Download as PDF, PPTX
Uploaded byGokul Alex
Decentralised AI through Distributed Ledger Technologies
The document discusses the evolution of artificial intelligence (AI) towards decentralization using distributed ledger technologies, highlighting challenges and potential solutions in computations, incentives, and data management. It covers various decentralized AI platforms and their applications across multiple sectors, such as finance and healthcare, while detailing the architecture and components necessary for implementing decentralized systems. Furthermore, it presents key concepts and innovations in blockchain technology that enhance collaboration, data sharing, and AI development.
More Related Content
Similar to Decentralised AI through Distributed Ledger Technologies
![Support, Monitoring, Continuous Improvement & Scaling Agentic Automation [3/3]](https://cdn.slidesharecdn.com/ss_thumbnails/agenticcommunityseries-day3-cfd-251120170304-ddef8112-thumbnail.jpg?width=640&height=640&fit=bounds)
Decentralised AI through Distributed Ledger Technologies
- 1. DECENTRALISED AI THROUGH DISTRIBUTEDLEDGERS National Seminar on Machine Intelligence 2020 - University of Kerala, Department of Computer Science,
- 2. The current contextof centralisation in AI • During the last few years, AI has evolved in centralised models across different aspects of the lifecycle of AI models. • Increasing gap between the AI innovations in large companies and small scale startups • Big companies control large datasets . Large datasets are used to create AI models. AI models generate high quality data
- 3. Computation Challenge forDecentralised AI • Decentralised distributed ledgers such as Blockchain protocols are still very limited in capability to execute computations such as the ones required by deep learning models. • Decentralised AI networks require off- chain computation models which requires scalable infrastructure.
- 4. Incentive Challenge forDecentralised AI • Decentralised AI models need to rely on incentive mechanisms to motivate different parties to participate in a network • Smart Contracts and Tokenisation implementations on Distributed Ledgers can help to realise incentives and rewards • However the incentive structures should be safeguarded against attack vectors from bad actors to manipulate the behaviour of a network
- 5. Decentralisation Roadmap • Decentralisationof Datasets • Decentralisation of Data Pipelines • Decentralisation of Training Algorithms • Decentralisation of Automation Algorithms • Decentralisation of Activation Functions • Decentralisation of Training Infrastructure • Decentralisation of Deployment Infrastructure
- 6. The challenges forcentralised AI
- 7. A contemporary dreamon a society of minds !
- 8. Key Components ofDecentralised AI
- 9. WHAT IS DISTRIBUTEDLEDGER TECHNOLOGY ?
- 10. A DIGITAL SYSTEMFOR RECORDING THE TRANSACTION OF ASSETS IN WHICH THE TRANSACTIONS ARE RECORDED IN MULTIPLE PLACES AT THE SAME TIME.
- 11. THERE IS NOCENTRAL ADMINISTRATOR OR CENTRALISED DATA STORAGE REQUIRED IN A DISTRIBUTED LEDGER.
- 12. A CONSENSUS OFREPLICATED, SHARED, AND SYNCHRONISED DIGITAL DATA GEOGRAPHICALLY SPREAD ACROSS MULTIPLE SITES, COUNTRIES, OR INSTITUTIONS.
- 13. A PEER TOPEER NETWORK AND A CONSENSUS ALGORITHM IS REQUIRED TO ENSURE REPLICATION ACROSS NODES.
- 14. It is importantto understand the difference between centralised and distributed ledger
- 15. There is adefinite differentiation between centralised, decentralised and distributed ledgers
- 16.
