Abstract: Systems and methods for machine learning architecture for time series data prediction. The system may include a processor and a memory storing processor-executable instructions. The processor-executable instructions, when executed, may configure the processor to: obtain time series data associated with a data query; generate a predicted value based on a sampled realization of the time series data and a continuous time generative model, the continuous time generative model trained to define an invertible mapping to maximize a log-likelihood of a set of predicted values for a time range associated with the time series data; and generate a signal providing an indication of the predicted value associated with the data query.

Abstract: A bearing device in a radial flow adsorber with a gas flow guide function includes a bearing platform, a supporting cylinder, a flow guide pipe, an annular bottom plate, a limiting block and a supporting frame, wherein an upper portion of the bearing platform is filled with adsorbent, an interior of a lower portion of the bearing platform is connected with the annular bottom plate through the supporting cylinder, a backing plate is arranged between the bearing platform and the supporting cylinder, on the supporting cylinder or the annular bottom plate is opened a hole and arranged a plurality of flow guide pipes for guiding gas flow, the annular bottom plate is placed on the supporting frame, the limiting block is arranged on the supporting frame, and the supporting frame is fixed on a cylinder body or a lower head of the adsorber by welding.

Abstract: A sensor package includes at least one die, a fingerprint sensor, a mold material, and a land grid array. The fingerprint sensor is electrically coupled to the at least one die. The mold material encapsulates the at least one die. The land grid array layer is electrically coupled to the at least one die. The land grid array layer, the fingerprint sensor, and the mold material each include a common footprint.

Abstract: A sensor package includes at least one die, a fingerprint sensor, a mold material, and a land grid array. The fingerprint sensor is electrically coupled to the at least one die. The mold material encapsulates the at least one die. The land grid array layer is electrically coupled to the at least one die. The land grid array layer, the fingerprint sensor, and the mold material each include a common footprint.

Abstract: An underground mining method for unexploited coal in a boundary open-pit mine is provided. A shaft construction platform is arranged at one rock step to two rock steps above a coal seam. Intermediate bridges are built starting from a pit bottom. Mining area clay is laid on a working slope where no intermediate bridge is built and on a side slope with an outcrop of the coal seam as a sealing layer to seal the slopes. Auxiliary vertical shafts and main inclined shafts are dug. The pit bottom is dug downward to form a digging space on a side close to the working slope between two adjacent ones of the intermediate bridges, and clay is filled into the digging space to form an artificial water barrier layer. A roadway communicating the main inclined shafts and the auxiliary vertical shafts is constructed, and a coal seam stope face is arranged.

Abstract: An underground mining method for unexploited coal in a boundary open-pit mine is provided. A shaft construction platform is arranged at one rock step to two rock steps above a coal seam. Intermediate bridges are built starting from a pit bottom. Mining area clay is laid on a working slope where no intermediate bridge is built and on a side slope with an outcrop of the coal seam as a sealing layer to seal the slopes. Auxiliary vertical shafts and main inclined shafts are dug. The pit bottom is dug downward to form a digging space on a side close to the working slope between two adjacent ones of the intermediate bridges, and clay is filled into the digging space to form an artificial water barrier layer. A roadway communicating the main inclined shafts and the auxiliary vertical shafts is constructed, and a coal seam stope face is arranged.

Abstract: A variational hyper recurrent neural network (VHRNN) can be trained by, for each step in sequential training data: determining a prior probability distribution for a latent variable from a prior network of the VHRNN using an initial hidden state; determining a hidden state from a recurrent neural network (RNN) of the VHRNN using an observation state, the latent variable and the initial hidden state; determining an approximate posterior probability distribution for the latent variable from an encoder network of the VHRNN using the observation state and the initial hidden state; determining a generating probability distribution for the observation state from a decoder network of the VHRNN using the latent variable and the initial hidden state; and maximizing a variational lower bound of a marginal log-likelihood of the training data. The trained VHRNN can be used to generate sequential data.

Abstract: A bearing device in a radial flow adsorber with a gas flow guide function includes a bearing platform, a supporting cylinder, a flow guide pipe, an annular bottom plate, a limiting block and a supporting frame, wherein an upper portion of the bearing platform is filled with adsorbent, an interior of a lower portion of the bearing platform is connected with the annular bottom plate through the supporting cylinder, a backing plate is arranged between the bearing platform and the supporting cylinder, on the supporting cylinder or the annular bottom plate is opened a hole and arranged a plurality of flow guide pipes for guiding gas flow, the annular bottom plate is placed on the supporting frame, the limiting block is arranged on the supporting frame, and the supporting frame is fixed on a cylinder body or a lower head of the adsorber by welding.

Abstract: Systems and methods for neural time series preprocessing and forecasting, dividing time series data to generate chunks of short time series, inputting each of the short time series to a data preprocessing neural network that includes differencing to transform non-stationary data to stationary data and to filter noise, generating and outputting, from the data preprocessing neural network, processed time series data, and inputting the processed time series data to a forecasting neural network. Parameters of the data preprocessing neural network and parameters of the forecasting neural network are learned end-to-end.

Abstract: Systems and methods for machine learning architecture for time series data prediction. The system may include a processor and a memory storing processor-executable instructions. The processor-executable instructions, when executed, may configure the processor to: obtain time series data associated with a data query; generate a predicted value based on a sampled realization of the time series data and a continuous time generative model, the continuous time generative model trained to define an invertible mapping to maximize a log-likelihood of a set of predicted values for a time range associated with the time series data; and generate a signal providing an indication of the predicted value associated with the data query.

