-
公开(公告)号：US20230153633A1
公开(公告)日：2023-05-18
申请号：US17766955
申请日：2019-10-07
IPC分类号： G06N3/098
CPC分类号： G06N3/098
摘要： A method for training a central model in a federated learning system is provide. The method includes receiving a first update from a first local model of a set of local models; receiving a second update from a second local model of the set of local models; enqueueing the first update and the second update in one more queues corresponding to the set of local models; selecting an update from the one or more queues to apply to a central model based on determining that a selection criteria is satisfied, the selection criteria being related to a quality of the central model; and applying the selected update to the central model or instructing a node to apply the selected update to the central model.
-
公开(公告)号：US20220408213A1
公开(公告)日：2022-12-22
申请号：US17773392
申请日：2019-10-30
发明人： Swarup Kumar MOHALIK , Saravanan M , Anshu SHUKLA
摘要： In a method performed by a first node, the node determines a plan based on: i) a first goal to be achieved according to capabilities of the first node, and a first set of first actions to be performed by the first node to achieve the first goal, and ii) for each second node of one or more second nodes: a) a respective goal to be achieved according to respective capabilities of each second node, and b) a respective set of first actions to be individually performed by each second node to achieve the respective goal. The plan is to collaboratively achieve each of the respective goals and the first goal, by determining a respective set of second actions to be respectively performed by each second node and the first node. The first node sends a respective indication indicating the determined second actions.
-
公开(公告)号：US20210365937A1
公开(公告)日：2021-11-25
申请号：US16963349
申请日：2018-02-14
发明人： Nanjangud Chandrasekhara Swamy NARENDRA , Ramachandran KRISHNASAMY , Sambit NAYAK , SYED NADEEMULLA R. , Anshu SHUKLA
摘要： It is provided a method for managing a smart contract in real-time. The method is performed in a contract manager and comprises the steps of: obtaining a base version of the smart contract between the supplier and a first purchaser; recommending amendments to the base version of the smart contract, based on historic contract compliance data of the supplier, wherein the historic contract compliance data is based on smart contracts with the supplier and a plurality of purchasers; receiving a signal indicating an agreed smart contract between the first purchaser and the supplier; receiving a real-time monitoring signal relating to a compliance of the supplier in relation to at least one condition of a smart contract between the supplier and a second purchaser; and recommending amendments to the agreed smart contract, based on the monitoring signal.
-
公开(公告)号：US20210234704A1
公开(公告)日：2021-07-29
申请号：US17053666
申请日：2018-05-18
摘要： A method and a network node for managing access, by a device, to a blockchain. The network node receives, a first request for creation of a first account. The first request includes information relating to a user of the device and a first hash value. The network node creates, based on the information and the first hash value, the first account on the blockchain, whereby a first address is obtained. The network node receives, a second request for recovering account information of the first account. The second request includes the first address of the first account, a source value from which the first hash value is derived and a second address of a second account. The network node generates a second hash value based on the source value. The network node initiates transfer of the account information of the first account to the second account.
-
公开(公告)号：US20220351039A1
公开(公告)日：2022-11-03
申请号：US17766025
申请日：2019-10-04
IPC分类号： G06N3/08
摘要： A method on a central node or server is provided. The method includes: receiving a first model from a first user device and a second model from a second user device, wherein the first model is of a neural network model type and has a first set of layers and the second model is of the neural network model type and has a second set of layers different from the first set of layers; for each layer of the first set of layers, selecting a first subset of filters from the layer of the first set of layers, for each layer of the second set of layers, selecting a second subset of filters from the layer of the second set of layers; constructing a global model by forming a global set of layers based on the first set of layers and the second set of layers, such that for each layer in the global set of layers, the layer comprises filters based on the corresponding first subset of filters and/or the corresponding second subset of filters; and forming a fully connected layer for the global model, wherein the fully connected layer is a final layer of the global set of layers.
-
6.发明申请公开(公告)号：US20210312472A1
公开(公告)日：2021-10-07
申请号：US17266890
申请日：2018-08-17
摘要： A method implemented by a network device to predict smart contract violations, where the network device is communicatively coupled to one or more computing nodes that execute one or more smart contracts that are configured to perform transactions on a blockchain. The method includes generating a control flow graph for each of the one or more smart contracts, retrieving, from the blockchain, a current state of a variable utilized by the one or more smart contracts following execution of a transaction that updates the variable, generating or updating a state space tree based on the current state of the variable and the control flow graphs generated for the one or more smart contracts, determining whether the state space tree meets a predefined violation requirement, and sending an alert to the one or more computing nodes in response to determining that the state space tree meets the violation requirement.
