Abstract: Semantic segmentation techniques and systems are described that overcome the challenges of limited availability of training data to describe the potentially millions of tags that may be used to describe semantic classes in digital images. In one example, the techniques are configured to train neural networks to leverage different types of training datasets using sequential neural networks and use of vector representations to represent the different semantic classes.

Abstract: A bionic nested structure fiber composite material includes a first fiber resin layer and a second fiber resin layer which are arranged in parallel, the first fiber resin layer and the second fiber resin layer are formed by a fiber bundle infiltrated with resin, and a bonded fiber unit is arranged between the first fiber resin layer and the second fiber resin layer, the bonded fiber unit are evenly distributed in a radial direction and a weft direction, the bonded fiber unit includes an inner core layer bonded fiber bundle, a middle core layer bonded fiber bundle and an outer core layer bonded fiber bundle, and the bonded fiber unit is performed 3D integrated layer-by-layer inner and outer nesting-and-weaving to form bionic nested structure.

Abstract: A method for recovering audio signals, a terminal and a storage medium are provided. The method includes: buffering an audio signal sampled at a preset number of sampling points each time, and then performing frequency spectrum analysis on the sampled audio signal by FFT; when it is determined that the audio signal is compressed, filtering a frequency point; recovering high-frequency signals based on audio signals before the frequency point; and performing phase recovery on the high-frequency signals. Thus, compressed high-frequency signals in the audio signals may be recovered.

Abstract: The present invention relates to an oxyhalide electrochemical cell comprising an anode of a Group IA metal and a cathode of a composite material prepared from a first electrochemically active carbonaceous material and a second electrochemically non-active carbonaceous material. The cathode material of the present invention provides increased discharge capacity compared to traditional lithium oxyhalide cells. In addition, the cathode material of the present invention is chemically stable which makes it particularly useful for applications that require increased rate capability in extreme environmental conditions such as those found in oil and gas exploration.

Abstract: The present disclosure provides a method for playing audio data, and belongs to the field of computer technology. The method includes: in a process of playing audio, partitioning left-channel audio data of the audio into audio data of a preset number of frequency sub-bands and partitioning right-channel audio data of the audio into audio data of the preset number of frequency sub-bands; calibrating the left-channel audio data of respective frequency sub-bands with left-channel calibration parameter values of respective frequency sub-bands to obtain left-channel audio data to be played; calibrating the right-channel audio data of respective frequency sub-bands with right-channel calibration parameter values of respective frequency sub-bands to obtain right-channel audio data to be played; and playing, over a left channel, the left-channel audio data to be played, and playing, over a right channel, the right-channel audio data to be played.

Abstract: A method includes: obtaining a transmission bandwidth granularity G1 of a target service and an optical-layer spectrum resource of a target fiber channel corresponding to the target service, where the optical-layer spectrum resource includes N consecutive frequency slots, and all of the N consecutive frequency slots have a same spectrum bandwidth; determining a target spectrum slice from the N frequency slots based on the transmission bandwidth granularity G1, where the target spectrum slice includes N1 consecutive frequency slots, and the target spectrum slice includes at least G1 consecutive idle frequency slots; and allocating the G1 consecutive idle frequency slots included in the target spectrum slice to the target service.

Abstract: This invention is directed to chimeric antigen receptors and cells comprising the same, wherein the cells further secrete monoclonal antibodies locally at a tumor site.

Abstract: A method performed by a control device includes: receiving forwarding information of many network devices in a network, where the many network devices in the network include a first edge device and a second edge device; determining, based on the forwarding information of the second edge device, information about a first network address that is reachable to the second edge device and that is outside the network; and estimating, based on the first network address information and the forwarding information of the plurality of network devices, whether one or more packets whose destination addresses are covered by the first network address information and that enter the network through the first edge device can be forwarded to the second edge device through the network.

Abstract: The present disclosure is related to a display apparatus. The display apparatus may include a display panel and a mirror switcher. The mirror switcher may be configured to reflect external incident light and block display light emitted by the display panel in a mirror mode of the display apparatus, and to transmit the display light emitted by the display panel in a display mode of the display apparatus.

Abstract: The present disclosure provides a method for playing audio data, and belongs to the field of computer technology. The method includes: in a process of playing audio, partitioning left-channel audio data of the audio into audio data of a preset number of frequency sub-bands and partitioning right-channel audio data of the audio into audio data of the preset number of frequency sub-bands; calibrating the left-channel audio data of respective frequency sub-bands with left-channel calibration parameter values of respective frequency sub-bands to obtain left-channel audio data to be played; calibrating the right-channel audio data of respective frequency sub-bands with right-channel calibration parameter values of respective frequency sub-bands to obtain right-channel audio data to be played; and playing, over a left channel, the left-channel audio data to be played, and playing, over a right channel, the right-channel audio data to be played.

