{"id":31205322,"url":"https://github.com/aygp-dr/liquid-neural-networks","last_synced_at":"2025-09-20T17:58:28.600Z","repository":{"id":308598758,"uuid":"1021381963","full_name":"aygp-dr/liquid-neural-networks","owner":"aygp-dr","description":"Liquid Neural Networks (LNN) - Continuous-time neural dynamics inspired by C. elegans. Parameter-efficient AI with 19-302 neurons for complex tasks. Hybrid Clojure/Python implementation.","archived":false,"fork":false,"pushed_at":"2025-07-28T11:40:24.000Z","size":133,"stargazers_count":5,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-09-14T20:01:45.069Z","etag":null,"topics":["clojure","neural-networks","python","research"],"latest_commit_sha":null,"homepage":"https://github.com/jwalsh/liquid-neural-networks","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/aygp-dr.png","metadata":{"files":{"readme":"README.org","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2025-07-17T10:06:19.000Z","updated_at":"2025-09-11T15:26:13.000Z","dependencies_parsed_at":"2025-08-06T20:52:29.095Z","dependency_job_id":"4555df70-c240-429b-adaa-44135a2fed50","html_url":"https://github.com/aygp-dr/liquid-neural-networks","commit_stats":null,"previous_names":["aygp-dr/liquid-neural-networks"],"tags_count":2,"template":false,"template_full_name":null,"purl":"pkg:github/aygp-dr/liquid-neural-networks","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aygp-dr%2Fliquid-neural-networks","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aygp-dr%2Fliquid-neural-networks/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aygp-dr%2Fliquid-neural-networks/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aygp-dr%2Fliquid-neural-networks/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/aygp-dr","download_url":"https://codeload.github.com/aygp-dr/liquid-neural-networks/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aygp-dr%2Fliquid-neural-networks/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":276134627,"owners_count":25591182,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-09-20T02:00:10.207Z","response_time":63,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["clojure","neural-networks","python","research"],"created_at":"2025-09-20T17:58:26.566Z","updated_at":"2025-09-20T17:58:28.591Z","avatar_url":"https://github.com/aygp-dr.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"#+TITLE: Liquid Neural Networks\n#+AUTHOR: Aidan Pace\n#+DATE: 2025-01-17\n#+OPTIONS: toc:2 num:nil ^:nil\n\n[[https://github.com/aygp-dr/liquid-neural-networks/blob/main/LICENSE][https://img.shields.io/badge/license-MIT-blue.svg]]\n[[https://github.com/aygp-dr/liquid-neural-networks][https://img.shields.io/badge/python-3.9+-blue.svg]]\n[[https://github.com/aygp-dr/liquid-neural-networks][https://img.shields.io/badge/clojure-1.11+-blue.svg]]\n[[https://github.com/aygp-dr/liquid-neural-networks][https://img.shields.io/badge/status-draft-orange.svg]]\n[[https://github.com/aygp-dr/liquid-neural-networks/releases][https://img.shields.io/badge/release-v0.1.0-blue.svg]]\n\n* Overview\n\n*🚧 Project Status: Draft/In Progress*\n\nThis project is in early development. Core algorithms and implementations are being actively developed.\n\nLiquid Neural Networks (LNNs) are a revolutionary approach to artificial intelligence that draws inspiration from biological neural systems, particularly the C. elegans nervous system. Unlike traditional neural networks, LNNs use continuous-time dynamics and can adapt in real-time to changing inputs.\n\n** Key Features\n\n- *Parameter Efficiency*: Solve complex tasks with as few as 19-302 neurons\n- *Continuous-Time Dynamics*: Based on ordinary differential equations (ODEs)\n- *Real-Time Adaptation*: Networks that evolve and adapt during inference\n- *Superior Interpretability*: Understand exactly how decisions are made\n- *Edge AI Ready*: Efficient enough for deployment on resource-constrained devices\n\n** Why Liquid Neural Networks?\n\nTraditional neural networks require millions of parameters and struggle with:\n- Adapting to new situations without retraining\n- Explaining their decision-making process\n- Running efficiently on edge devices\n- Handling time-series data naturally\n\nLNNs address these limitations by mimicking biological neurons more closely, using differential equations to model continuous-time dynamics.