Overview
terms used
Cognitive Blocks
The Cognitive Blocks are structural units generated as the output result of the Ai real-time processing done on any raw data in the active cortex component of the Ethernos Wallet during the cognitive extraction phase. A set of different Ai modules interpret the fed data and extra the cognitive information out of it comparing raw data against previously-stored ML cognitive models from a local wallet DB. The extracted semantic information is virtually stored in cognitive blocks. Together with the factual block, the cognitive blocks are the smallest semantic storage unit in Ethernos.
Each cognitive block has an ML signature that is inherited from the model that helped to extract it from the raw data.
Factual Blocks
The factual blocks are the output of the 2nd step of the processing executed in the active cortex during the factual extraction phase. The factual blocks contain more precise information that defines the characteristics of the cognitive extraction and if any the relation with personal ML Models. The factual blocks are always complimentary information of a cognitive block. E.g guitar, old guitar. CB1FB1- An old guitar. The factual blocks are extracted on the basis of understanding the cognitive block bonding to them. E.g. A guitar object is detected, during the factual check characteristics of the guitar are extracted, in the current example is 'old'.
During the factual extraction phase, the factual blocks are also compared against ML personal models from the Childhood learning database to see if there are found in the previous memory or if they are different. If a match is found, the processing factual block found in the childhood learning DB, For E.g. 'An old guitar' block found during the processing looking like 'my old guitar' compared on the basis of 'guitar' cognitive block signature will be indexed in the meta chain ledger with the factual block identifier of my old guitar- CB1 FBasafa221
If a match is not found the factual block keeps the virtual naming obtained during processing- FB1 and represents on a semantic output level an old guitar, which in the given case example is not ' my old guitar' but another old guitar, that will be added to the memory meta chain ledger as FB1 in meta pair with the cognitive block CB1 as CB1FB1.
The Block Ledger
Is a leger that holds all the timestamped factual and cognitive blocks of information of one user and the signature of the ML model that was used to extract the cognitive and factual information of that specific block.
The ML Model Signature
In the Ethernos the real-time AI modules process in thread batches all the new information. During the processing, each module uses different types of ML models to extract the semantics of the raw files processed. The ML model gallery is composed of generally available models for generic semantic extraction and user's particular models manually trained by the user during the childhood learning phase. The user trained models represent important elements defining the day-to-day life E.g. 'the ML model of the wife' or the 'model of my dog'. Each of these models has its unique signature that during the factual extraction and during the memory staking helps to understand in what entry or meta chain of memory the same model was used before and based on that to create affinity pools of similar memories.
Childhood learning phase
Is a process of manual labeling of different ML models. It is done manually and step by step by the user from the Ethernos Wallet interface as a first step prior to usage. It is a critical process and it represents the reference outline of the memory of each user and the foundation for the automatic processing and staking. During this process, the user trains Ethernos what are the important pieces that define their past memory and their day-to-day activities. Key moments, memories, events, locations, objects, people, feelings are transformed in personal ML models.
These personal models are stored in a DB inside the wallet DB. Each personal model inherits the cognitive block associated with it . E.g my father, cognitive block for-Man, factual block -my father and a unique signature that helps a faster identification within the meta chain ledger.
The generic ML models such as for e.g. Guitar, hold the cognitive block for 'guitar' logged in the block ledger and are identical to all users within Ethernos. The meta pair for factual is usually empty.
The ML models have a unique signature that is used during factual extraction for 1 to 1 comparison between old memories and processing memories and also during the anti-proof of history to remove the block duplicates in the meta chain ledger and for the same facts to always use one reference element.
Memory Meta Pairs
Are structures of connected factual and cognitive blocks.
Memory Meta chains
Are the fundamental organizational structures of memory that contain the composing cognitive and factual blocks, the meta pairs, the raw files, and the file metadata associated with a staked indexed memory. They are stored and time stamped as entries in the Meta Chain Ledger of each user.
Chain Sequences
Are a chronological set of meta chains that represent successive memories ordered by chronological or geolocation signature.
Pre-Cortex
Is the component inside any of the wallets responsible for analyzing data in real-time, extracting and understanding if that exact data file was interpreted before or not. The main responsibility is to bypass the processing of existing files and prepare any unprocessed file for the Active cortex engine.
Pre-Cortex Ledger
Is the ledger that holds time-stamped information about all the new data files and data sources that were processed and is used by the Pre-cortex as a source of truth to look for similar files a
Pre-Cortex Staking
Is the process of validating if a file was processed or not before by a wallet and if yes in what memories that happened. Alternatively the new files are logged in the Pre-cortex ledger and processed further to the active cortex component.
Active Cortex Engine
Is the wallet component that runs all the AI detection, responsible for factual and cognitive information extraction. At the end of the file processing cycle, a memory meta chain is created and a new entry is written in the user meta chain ledger.
Cortex Caching layer
Is a virtual cache running inside the cortex that contains and holds the last 100 meta chains of memory and the meta chain ledger entries associated with them. The cache also contains information about the affinity pools associated with any of the 100 last memory meta chains. The main responsibility of it is to boost the processing speed by acting as a shortlist of the last memories that were processed. Accessing the recent memories more rapidly also boosts the performance of the anti-proof of history process and optimizes the speed of indexing of the new memories in affinity pools.