Abstract: A variational hyper recurrent neural network (VHRNN) can be trained by, for each step in sequential training data: determining a prior probability distribution for a latent variable from a prior network of the VHRNN using an initial hidden state; determining a hidden state from a recurrent neural network (RNN) of the VHRNN using an observation state, the latent variable and the initial hidden state; determining an approximate posterior probability distribution for the latent variable from an encoder network of the VHRNN using the observation state and the initial hidden state; determining a generating probability distribution for the observation state from a decoder network of the VHRNN using the latent variable and the initial hidden state; and maximizing a variational lower bound of a marginal log-likelihood of the training data. The trained VHRNN can be used to generate sequential data.

Abstract: A constructing method for a concrete cylinder of a construction steel bar of a high-rise steel structure adopts a constructing method of supporting inside and climbing outside. The outside of the cylinder adopts hydraulic mutual-climbing adhesive lifting scaffold creeping formwork system (100), and also has an outside wallboard mechanical formwork erection function and an outer frame protection function. For the inside of the cylinder, except that an appropriate structure is left for later processing according to a schedule requirement, beam walls all adopt a steel cylinder support to perform formwork erection, so as to greatly reduce the workload of repeatedly mounting/dismantling the scaffold and the formwork at a high place. The scaffold itself carries the formwork erection system, so formwork erection and formwork removal can be mechanized, thereby reducing the risk of formwork erection and formwork removal at a high place.

Abstract: A constructing method for a concrete cylinder of a construction steel bar of a high-rise steel structure adopts a constructing method of supporting inside and climbing outside. The outside of the cylinder adopts hydraulic mutual-climbing adhesive lifting scaffold creeping formwork system (100), and also has an outside wallboard mechanical formwork erection function and an outer frame protection function. For the inside of the cylinder, except that an appropriate structure is left for later processing according to a schedule requirement, beam walls all adopt a steel cylinder support to perform formwork erection, so as to greatly reduce the workload of repeatedly mounting/dismantling the scaffold and the formwork at a high place.

Abstract: A leadframe and semiconductor device package with multiple semiconductor device die paddles for accepting multiple semiconductor devices is disclosed, wherein the leadframe increases semiconductor device density and reduces cost by integrating the multiple dies into a semiconductor device package with a relatively small footprint. The leadframe may include at least one full-metal die paddle and at least one reduced-metal die paddle, which may form a unified or hybrid die paddle. The leadframe may enable electrical coupling of multiple semiconductor devices to a common leadfinger and/or die paddle, where internal leadfingers coupled to the common leadfingers and/or die paddles may receive the electrical coupling means from the semiconductor device. Surfaces of one or more die paddles of the leadframe may be exposed to the outside of the semiconductor device package to enable electrical testing of and/or provide heat dissipation from one or more of the semiconductor devices attached to the leadframe.

Abstract: A method and packaging for semiconductor devices and integrated circuits is disclosed that eliminates warpage stress on packages caused by coefficient of thermal expansion (CTE) mismatch between the device, lead frame or die paddle and a molding compound. Generally, the method includes steps of: (i) mounting the die on which the device is fabricated to a die paddle of a leadframe; and (ii) encapsulating the die on the die paddle and at least a portion of the leadframe in a molding compound, wherein a difference between a first volume of molding compound above a plane of the leadframe and a second volume of molding compound below the plane of the leadframe is sufficiently reduced to substantially eliminate warpage of the finished package due to mismatch of CTEs of the device, lead frame and packaging compound. The die paddle may be etched or reduced to facilitate molding compound flowing under the plane of the leadframe. Other embodiments are also disclosed.

Abstract: An integrated circuit packaging device includes a laminate substrate. A first surface of the substrate can be mounted on an integrated circuit and the second surface can be mounted on a surface of a printed circuit board. The device can also include an array of lead contact pads on the first surface that can provide wire bond connections to circuit contact pads in the integrated circuit, and an array of solder ball contact pads on the second surface. Routing layers can provide electrical coupling between the lead contact pads on the first surface and the solder ball contact pads on the second surface. A dedicated contact pad on the first surface is electrically coupled to the laminate substrate.

Abstract: Embodiments of the present invention relate to a method and mechanism for testing wire bonds in an integrated circuit package. The method comprises bonding an integrated circuit silicon die to a package substrate. Next, wire connections are formed between pads in the integrated circuit silicon die and contact leads in the package substrate and testing each of the wire connections in order to detect non-stick failures using electrical continuity provided by the integrated circuit silicon die substrate. Electrical continuity is provided through dedicated pads in the package substrate that contact the underside of the silicon die substrate. The dedicated contact pads in each package substrate of the molded laminate array are connected to each other and to the mold gate. The continuity thus provided allows a non-stick-on-pad test by ensuring continuity between the wire spool through the die to the mold gate.

Abstract: An apparatus and method for providing delamination-resistant, array type molding of chip laminate packages such that larger chip array block sizes may be employed. An advanced mold die provides multiple wells for the formation of ejector pin tabs to be formed integrally to the mold cap of a chip laminate package. The die further provides for an ejector pin hole to be located at each ejector pin tab such that the ejector pins, when pressed for release of the laminate package from the mold die, bear against the integrally formed pin tabs rather than against the substrate of the chip/substrate assembly. The placement of the ejector pins for bearing against the pin tabs precludes the loading of the interface within the laminate package between the mold cap and the chip/substrate assembly. Substantially reduced delamination of the chip laminate package is achieved allowing for the use of larger chip array block sizes and providing for a substantial reduction in chip laminate package moisture sensitivity.