\n\n* Quick Start\n\n** Installation\n\n*** Python\n#+begin_src bash\n# Using pip\npip install liquid-neural-networks\n\n# Using uv (recommended)\nuv pip install liquid-neural-networks\n\n# Development installation\ngit clone https://github.com/aygp-dr/liquid-neural-networks\ncd liquid-neural-networks\nuv pip install -e \".[dev]\"\n#+end_src\n\n*** Clojure\n#+begin_src bash\n# Add to your deps.edn\n{:deps {aygp-dr/liquid-neural-networks {:git/url \"https://github.com/aygp-dr/liquid-neural-networks\"\n                                        :git/sha \"LATEST_SHA\"}}}\n\n# Or use from source\ngit clone https://github.com/aygp-dr/liquid-neural-networks\ncd liquid-neural-networks\nclojure -M:dev\n#+end_src\n\n** Basic Usage\n\n*** Python Example\n#+begin_src python\nfrom liquid_neural_networks import LiquidNeuron, LiquidNetwork\n\n# Create a simple liquid neural network\nnetwork = LiquidNetwork(\n    input_size=10,\n    hidden_size=32,  # Just 32 neurons!\n    output_size=2\n)\n\n# Train on time-series data\nfor epoch in range(100):\n    outputs = network(inputs, time_constants)\n    loss = criterion(outputs, targets)\n    loss.backward()\n#+end_src\n\n*** Clojure Example\n#+begin_src clojure\n(require '[liquid-neural-networks.core :as lnn])\n\n;; Create a liquid network\n(def network (lnn/create-network {:input-size 10\n                                  :hidden-size 32\n                                  :output-size 2}))\n\n;; Process time-series data\n(def result (lnn/forward network input-data time-constants))\n#+end_src\n\n* Applications\n\n** Autonomous Systems\n- Drone navigation with 19 neurons\n- Self-driving car control\n- Robotic arm manipulation\n\n** Time-Series Analysis\n- Financial market prediction\n- Weather forecasting\n- Sensor data processing\n\n** Medical Diagnostics\n- ECG analysis\n- Brain signal interpretation\n- Disease progression modeling\n\n** Edge AI\n- IoT device intelligence\n- Embedded system control\n- Real-time anomaly detection\n\n* Architecture\n\nLNNs consist of three main components:\n\n1. *Liquid Time-Constant (LTC) Neurons*: Neurons with adaptive time constants that change based on input\n2. *Continuous-Time Dynamics*: ODEs that govern neuron behavior\n3. *Sparse Connectivity*: Efficient wiring patterns inspired by biological systems\n\nThe mathematical foundation:\n#+begin_example\ndx/dt = -x/τ(t) + f(Wx + b)\nwhere τ(t) is the adaptive time constant\n#+end_example\n\n* Performance\n\nBenchmark results comparing LNNs to traditional architectures:\n\n| Task | Traditional NN | LNN | Parameters Reduction |\n|------+---------------+-----+---------------------|\n| Drone Control | 100K params | 19 neurons | 99.98% |\n| Time-Series | 1M params | 302 neurons | 99.97% |\n| Image Classification | 25M params | 1K neurons | 99.99% |\n\n* Contributing\n\nWe welcome contributions! See our [[file:CONTRIBUTING.org][Contributing Guide]] for:\n- Code style guidelines\n- Testing requirements\n- Pull request process\n- Development setup\n\n* Documentation\n\n- [[file:docs/tutorials/][Tutorials]]: Step-by-step guides\n- [[file:docs/api/][API Reference]]: Detailed documentation\n- [[file:examples/][Examples]]: Working code samples\n- [[file:SETUP.org][Development Setup]]: For contributors\n\n* Research\n\nThis implementation is based on:\n- Hasani et al. \"Liquid Time-constant Networks\" (2021)\n- Lechner et al. \"Neural Circuit Policies\" (2020)\n- MIT CSAIL research on continuous-time neural models\n\n* License\n\nMIT License - see [[file:LICENSE][LICENSE]] for details.\n\n* Community\n\n- [[https://github.com/aygp-dr/liquid-neural-networks/discussions][GitHub Discussions]]\n- [[https://github.com/aygp-dr/liquid-neural-networks/issues][Issue Tracker]]\n- Research papers and citations in [[file:docs/papers/][docs/papers/]]","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faygp-dr%2Fliquid-neural-networks","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Faygp-dr%2Fliquid-neural-networks","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faygp-dr%2Fliquid-neural-networks/lists"}