-
公开(公告)号：US20210194890A1
公开(公告)日：2021-06-24
申请号：US17273087
申请日：2018-09-14
发明人： Swarup Kumar MOHALIK , Ramamurthy BADRINATH , Sandhya BASKARAN , Perepu SATHEESH KUMAR , Anshu SHUKLA
摘要： A method and a system are presented for enabling coordinated executions of actions in a distributed computing system with untrusted local computing agents. A smart contract including plans is recorded in a blockchain database. Each plan includes actions to be executed by a respective one of the local computing agents. Execution of a first action of a first set of actions to be executed by a first local computing agent is requested. Execution of a second action of a second set of actions to be executed by a second local computing agent is requested. Responsive to determining, based on the smart contract, that the first action can be executed, the first local computing agent is caused to execute the first action and responsive to determining, based on the smart contract, that the second action cannot be executed, the second local computing agent is caused to not execute the second action.
-
8.发明申请公开(公告)号：US20200322308A1
公开(公告)日：2020-10-08
申请号：US16956666
申请日：2017-12-29
摘要： A method and a network device in a cloud computing system, including a first blockchain component that is one of a plurality of blockchain components forming a blockchain system, of dynamic discovery of another blockchain component of the blockchain system are described. A multicast address including a multicast group identifier is generated. The multicast group identifier is generated at least in part based on a genesis block identifier that uniquely identifies a blockchain serviced by the blockchain system. The network device joins a multicast group identified by the multicast group identifier; and transmits a message destined to the multicast address, where the message includes a request for a unicast address of another component of the blockchain system.
-
公开(公告)号：US20230004322A1
公开(公告)日：2023-01-05
申请号：US17780427
申请日：2019-11-29
发明人： Sambit NAYAK , Fredrik SVENSSON , Anshu SHUKLA
IPC分类号： G06F3/06
摘要： A method for managing provenance information associated to one or more interconnected provenance entities in a provenance system for data processing pipelines in a distributed cloud environment over a network interface, wherein each of the data processing pipelines is configured to read in data, transform the data, and output transformed data is disclosed. The method comprises steps being performed by a configuration component of obtaining at least one declarative intent representing a configuration indicative of requirements and levels of priority for storage of provenance information for each of the data processing pipelines, deriving the requirements and levels of priority for storage of provenance information for each of the data processing pipelines based on the obtained at least one declarative intent, wherein one of the levels of priority—first level of priority—is higher than the other levels of priority—second levels of priority, estimating storage capacity for storage of provenance information in the provenance system based on the derived requirements and levels of priority, storing the provenance information according to the derived requirements and levels of priority for storage of provenance information and for each of the data processing pipelines, and when actual storage consumption for storage of provenance information in the provenance system meets a threshold of storage capacity set based on the estimated storage capacity: reducing a data amount for storage of provenance information of the second levels of priority in the provenance system. Corresponding computer program product, arrangement, configuration component, and system are also disclosed.
-
10.发明申请公开(公告)号：US20220094646A1
公开(公告)日：2022-03-24
申请号：US17422912
申请日：2019-01-17
IPC分类号： H04L49/9005 , G06F15/173 , H04L49/901 , H04L49/90 , H04L67/1097
摘要： A method and network interface card providing central processor unit efficient storing of data. The NIC receives request for registering a memory address range in the NIC, the request comprising a rewrite protection granularity for the memory address range. When receiving data from a client process, subsequent to registering of said memory address range, said data having an address within the memory address range, the NIC determines whether the rewrite protection granularity of the NIC is reached, when receiving said data. In the event that the rewrite protection granularity is reached, the NIC inactivates the memory address range according to said reached rewrite protection granularity. The auto-inactivated memory address range also provides a rewrite protection of data when storing data. Remote logging or monitoring of data is also enabled, wherein the logging or monitoring may be regarded to become server-less.
-
-
-
-
-
-
-
-
-
搜索结果
12 条记录 (耗时 0.019 秒)
