Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for performing protein design. In one aspect, a method comprises: processing an input characterizing a target protein structure of a target protein using an embedding neural network having a plurality of embedding neural network parameters to generate an embedding of the target protein structure of the target protein; determining a predicted amino acid sequence of the target protein based on the embedding of the target protein structure, comprising: conditioning a generative neural network having a plurality of generative neural network parameters on the embedding of the target protein structure; and generating, by the generative neural network conditioned on the embedding of the target protein structure, a representation of the predicted amino acid sequence of the target protein.

Abstract: Techniques for proving authenticity of data files, such as digital photos. An authentication/indexing server content from a client device such as a mobile device, and generates or receives other related data such as a reference to the content object, a fingerprint for the content object, and other metadata indicating for example a timestamp related to the content object. The fingerprint is inserted into a distributed ledger, with the distributed ledger returning a ledger address. A manifest object containing the transaction object, the ledger address, and a list of the content object and, optionally, related edits is then created and stored. The server may index fingerprints. Thus, a fingerprint can be used to authenticate subsequently received object against the stored metadata.

Abstract: A digital asset is represented and verified as a set of related digital asset or other content objects. Related metadata is stored on an immutable distributed ledger separately from the content objects themselves. For example, a transaction object includes metadata such as identifiers for two or more content objects, fingerprints for the content objects. The content objects may be stored in a local or cloud object repository. Validation of a later identified content object may include determining a fingerprint for the later identified content object, mapping that fingerprint to an address within the immutable distributed ledger to retrieve metadata previously mapped, and comparing the two fingerprints. Visual validation may be provided when the first and second fingerprints match, such as by displaying a positive icon adjacent the later identified object.

Abstract: Techniques for authenticating digital media asset, and particularly to verifying the authenticity of digital image or video files which may have been redacted or otherwise altered in some way. The approach, which we call perceptual video fingerprinting, compares media assets based on what humans perceive, rather than exactly which bits match, by using neural network algorithms.

Abstract: A digital asset is represented and verified as a set of related digital asset or other content objects. Related metadata is stored on an immutable distributed ledger separately from the content objects themselves. For example, a transaction object includes metadata such as identifiers for two or more content objects, fingerprints for the content objects. The content objects may be stored in a local or cloud object repository. Validation of a later identified content object may include determining a fingerprint for the later identified content object, mapping that fingerprint to an address within the immutable distributed ledger to retrieve metadata previously mapped, and comparing the two fingerprints. Visual validation may be provided when the first and second fingerprints match, such as by displaying a positive icon adjacent the later identified object.

Abstract: A digital asset is represented and verified as a set of related digital asset or other content objects. Related metadata is stored on an immutable distributed ledger separately from the content objects themselves. For example, a transaction object includes metadata such as identifiers for two or more content objects, fingerprints for the content objects. The content objects may be stored in a local or cloud object repository. Validation of a later identified content object may include determining a fingerprint for the later identified content object, mapping that fingerprint to an address within the immutable distributed ledger to retrieve metadata previously mapped, and comparing the two fingerprints. Visual validation may be provided when the first and second fingerprints match, such as by displaying a positive icon adjacent the later identified object.

Abstract: Techniques for authenticating digital media asset, and particularly to verifying the authenticity of digital image or video files which may have been redacted or otherwise altered in some way. The approach, which we call perceptual video fingerprinting, compares media assets based on what humans perceive, rather than exactly which bits match, by using neural network algorithms.

Abstract: Techniques for proving authenticity of data files, such as digital photos. An authentication/indexing server content from a client device such as a mobile device, and generates or receives other related data such as a reference to the content object, a fingerprint for the content object, and other metadata indicating for example a timestamp related to the content object. The fingerprint is inserted into a distributed ledger, with the distributed ledger returning a ledger address. A manifest object containing the transaction object, the ledger address, and a list of the content object and, optionally, related edits is then created and stored. The server may index fingerprints. Thus, a fingerprint can be used to authenticate subsequently received object against the stored metadata.

Abstract: A digital asset is represented and verified as a set of related digital asset or other content objects. Related metadata is stored on an immutable distributed ledger separately from the content objects themselves. For example, a transaction object includes metadata such as identifiers for two or more content objects, fingerprints for the content objects. The content objects may be stored in a local or cloud object repository. Validation of a later identified content object may include determining a fingerprint for the later identified content object, mapping that fingerprint to an address within the immutable distributed ledger to retrieve metadata previously mapped, and comparing the two fingerprints. Visual validation may be provided when the first and second fingerprints match, such as by displaying a positive icon adjacent the later identified object.

Abstract: A method for assessing the risk and cost for data loss and disaster recovery (DR) plans includes providing an application having a graphical user interface (GUI) comprising first and second windows arranged adjacent to each other. The first window comprises a catalog of components used to generate data disaster recovery (DR) configurations and the second window displays the generated DR configurations. A first DR configuration is generated in the second window and components are added to the first DR configuration by dragging and dropping components from the catalog into appropriate locations of the second window. Metrics for the first DR configuration are calculated and reported in the second window. A second configuration is also similarly generated in the second window and the metrics results are graphically compared to each other.

Abstract: A cloud storage replica of a production device is created and asynchronously updated. A snap, clone or portion of the replica is copied to a cloud gateway in a failover situation. Access to replica images may be obtained via other cloud gateways.

