM

Fields per Model

Model	Main Fields
Yakopcic, MMS, VTEAM	Amp, Freq, Dop, Rs, U_m, I_m, t, min_r, param, X, G, V, I, cost_function2compare, U_sin
stat	Amp, Freq, Dop, Rs, U_m, I_m, t, std, wsk, eps, I_m_b_interp, U_m_b_interp, U_sin

---

🔬 Methodology

1️⃣ Regression — Current Prediction

• Predict I from V and device parameters
• Models: Linear Regression & Neural Networks
• Evaluation: Predicted vs. Actual scatter plots

2️⃣ Classification — Resistance States

• Compute resistance: R = V / I
• Define High / Low resistance (median or percentile thresholds)
• Model: Random Forest Classifier
• Metrics: Accuracy, Confusion Matrix, Visualizations

3️⃣ Sequential Modeling — LSTM

• Capture temporal dynamics of current over time t
• Create sliding-window sequences of V, I, and parameters
• Train LSTM to predict next-step current
• Visualize predicted vs. actual sequences

4️⃣ Advanced Analysis

• 🔄 Hysteresis Loops: R-V curves for each model
• ⚙️ Parameter Sensitivity: Effects of amplitude, frequency, doping
• 📐 Feature Engineering: Derived features such as dV/dt and dI/dt to improve ML models

---

🚀 How to Run

Follow these steps to set up and run the project.

1. Clone the Repository

git clone https://github.com/Lily-Evan/Modeling-Memristors-with-Machine-Learning.git
cd Modeling-Memristors-with-Machine-Learning

Author: Lily-Evan (original repo) / Adapted by Panagiota G for improved README

Kaggle Notebook: Modeling Memristors with Machine Learning