- 17. Recent Innovations inLedger Technology
- 18. Focused History ofDistributed Ledgers
- 19. An extended historyof distributed ledgers
- 22. Principles of DistributedLedgers
- 23. Blockchain Network andNode Anatomy
- 24. A simple viewon blockchain data structure
- 25. Hash Pointers ina Blockchain Data Structure
- 26. The end toend transaction in a Bitcoin Network
- 27. Connection of Blocksand Collection of Transactions
- 28. Structure of Blockchainin Ethereum Network
- 29. The Internals ofHashing Algorithms in Blockchain Data Structure
- 30. Development of WalletAddress through Cryptographic Algorithms
- 31. Key Management ina Bitcoin Blockchain
- 33. The message paddingin secure hashing algorithm
- 34. Essential Properties ofHash Functions
- 35. Message Blocks andBlock Ciphers in SHA - 256
- 36. Block Data Hashingin a Bitcoin Blockchain
- 37. Fundamentals of EllipticalCurve Cryptography based Digital Signature Algorithm
- 38. Depiction of EllipticalCurve Signing and Verification Actions
- 39. End to EndSteps in Digital Signature Verification
- 40. Transformation of Datain a Digital Signature
- 41. Blockchain Protocol Componentsand Categories
- 42. Architecture Layers -Important Views
- 43. Architecture Layers -Integration Models
- 44. Consensus Algorithms -Reference Architecture
- 45. Consensus Algorithms -Security Models
- 46. Consensus Algorithms -Important Milestones
- 47. Consensus Algorithms -Protocols and Platforms
- 48. Consensus Algorithms -Deep Dive
- 49. Consensus Algorithms -A Comparison
- 50. Nakamoto Consensus Algorithm- Components
- 51. Nakamoto Consensus Algorithm- A Deep Dive
- 52. PBFT Consensus -Hyperledger Implementation
- 53. Client Interaction andData Storage on a Blockchain Protocol Implementation
- 54. Additional Hybrid andPrivate Networks on Blockchain Protocol and Data Model
- 55. Enterprise Blockchain ReferenceArchitecture
- 56. A Cloud basedProduction Scale Blockchain Deployment Architecture
- 57. Healthcare Provider Patient DataManagement on Blockchain Protocol
- 58. A Wireless SensorNetwork on Blockchain Protocol
- 59. Big Data IntegrationArchitecture for Blockchain Driven Data Marketplaces
- 60. Blockchain for AIAlgorithms • SingularityNET - SingularityNET is a distributed AI platform on the Ethereum blockchain, with each blockchain node backing up an AI algorithm. • Intuition Fabric - The goal of Intuition Fabric is to democratise access to AI through a network of deep learning models that are stored on the interplanetary file system and accessed through the Ethereum blockchain.
- 61. Blockchain for AIAlgorithms • Cerebrum - Cerebrum is a decentralised platform for crowdsourced machine learning. Cerebrum enables anyone to encrypt their data and host machine learning competitions to utilise crowd- sourced machine learning models. • OpenMined - OpenMined is a community focused on building open- source technology for the decentralised ownership of data and intelligence. With OpenMined, AI can be trained on data that it never has access to.
- 62. Blockchain for AIAlgorithms • Effect.ai - Effect.ai is a blockchain- powered, decentralised platform for Artificial Intelligence development and AI related services. • Decentralised Machine Learning - DML aims to create a blockchain-based decentralised machine learning protocol and ecosystem. • Raven Protocol - Raven Protocol is a decentralised and distributed deep- learning training protocol.
- 63. Blockchain for AIAlgorithms • Thought Network - Thought's blockchain-enabled Fabric fundamentally changes applications by embedding artificial intelligence into every bit of data making it agile, actionable and inherently secure. • Matrix AI - Thought's blockchain- enabled Fabric fundamentally changes applications by embedding artificial intelligence into every bit of data making it agile, actionable and inherently secure.
- 64. Blockchain for AIAlgorithms • Pandora Boxchain - A marketplace for AI kernels, computations and big data powered by Proof of Cognitive Work (PoCW). • Fitchain - Fitchain is a blockchain-based machine learning factory that allows data scientists to train models, tracing them from inception and throughout their lifetime. • Cortex Labs - Cortex Labs is a decentralised AI platform with a virtual machine that allows you to execute AI programs on-chain.
- 65. Blockchain for Datasets •Ocean Protocol - Ocean Protocol is a decentralised data exchange protocol that lets people share and monetise data while guaranteeing control, auditability, transparency and compliance to all actors involved. Its network handles storing of the metadata (i.e. who owns what), links to the data itself, and more. • Neuromation - Distributed synthetic data platform for Deep Learning Applications.
- 66. Blockchain for Datasets •Synapse AI - A decentralised global data marketplace built on the Ethereum blockchain. • Bottos AI - A decentralised AI data sharing network that uses a consensus- based application platform to allow AI projects to gather training data. • Computable - A decentralised data marketplace for artificial intelligence applications.
