Abstract: A memory array of a compute tile may store activations or weights of a DNN. The memory array may include databanks for storing contexts, context MUXs, and byte MUXs. A databank may store a context with flip-flop arrays, each of which includes a sequence of flip-flops. A logic gate and an ICG unit may gate flip-flops and control whether states of the flip-flops can be changed. The data gating can prevent a context not selected for the databank from inadvertently toggling and wasting power A context MUX may read a context from different flip-flop arrays in a databank based on gray-coded addresses. A byte MUX can combine bits from different bytes in a context read by the context MUX. The memory array may be implemented with bit packing to reduce distance between the context MUX and byte MUX to reduce lengths of wires connecting the context MUXs and byte MUXs.

Abstract: This application relates to apparatus and methods for optimizing hyperparameters for machine learning models. In some examples, a computing device configures a machine learning model with a first set of hyperparameters from a pool of hyperparameters. The computing device may execute the machine learning model to generate a validation score, and may update parameters of a probability determination model based on the validation score. Further, the computing device may execute the probability determination model to generate probabilities corresponding to the first set of hyperparameters. The computing device may also determine a second set of hyperparameters from the pool of hyperparameters based on the generated probabilities, and may configure the machine learning model with the second set of hyperparameters.

Abstract: Systems and methods for message delivery prioritization that can include receiving a messaging request through an application programming interface, the messaging request being associated with message-associated delivery attributes; selecting a message transmission plan based on the message-associated delivery attributes, the message transmission plan including at least one timing priority related property; and executing the message transmission plan with a timing of transmission based in part on the timing priority property.

Abstract: A method for a fully distributed privacy-preserving multi-lateral netting between parties is disclosed. The method comprising collecting pending transactions between at least a part of the parties, publishing a commitment of pending transactions to a bulletin board, secretly sharing de-commitments among the parties, and publishing, by each party, its determined liquidity status, to the bulletin board. Furthermore, the method comprises upon determining a failure in the publishing of any liquidity status, performing the following MPC by determining whether all pending transactions are settable maintaining positive liquidity for each party, and upon determining a failure in the determining, for each party, whether all pending transactions are nettable determining an optimized subset of transactions that are nettable without violating liquidity requirements for each party.

Abstract: An apparatus is provided which comprises: a multi-bit quad latch with an internally coupled level sensitive scan circuitry; and a combinational logic coupled to an output of the multi-bit quad latch. Another apparatus is provided which comprises: a plurality of sequential logic circuitries; and a clocking circuitry comprising inverters, wherein the clocking circuitry is shared by the plurality of sequential logic circuitries.

Abstract: This application relates to apparatus and methods for training machine learning models. In some examples, a pool of worker pods are generated that can execute tasks to train a machine learning model. The pool of work pods are assigned tasks by a master that communicates with the worker pods using a work queue. Each worker pod can provide output using a results queue. The embodiments may operate with less reliable memory, such as object stores, which may be less costly than other types of storage mechanisms. To operate in less reliable environments, each worker pod can include a checkpoint mechanism that can recover from interruptions, such as interruptions due to node failure or preemption. For example, the checkpoint mechanism may allow a worker pod to continue processing a task, when the task is interrupted, from a last checkpoint. Processing results are provided to a results queue when a task completes.

Abstract: A new family of shared clock single-edge triggered flip-flops that reduces a number of internal clock devices from 8 to 6 devices to reduce clock power. The static pass-gate master-slave flip-flop has no performance penalty compared to the flip-flops with 8 clock devices thus enabling significant power reduction. The flip-flop intelligently maintains the same polarity between the master and slave stages which enables the sharing of the master tristate and slave state feedback clock devices without risk of charge sharing across all combinations of clock and data toggling. Because of this, the state of the flip-flop remains undisturbed, and is robust across charge sharing noise. A multi-bit time borrowing internal stitched flip-flop is also described, which enables internal stitching of scan in a high performance time-borrowing flip-flop without incurring increase in layout area.

Abstract: A fast Mux-D scan flip-flop is provided, which bypasses a scan multiplexer to a master keeper side path, removing delay overhead of a traditional Mux-D scan topology. The design is compatible with simple scan methodology of Mux-D scan, while preserving smaller area and small number of inputs/outputs. Since scan Mux is not in the forward critical path, circuit topology has similar high performance as level-sensitive scan flip-flop and can be easily converted into bare pass-gate version. The new fast Mux-D scan flip-flop combines the advantages of the conventional LSSD and Mux-D scan flip-flop, without the disadvantages of each.