Abstract: Disclosed are an optical module unlocking device and an optical module. For the optical module unlocking device, a boss is arranged on a metal front end of a pull ring. The boss abuts against the skewed slot of an upper cover. The metal front end located in the upper cover can slide relatively along the limiting post of the upper cover and at least one boss is driven to slide into the skewed slot of the upper cover and then hide into the bottom of the skewed slot or abut against a bevel edge of the at least one skewed slot and then bulge. After the at least one boss abuts against the bevel edge of the at least one skewed slot and then bulges, a shielding cage spring strip which is about to be cooperated with the boss is opened by being pushed upwards, and the optical module can be unlocked. The reliability of the optical module unlocking device is high and the unlocking motion of the optical module is simple.

Abstract: Provided are systems and techniques that provide an output phrase describing an image. An example method includes creating, with a convolutional neural network, feature maps describing image features in locations in the image. The method also includes providing a skeletal phrase for the image by processing the feature maps with a first long short-term memory (LSTM) neural network trained based on a first set of ground truth phrases which exclude attribute words. Then, attribute words are provided by processing the skeletal phrase and the feature maps with a second LSTM neural network trained based on a second set of ground truth phrases including words for attributes. Then, the method combines the skeletal phrase and the attribute words to form the output phrase.

Abstract: The present disclosure provides a method for preparing an alkali and co-producing gypsum, and belongs to the technical field of chemical production. The method comprises the steps of placing a cation exchange membrane into an electrolytic cell, adding a solution of sodium salt of a weak acid and a compound MH to an anode region as an anode electrocatalyst, adding sodium carbonate or sodium hydroxide to a cathode region, adding a compound M as a cathode electrocatalyst, and applying a DC power supply between a cathode electrode and an anode electrode.

Abstract: A novel wound electrode assembly for a lithium oxyhalide electrochemical cell is described. The electrode assembly comprises an elongate cathode of an electrochemically non-active but electrically conductive carbonaceous material disposed between an inner elongate portion and an outer elongate portion of a unitary lithium anode. That way, lithium faces the entire length of the opposed major sides of the cathode. This inner anode portion/cathode/outer anode portion configuration is rolled into a wound-shaped electrode assembly that is housed inside a cylindrically-shaped casing. A cylindrically-shaped sheet-type spring centered in the electrode assembly presses outwardly to limit axial movement of the electrode assembly. In one embodiment, all the non-active components, except for the cathode current collector which is nickel, are made of stainless-steel. This provides the cell with a low magnetic signature without adversely affecting the cell's high-rate capability.

Abstract: Disclosed are an optical module unlocking device and an optical module. For the optical module unlocking device, a boss is arranged on a metal front end of a pull ring. The boss abuts against the skewed slot of an upper cover. The metal front end located in the upper cover can slide relatively along the limiting post of the upper cover and at least one boss is driven to slide into the skewed slot of the upper cover and then hide into the bottom of the skewed slot or abut against a bevel edge of the at least one skewed slot and then bulge. After the at least one boss abuts against the bevel edge of the at least one skewed slot and then bulges, a shielding cage spring strip which is about to be cooperated with the boss is opened by being pushed upwards, and the optical module can be unlocked. The reliability of the optical module unlocking device is high and the unlocking motion of the optical module is simple.

Abstract: The present disclosure, belonging to the field of audio technology, provides a method and apparatus for recovering audio signals. The method includes: buffering an audio signal sampled at a preset number of sampling points each time, and then performing frequency spectrum analysis on the sampled audio signal by FFT; when it is determined that the audio signal is compressed, filtering a frequency point; recovering high-frequency signals based on audio signals before the frequency point; and performing phase recovery on the high-frequency signals. The present disclosure provides the method for recovering audio signals.

Abstract: The present invention relates to an oxyhalide electrochemical cell comprising an anode of a Group IA metal and a cathode of a composite material prepared from a first electrochemically active carbonaceous material and a second electrochemically non-active carbonaceous material. The cathode material of the present invention provides increased discharge capacity compared to traditional lithium oxyhalide cells. In addition, the cathode material of the present invention is chemically stable which makes it particularly useful for applications that require increased rate capability in extreme environmental conditions such as those found in oil and gas exploration.

Abstract: A data encryption method performed at a computing device includes: receiving a data encryption request, the data encryption request indicating original data that needs to be encrypted and at least two target storage devices that are communicatively connected to the computing device; in response to the data encryption request: separately obtaining unique device information of the at least two target storage devices; generating, based on the unique device information, a public key according to a preset policy; encrypting the original data by using the public key to obtain ciphertext; and destructing relevant data of the public key from the computing device, and storing the ciphertext into the at least two target storage devices.

Abstract: In embodiments of event image curation, a computing device includes memory that stores a collection of digital images associated with a type of event, such as a digital photo album of digital photos associated with the event, or a video of image frames and the video is associated with the event. A curation application implements a convolutional neural network, which receives the digital images and a designation of the type of event. The convolutional neural network can then determine an importance rating of each digital image within the collection of the digital images based on the type of the event. The importance rating of a digital image is representative of an importance of the digital image to a person in context of the type of the event. The convolutional neural network generates an output of representative digital images from the collection based on the importance rating of each digital image.