Abstract: Example solutions for zero-shot domain transfer with a text-to-text model train a text-to-text model for a target domain using unlabeled in-domain text training data, and concurrently train the model using labeled general-domain task training data. The in-domain training comprises masked language modeling (MLM) training, and the task training comprises both natural language generation (NLG) training and natural language understanding (NLU) training. The NLG training comprises natural language inference (NLI) training and the NLU training comprises summarization training. The trained model acquires domain-specific task competency, sufficient to perform a language task within the target domain. Suitable target domains include radiology, biomedical, and other medical, legal, and scientific domains.

Abstract: Embodiments of the invention provide a system including a first logic module for receiving a data stream that includes at least one neural network configured to generate at least one first password sample based at least in part on at least a portion of the data stream. A second logic module can be operatively coupled to the first logic module to receive the first password sample and at least one input dataset including a second password sample. The system can perform calculations to distinguish between at least one password of the first password sample and at least one password of the second password sample. Further, the system can iteratively learn and produce a feedback dataset based on the calculations, where the feedback dataset is configured to be provided to the first logic module.

Abstract: In an embodiment, a system for synchronizing the rotation of multiple mainframes of an airship includes multiple belt drive systems configured to mechanically rotate the mainframes, a central control system for sending a timing instruction to cause the mainframes to rotate synchronously about their respective rotational axis, wherein the mainframes are axis-aligned about their respective rotational axes and the timing instruction specifies a desired angular displacement of the mainframes, and multiple control units for controlling the belt drive systems to rotate the mainframes, respectively, wherein, for each mainframe, the associated control unit is configured to: receive the timing instruction from the central control system; determine, according to the timing instruction, a rotation instruction based on a size of the mainframe and the desired angular displacement; and instruct the belt drive system controlled by the control unit to rotate the mainframe based on the rotation instruction.

Abstract: In an embodiment, a system for synchronizing the rotation of multiple mainframes of an airship includes multiple belt drive systems configured to mechanically rotate the mainframes, a central control system for sending a timing instruction to cause the mainframes to rotate synchronously about their respective rotational axis, wherein the mainframes are axis-aligned about their respective rotational axes and the timing instruction specifies a desired angular displacement of the mainframes, and multiple control units for controlling the belt drive systems to rotate the mainframes, respectively, wherein, for each mainframe, the associated control unit is configured to: receive the timing instruction from the central control system; determine, according to the timing instruction, a rotation instruction based on a size of the mainframe and the desired angular displacement; and instruct the belt drive system controlled by the control unit to rotate the mainframe based on the rotation instruction.

Abstract: A method for reducing a volume of air within an electrical distribution system. The method includes providing at least one cable, surrounding the at least one cable with an isolating unit, placing a first shielding body on a top surface of the isolating unit, placing a second shielding body on a bottom surface of the isolating unit, wherein placing the first shielding body and the second shielding body creates a gap X1 between the first shielding body or the second shielding body and the isolating unit, and crimping the first shielding body to the second shielding body to create a shield housing, so that a gap X2 is created between the first shielding body or the second shielding body and the isolating unit, wherein X2 is less than X1. The at least a portion of the first shielding body may overlap with the second shielding body, and the first shielding body or the second shielding body may include at least one flared portion.

Abstract: This invention provides a process of obtaining a functional dermal substitute from decellularized amniotic membrane from placenta in combination with mammalian keratinocytes and its use as a tissue regeneration agent of skin, which the process for the generation of an artificial skin graft in the laboratory using specialized techniques in biotechnology such as the mammalian cells culture, preferably human, in which keratinocytes are cultured in aseptic conditions that form part of the epidermal layer of the skin, which are combined with a mesh rich in collagen which is obtained from cells that compose the structure of the skin in combination with the amniotic membrane of the placenta to generate two layers that can be used in burn patients or with skin lesions.