Abstract: Systems, methods, and non-transitory computer-readable media can be configured to determine a request for a video from a client device, wherein the video is associated with a plurality of reframed versions of the video. A reframed version of the video to provide in response to the request can be determined based at least in part on a consumption surface associated with the client device. The reframed version of the video can be provided to the client device for display.

Abstract: An unstructured data input is accessed that includes an electronic communication. Content of the unstructured data is parsed to determine one or more terms in the unstructured data input. It is determined that one or more particular elements defined in a structured business data model correspond to the terms. Tags are assigned to the unstructured data based on the terms corresponding to the one or more particular elements, where the tags define an association between the unstructured data and the structured data model.

Abstract: An example method facilitates enabling Key Encryption Key (KEK) rotation for a running multi-tenant system without requiring system downtime or interruption. The example method facilitates decrypting a set of one or more DEKs using a preexisting KEK; using a new KEK to re-encode the DEKs using the new KEK, all while simultaneously enabling servicing of tenant requests. This is enabled in part, by strategic caching of tenant DEKs in a secure local memory, wherein the cached tenant DEKs are maintained in the clear and are readily accessible to running processes that are using the DEKs to decrypt and access tenant data, irrespective of the state of a background process used to implement the KEK rotation to the new KEK.

Abstract: A webpage serving system is provided and generally includes a server and a computing device. The server can receive a webpage request from the computing device, and determine a type of webpage being requested. The server may provide a document, such as an HTML document, to the computing device based on the type of webpage. The document may include instructions to download a particular bundle, which may include one or more files, from a server. The bundle identified in the instructions may be based on the type of webpage the document is for. The customer computing device may receive the requested bundle, and display the requested webpage based on the information received in the document and in the bundle.

Abstract: An interactive user interface is generated to illustrate a graphical view of a network of plan models. The network of plan models includes a plurality of plan models, each plan model including one or more measures and one or more dimensions. At least one user-entered text character is entered in a query field and the plan models are searched for names of measures and/or dimensions in the plan models that correspond to the user entry. The names are presented as search term suggestions in the search query field and at least one suggestion is selected and adopted as at least a portion of a search query. The network of plan models is searched to generate search results corresponding to the search query. The graphical view of the network of plan models can be revised based on the search results.

Abstract: Described is a system for contrast and entropy-based perception adaption to optimize perception. The system is operable for receiving an input image of a scene with a camera system and detecting one or more objects (having perception data) in the input image. The perception data of the one or more objects is converted into probes, which are then converted into axioms using probabilistic signal temporal logic. The axioms are evaluated based on probe bounds. If the axioms are within the probe bounds, then results are provided; however, if the axioms are outside of the probe bounds, the system estimates optimal contrast bounds and entropy bounds as perception parameters. The contrast and entropy in the camera system are then adjusted based on the perception parameters.

Abstract: Systems and methods for loyalty estimation are disclosed. A set of assortments each including a subset of a plurality of items is received and a loyalty score is calculated for each item in the plurality of items. The loyalty value is calculated based on a relationship between assortments including the item and assortments without the item.

Abstract: A system ground surface projection for autonomous driving of a host vehicle is provided. The system includes a LIDAR device of the host vehicle and a computerized device. The computerized device is operable to monitor data from the LIDAR device including a total point cloud. The total point cloud describes an actual ground surface in the operating environment of the host vehicle. The device is further operable to segment the total point cloud into a plurality of local point cloud and, for each of the local point clouds, determine a local polygon estimating a portion of the actual ground surface. The device is further operable to assemble the local polygons into a total estimated ground surface and navigate the host vehicle based upon the total estimated ground surface.

Abstract: Described is a system to evaluate and reduce perception error in object detection and recognition. The system includes a perception module that receives perception data (of an object(s)) from an environment proximate a mobile platform. Perception probes are generated that describe one or more characteristics of the objects. The perception probes are converted into probabilistic signal temporal logic (PSTL)-based constraints that provide axioms having statistical analysis of the perception probes. The axioms are evaluated to classify the perception probes as valid or erroneous. Optimal perception parameters are generated by solving an optimization problem based on the axioms, which allows the system to adjust the perception module based on the optimal perception parameters.