Welcome to the comprehensive, hands-on course for Mneme: a system for detecting field-like memory structures in biological systems. This program takes you from foundational theory to confident, practitioner-level use of Mneme’s CLI and Python APIs—with exercises, projects, and optional advanced modules.
- Audience: Scientists, ML/DS engineers, biophysicists, and curious generalists
- Prerequisites: Python fundamentals; basic linear algebra and probability; comfort with NumPy; curiosity about fields and topology
- Compute: CPU is sufficient for the MVP; GPU optional (PyTorch, heavy models)
- Duration: ~12–18 hours total (self-paced)
- Foundations (Modules 1–4): First principles, environment, CLI, pipeline anatomy
- Practitioner (Modules 5–9): Reconstruction, topology, attractors, visualization, experiments
- Advanced (Modules 10–11): Performance/monitoring; optional symbolic regression (PySR)
- Capstone: End-to-end experiment with reporting
- First Principles: Fields, Topology, Attractors
- Environment Setup and Sanity Checks
- CLI Quickstart: Generate → Analyze → Visualize
- Pipeline Anatomy and Configuration
- Field Reconstruction (IFT and GP)
- Topology (Cubical, Rips, Alpha) and Features
- Attractor Detection (Recurrence, Lyapunov, Clustering)
- Visualization and Reporting
- Designing Experiments and Reproducibility
- Performance and Monitoring (MVP Tools)
- Optional: Symbolic Regression with PySR
- Each module includes learning objectives, short readings, and exercises
- Exercises are designed to run in minutes on CPU
- Solutions are outlined inline after exercises (concealed by headings)
- Core project docs: Project Structure, Development Setup, API Design, Data Pipeline
- Run Logs: Each module will gain a short “Run log” section as we execute the exercises end-to-end, noting successes and any failures with fixes.
- Source:
src/mneme/(seeanalysis/,core/,data/,utils/)
Happy exploring!