- 67. Blockchain for Computation •Neureal - Open source, peer-to-peer AI supercomputing powered by blockchain. • TrueBit - TrueBit gives Ethereum smart contracts a computational boost. • DeepBrain Chain - A decentralised AI computing platform that supplies processing power to companies looking to develop A.I. technologies. • AI Crypto - AI Crypto is a blockchain-based ecosystem where AI resources such as GPU, models, and data are distributed in order to lower the cost of AI development.
- 68. Blockchain for AI inFinance • Numerai - Numerai is a hedge fund powered by a network of anonymous data scientists that build machine learning models to operate on encrypted data and stake cryptocurrency to express confidence in their models. • Erasure - Erasure is a decentralised protocol and data marketplace for financial predictions. • Cindicator - Cindicator is a crowd-sourced prediction engine for financial and crypto indicators.
- 69. Blockchain for AI inMedicine • Doc.ai - doc.ai aims to decentralise precision medicine on the blockchain by using AI. • BurstIQ - Healthcare data marketplace with granular ownership and granular consent of data. By using on-chain storage on a custom blockchain, BurstIQ can comply with HIPAA, GDPR, and other regulations. • Vitalyx - Vytalyx is a health technology company that plans to use AI and blockchain to store and analyse medical data.
- 70.
- 71. THE SINGULARITYNET PLATFORMWAS INITIALLY CONCEIVED BY BEN GOERTZEL, SIMONE’ GIACOMELLI AND DAVID HANSON IN A SERIES OF BRAINSTORMING SESSIONS AT HANSON ROBOTICS IN HONG KONG DURING THE FIRST HALF OF 2017.
- 72. Ben Goertzel on ArtificialGeneral Intelligence Intelligence is the ability to detect patterns in the world and in the agent itself, measurable in terms of emergent behaviour of "achieving complex goals in complex environments”. A "baby-like" artificial intelligence is initialised, then trained as an agent in a simulated or virtual world such as Second Life to produce a more powerful intelligence.
- 73. Ben Goertzel on ArtificialGeneral Intelligence Knowledge is represented in a network whose nodes and links carry probabilistic truth values as well as "attention values", with the attention values resembling the weights in a neural network. Several algorithms operate on this network, the central one being a combination of a probabilistic inference engine and a custom version of evolutionary programming.
- 74.
- 75.
- 76.
- 77.
- 78. SingularityNet - CollaborativeTraining Model
- 79.
- 80.
- 81. OPENCOG PRIME ISAN ARCHITECTURE FOR ROBOT AND VIRTUAL EMBODIED COGNITION THAT DEFINES A SET OF INTERACTING COMPONENTS DESIGNED TO GIVE RISE TO HUMAN-EQUIVALENT ARTIFICIAL GENERAL INTELLIGENCE (AGI) AS AN EMERGENT PHENOMENON OF THE WHOLE SYSTEM.
- 82. OpenCog Design Overview •A graph database that hold terms, formulaes, proofs, sentences, relationships together with interpretations • A collection of pre-defined knowledge representations • A collection of pre-defined type sub system • A collection of pre-defined imperative and functional programming compilers • A collection of pre-defined satisfiability modulo theory solvers • A generic rule engine including a forward chainer and backward chainer, that is able to chain rules together • Inference engines and reasoning systems, such as Bayesian Inference, Fuzzy logic, Constraint Solvers, Motion Planners • An attention allocation subsystem based on economic theory • A probabilistic reasoning engine based on probabilistic inference network • A probabilistic generic program evolver called MOSES • An NLP system consisting of Link Grammar, inspired by Meaning Text Theory and Word Grammar in relation to Dependency Grammar • OpenPsi - An implémentation of Psi- Theory for handling emotional states
- 83.
- 84. OpenCog Language andPerception Model
- 85.
- 86.
- 87. OpenCog Atomspace GraphDatabase
- 88.
- 89. OpenCog Cognitive StateGraphs
- 90. OpenCog Cognitive StateGraphs
- 94.
- 95. Ocean Protocol DataArchitecture