User Meta chain ledger
Is the fundamental timestamped and ordered ledger that holds all the memory meta chain entries and their relation with factual blocks, cognitive blocks, meta pairs and affinity pools.
Affinity pools
Are virtual groups of similarities between entire memory meta chains, cognitive and factual blocks or combinations of many of them. Affinity pools are critical because they help generate not only deep relational structures but also shortcut the navigation between memories. Affinity pools are very useful to identify the relations between memories without the need to parse entirely the meta chain ledger. They facilitate direct jumps or zoom in to specific memories. To understand affinity pools as a parallel to the way we think about things is like for e.g. 'thinking about your favorite food and you remember all the memories when you had that food'- these memories are indexed by Ethernos in affinity pools.
Anti-proof of History
The local memory meta chain staking mechanism is responsible for keeping the ledger ordered chronologically, respecting the entry time of a memory, and removing the duplication entries of identical memories in the ledger. The process of indexing is based on similarities to all the new memories in existing affinity pools. It checks for similarities between new meta chains and geolocation matching old meta chains without an affinity pool and when similarities are found it generates new affinity pools that contain the respective similar meta chains.
During the Anti-proof of history, the Cortex caching layer is updated and the last 100 memory meta chains are replaced one by one by newer entries. During the processing, the caching list is also updated with the newly organized affinity pool of each new memory.
During the anti-proof of history staking, the raw files of the new factual blocks that have no ML personal signature are always compared against all other blocks without an ML signature, but that have the same cognitive path. If the compared extracted data from the raw files is the same the block reference is replaced by the older block in both of the memory meta chains entries and the affinity pool is staked correspondingly. If no matching is found the new blocks are indexed as individual meta chains in the ledger.
E.g. Blue armchair having the signature of the armchair model is compared during the anti proof of history with all the blocks where the armchair signature was found and the ones that look the same are indexed as the first armchair block, the repetitive alike block name is always replaced by the oldest block. This mechanism is very important for extracting relations between repetitive unlabelled memory blocks and evolving the storage efficiency for all the new data sources.
Ethernos Wallet
An application for storing memories with a user-friendly interface for labeling, editing, removing, and adding memories as well as sharing with other people and exploring other people's memories. The wallet can be used on mobile devices and computers. It is also responsible for real-time captured data processing and memory staking. The pre-cortex, cortex and the anti-proof of history processes run natively on the hardware where the wallet is installed.
A user can use multiple wallets that can capture and stake memories in parallel and synchronize based on the historical order of meta chains in the ledger and re-indexing of the overlapping memories.
Master Wallet & Secondary Wallets
A user that uses multiple wallets must choose which of these wallets is the Master. This is the destination of backup to of any secondary wallet. The master wallet is also the one running the vault staking.
Ethernos Cloud Brain Storage
Is an encrypted cloud container dedicated to a unique users. The cloud brain is responsible for keeping all the historical raw files associated with the processed memories.
Synching
The process of synching meta chain ledgers of multiple devices that collected data asynchronous on behalf of the same user. The synching can be done via cloud or via local connectivity on-premise means.
VAULT Staking
Is the process of updating the MetaChain Ledgers from multiple devices and validating potential memory duplicates and affinities when multiple chains overlap the same timeline after a synching process takes place.
If data is added to the meta chain ledger from multiple sources in parallel the anti-proof of history orders chronologically the meta chains processed by multiple wallets of the same user, by respecting the rule of time. A double verification takes place and all the meta chain entries from all the sources are reanalyzed. In this particular case, the caching layer is cleaned on all the wallets and updated with a list of the last 100 staked memory meta chains. Ethernos Skull
The Skull is a storage capsule running on the same device with the master wallet and/or in the cloud and is always associated with a wallet ID. The skull contains an up-to-date clone of the user meta chain ledger and the raw files associated with the memory.
Overlapping multiple user memories in the network can be identified by colliding different SKULLS and verifying meta chains where between geo-location, timestamp, device, and device connectivity network exists a match. This helps to rapidly establish a similarity mapping between multiple people's memories on a physical overlapping point in time without keeping the details on the public network layer. Granular association of the memories can be extracted on the spot by verifying the ML model signatures and cognitive activity similarities from the user meta chain ledgers. On the fly, cross skull user affinities can be generated as a result of detecting similar cognitive elements within the memories of the users without accessing the raw files. The readability of the meta chain ledger of any user is subjected to a configurable privacy setup of the ledger itself or of the memory itself that is logged in the meta chain entry.
DIMSP- Decentralized Internet Memory Storage Protocol
Is a brain sharding backup process of storing over the Internet- on encrypted storage locations memories and affinity pools of memories using the brain pre-cortex ledger and meta chains ledger, referencing in a history log of those the locations of each file and the encryption signature key to decrypt and facilitate the access to those.
This requires each storage point to run an ETHERNOS SKULL- a container local application responsible for storing, encrypting, and validating the handshake with peer requests coming from third parties trying to recreate the integrity of a memory or affinity pool
DIMSP Router
Is the mechanism responsible for assigning to each new memory the external backup and storage path to the cloud backup wallet.
DeDS
Decentralized Data Storage as technology and architecture under the DeFi umbrella, focused on storing large data in sharded locations using cross-device blockchain ledgers as proof of history and validation
Last updated
Was this helpful?