Abstract: A cloud storage gateway is used to migrate a volume from a source cloud storage provider to a target cloud storage provider. If necessary, the gateway reconfigures the size of a logical cache associated with local storage devices so that the entire volume can be stored locally. Flushing is disabled and all of the locally stored volume pages are marked as dirty. After compression and encryption policies associated with the source and target are verified to be compatible, flushing is enabled and metadata is updated.

Abstract: A system for resource sharing across multi-cloud storage arrays includes a plurality of storage arrays and a cloud array storage (CAS) application. The plurality of storage resources are distributed in one or more cloud storage arrays, and each storage resource comprises a unique object identifier that identifies location and structure of the corresponding storage resource at a given point-in-time. The cloud array storage (CAS) application manages the resource sharing process by first taking an instantaneous copy of initial data stored in a first location of a first storage resource at a given point-in-time and then distributing copies of the instantaneous copy to other storage resources in the one or more cloud storage arrays.

Abstract: A method for data deduplication includes the following steps. First, segmenting an original data set into a plurality of data segments. Next, transforming the data in each data segment into a transformed data representation that has a band-type structure for each data segment. The band-type structure includes a plurality of bands. Next, selecting a first set of bands, grouping them together and storing them with the original data set. The first set of bands includes non-identical transformed data for each data segment. Next, selecting a second set of bands and grouping them together. The second set of bands includes identical transformed data for each data segment. Next, applying a hash function onto the transformed data of the second set of bands and thereby generating transformed data segments indexed by hash function indices. Finally, storing the hash function indices and the transformed data representation of one representative data segment in a deduplication database.

Abstract: A system for resource sharing across multi-cloud storage arrays includes a plurality of storage arrays and a cloud array storage (CAS) application. The plurality of storage resources are distributed in one or more cloud storage arrays, and each storage resource comprises a unique object identifier that identifies location and structure of the corresponding storage resource at a given point-in-time. The cloud array storage (CAS) application manages the resource sharing process by first taking an instantaneous copy of initial data stored in a first location of a first storage resource at a given point-in-time and then distributing copies of the instantaneous copy to other storage resources in the one or more cloud storage arrays.

Abstract: A multi-cloud data replication method includes providing a data replication cluster comprising at least a first host node and at least a first online storage cloud. The first host node is connected to the first online storage cloud via a network and comprises a server, a cloud array application and a local cache. The local cache comprises a buffer and a first storage volume comprising data cached in one or more buffer blocks of the local cache's buffer. Next, requesting authorization to perform cache flush of the cached first storage volume data to the first online storage cloud. Upon receiving approval of the authorization, encrypting the cached first storage volume data in each of the one or more buffer blocks with a data private key. Next, assigning metadata comprising at lest a unique identifier to each of the one or more buffer blocks and then encrypting the metadata with a metadata private key.

Abstract: A method for data deduplication includes the following steps. First, segmenting an original data set into a plurality of data segments. Next, transforming the data in each data segment into a transformed data representation that has a band-type structure for each data segment. The band-type structure includes a plurality of bands. Next, selecting a first set of bands, grouping them together and storing them with the original data set. The first set of bands includes non-identical transformed data for each data segment. Next, selecting a second set of bands and grouping them together. The second set of bands includes identical transformed data for each data segment. Next, applying a hash function onto the transformed data of the second set of bands and thereby generating transformed data segments indexed by hash function indices. Finally, storing the hash function indices and the transformed data representation of one representative data segment in a deduplication database.

Abstract: An asynchronous peer-to-peer data replication method implemented within a replication cluster comprising at least one master node and at least a first client node includes entering an update in a data volume of the master node and storing the update in a master node storage. Next, updating a first active session in a master log and then sending a first message from the master node to the first client node. The first message comprises a first message content and first “piggybacked” data indicating that the first active session in the master log was updated. Next, receiving the first message by the first client node, registering that the first active session in the master log was updated and signaling internally that the first active session in the master log was updated. Next, sending an update request from the first client node to the master node, processing the update request by the master node and sending the update to the first client node.

Abstract: A system for resource sharing across multi-cloud storage arrays includes a plurality of storage arrays and a cloud array storage (CAS) application. The plurality of storage resources are distributed in one or more cloud storage arrays, and each storage resource comprises a unique object identifier that identifies location and structure of the corresponding storage resource at a given point-in-time. The cloud array storage (CAS) application manages the resource sharing process by first taking an instantaneous copy of initial data stored in a first location of a first storage resource at a given point-in-time and then distributing copies of the instantaneous copy to other storage resources in the one or more cloud storage arrays.

Abstract: A multi-cloud data replication method includes providing a data replication cluster comprising at least a first host node and at least a first online storage cloud. The first host node is connected to the first online storage cloud via a network and comprises a server, a cloud array application and a local cache. The local cache comprises a buffer and a first storage volume comprising data cached in one or more buffer blocks of the local cache's buffer. Next, requesting authorization to perform cache flush of the cached first storage volume data to the first online storage cloud. Upon receiving approval of the authorization, encrypting the cached first storage volume data in each of the one or more buffer blocks with a data private key. Next, assigning metadata comprising at lest a unique identifier to each of the one or more buffer blocks and then encrypting the metadata with a metadata private key.