Abstract: Embodiments of the invention provide a system including a first logic module for receiving a data stream that includes at least one neural network configured to generate at least one first password sample based at least in part on at least a portion of the data stream. A second logic module can be operatively coupled to the first logic module to receive the first password sample and at least one input dataset including a second password sample. The system can perform calculations to distinguish between at least one password of the first password sample and at least one password of the second password sample. Further, the system can iteratively learn and produce a feedback dataset based on the calculations, where the feedback dataset is configured to be provided to the first logic module.

Abstract: A system and method for reducing air volume in a splitter. The system includes providing a isolator with a body closure of a splitter. The isolator my reduce the overall gas volume within the splitter. The method includes crimping a shield housing to reduce the amount of ambient space within the splitter.

Abstract: The method includes at least one processor estimating a first set of bias errors of a first set of base stations by obtaining a first set of time-of-arrival (ToA) measurements associated with the first set of base stations and a first set of devices, the first set of devices having known physical locations. The processor determines input parameters using the first set of bias errors, the input parameters including a first set of antenna locations for the first set of base stations, and a second set of ToA measurements associated with the first set of base stations and a first object, and determines a first object location for the first object using the input parameters.

Abstract: Various embodiments provide a method and apparatus for providing improved anchor-anchor clock synchronization. In particular, the skew and offset of a second clock in reference to a first clock is determined based on measured transmit and receive times of at least two two-way transmissions between a first anchor using the first clock and a second anchor using the second clock.

Abstract: When writing program code, syntax errors can be detected and a diagnostic error message can be provided. One of the causes of errors in a C-based programming language such as but not limited to C++ is a missing definition. The definition can exist in a file that is separate from the source program file. The file may be designated by a particular statement such as a “#include” statement which points to the file in which the token used in the code is defined. The list of files and their relationships form an include graph. A feature inside an IDE can suggest potential “#include” statements to correct a missing “#include” statement error. By using the include graph a more accurate list of potential corrections can be suggested so that standard and platform independent code can be written.

Abstract: Various embodiments provide a method and apparatus for providing improved anchor-anchor clock synchronization. In particular, the skew and offset of a second clock in reference to a first clock is determined based on measured transmit and receive times of at least two two-way transmissions between a first anchor using the first clock and a second anchor using the second clock.

Abstract: When writing program code, syntax errors can be detected and a diagnostic error message can be provided. One of the causes of errors in a C-based programming language such as but not limited to C++ is a missing definition. The definition can exist in a file that is separate from the source program file. The file may be designated by a particular statement such as a “#include” statement which points to the file in which the token used in the code is defined. The list of files and their relationships form an include graph. A feature inside an IDE can suggest potential “#include” statements to correct a missing “#include” statement error. By using the include graph a more accurate list of potential corrections can be suggested so that standard and platform independent code can be written.

Abstract: Various embodiments provide a method and apparatus for determining non-line of sight (NLOS) bias estimation based on estimated NLOS bias distributions at a number of anchors and the time of arrival (ToA) of a number of tag messages received at the anchors. In particular, using anchor redundancy, tag locations corresponding to the tag messages are estimated based on the NLOS bias distributions and the ToA and then the tag locations are used to update the NLOS bias distributions. The process of determining tag locations and updating the NLOS bias distributions is repeated until the NLOS bias distributions converge.

Abstract: Method and system for channel-invariant robust audio hashing is provided with a robust hash extraction step where a robust hash is extracted from audio content dividing the audio content in frames; applying a transformation procedure on the frames to compute, for each frame, transformed coefficients; applying a normalization procedure on the transformed coefficients to obtain normalized coefficients, where the normalization procedure computes the product of the sign of each coefficient of the transformed coefficients by an amplitude-scaling-invariant function of any combination of the transformed coefficients; applying a quantization procedure on the normalized coefficients to obtain the robust hash of the audio content; and a comparison step where the robust hash is compared with reference hashes to find a match.

Abstract: Disclosed embodiments include methods, apparatuses, and systems for secured image processing, image recognition, biometric recognition, and face recognition in untrusted environments. The disclosure includes a system for secure image recognition that comprises a secure biometric recognition system configured to work directly with encrypted signals, and the secure biometric recognition system comprises an input quantization system and a homomorphic encryption system configured for noninteractive biometric recognition.