Abstract: An adaptive document curation method. A machine learning model is trained on a first library of documents to assign staleness scores to documents, each indicating a confidence that the document to which it is assigned should be archived. If a document has a staleness score at least equal to a threshold, the document is identified as a candidate to be archived or deleted.

Abstract: Methods, computer readable media, and devices for machine learning for automated tagging of requests for e-commerce conversion funnel tracking. One method may include retrieving a plurality of requests previously grouped into a user session, generating preprocessed request information by determining a request type, a relative time, and a keyword count for each of the plurality of requests and determining a determining a relative time distribution for each of the plurality of keywords, and applying a machine learning model to the plurality of requests and the preprocessed request information to tag each of the plurality of requests as being associated with one of the plurality of conversion classifications.

Abstract: Methods, computer readable media, and devices for predictive product availability for grocery delivery are presented. A method may include determining a shopping location and a future delivery window. A predicted availability of one or more grocery product items may be generated based on the location and delivery window. If the predicted availability of the items exceeds a threshold, the items may be presented to a user for selection as part of an order. If the predicted availability of the items does not exceed a threshold, alternative shopping locations and/or alternative future delivery windows may be presented to the user for selection. A machine learning algorithm may be implemented to generate the predicted availability of the one or more grocery product items, the one or more alternative shopping locations, and/or the one or more future delivery windows.

Abstract: An adaptive document curation method. A machine learning model is trained on a first library of documents to assign staleness scores to documents, each indicating a confidence that the document to which it is assigned should be archived. If a document has a staleness score at least equal to a threshold, the document is identified as a candidate to be archived or deleted.

Abstract: An adaptive document curation method. A machine learning model is trained on a first library of documents to assign staleness scores to documents, each indicating a confidence that the document to which it is assigned should be archived. If a document has a staleness score at least equal to a threshold, the document is identified as a candidate to be archived or deleted.

Abstract: An adaptive document curation method. A machine learning model is trained on a first library of documents to assign staleness scores to documents, each indicating a confidence that the document to which it is assigned should be archived. If a document has a staleness score at least equal to a threshold, the document is identified as a candidate to be archived or deleted.

Abstract: A method of identifying a task to be completed in a task-tracking system. A first task to be completed by a user is used to identify a second task that can be completed by the user. A similarity score indicating a similarity of the second task to the first task can be used to identify the second task as being related to the first task.

Abstract: An apparatus includes a fuel cell subsystem, a first circuit and a second circuit. The fuel cell subsystem produces power for a load in response to receiving a flow of a reactant, and the first circuit is coupled to the fuel cell subsystem to halt the flow to the fuel cell stack and isolate the fuel cell stack from the load after the flow is halted. The second circuit monitors a characteristic of the fuel cell subsystem and receives power from the stack to dissipate a portion of the reactant that remains in the stack while the flow is halted.

Abstract: A system includes a fuel cell subsystem, switches and a controller. The fuel cell subsystem is adapted to provide power that is capable of being consumed by residential loads, and the fuel cell subsystem is sized to provide power up to a first power threshold that is less than a maximum power threshold that is capable of being consumed by the residential loads. The controller is adapted to determine the power that is consumed by the residential loads and based on the determined power, operate the switches to selectively regulate electrical connections between the residential loads and the fuel cell subsystem to keep the power approximately below the first power threshold.

Abstract: A method includes establishing a fuel flow through a fuel cell stack to produce a current. The voltages of the fuel cell stack are scanned to determine the minimum cell voltage. A maximum limit on the current is set based at least in part on the minimum cell voltage.

Abstract: An apparatus includes a fuel cell subsystem, a first circuit and a second circuit. The fuel cell subsystem produces power for a load in response to receiving a flow of a reactant, and the first circuit is coupled to the fuel cell subsystem to halt the flow to the fuel cell stack and isolate the fuel cell stack from the load after the flow is halted. The second circuit monitors a characteristic of the fuel cell subsystem and receives power from the stack to dissipate a portion of the reactant that remains in the stack while the flow is halted.

Abstract: A fuel cell system includes a fuel cell stack, an inverter, a reformer and a circuit. The fuel cell stack is adapted to produce heat in response to a fuel flow, and the inverter is coupled to the fuel cell stack and is adapted to use electrical energy produced by the fuel cell stack to furnish electrical power that is received by a power grid. The reformer is adapted to produce the fuel flow, and the circuit is adapted to interact with the reformer to produce an approximate predetermined amount of heat. The circuit also determines cell voltages of the fuel cell stack, selects the minimum cell voltage, and interacts with the inverter to regulate the amount of the electrical power that is furnished by the inverter based on the selected voltage.

Abstract: A system includes a divider network, a memory and a circuit. The divider network is adapted to be coupled to fuel cells of a fuel cell stack and provide pairs of signals. Each pair of signals is associated with a different fuel cell and indicates a terminal voltage of the associated cell and another voltage common to the pair of signals. The memory stores indications of different common mode gains, and each common mode gain indication is associated with a different one of the pairs of signals. The circuit is coupled to the memory and adapted to generate an indication of the terminal voltage from each pair based on the associated common mode gain indication.

Abstract: A pair of small diameter optical fibers are arranged in a side-by-side configuration, the first fiber providing a passive guide for pump light, and the second fiber formed of a material, such as Nd:YAG which has amplifying properties at the signal frequency. The signal to be amplified propagates through the second fiber from the active material, resulting in amplification of the signal. The first fiber is pumped from one or both ends, utilizing a cone shaped rod which is optically coupled thereto. A large diameter beam of collimated pump light enters the cone shaped rod, where total internal reflections on the cone walls compress the beam to a small, fiber-size diameter for absorption by the active fiber. The light input to the cone-shaped rod is produced by plural, high power laser diodes, and is collimated by microlenses. By utilizing the cone-shaped rod to couple light from the high power diode lasers to the pump fiber, high pump power densities can be achieved, yielding increased amplification.

Abstract: A selected point in an electric power system is monitored for subsequent fault analysis. The voltage and current at the point is continuously sampled with samples for a predetermined period of time being temporarily stored in memory. In response to a Fault Event, voltage and current samples at differing sample rates and for differing periods of time are recorded. Voltage and current samples after a Cold Load Pickup are recorded, also the recorded samples allow later analysis of the fault and identification of the cause of the fault.