Abstract: The present disclosure describes systems and methods for implementing input-adaptive neural networks. Several approaches and systems are presented for creating flexible and computationally-efficient deep neural networks that adapt their complexity based upon input data such as video streams. The approach may use fine-grained design and an automatic approach for generating optimal network architectures based on early-exit mechanisms. A neural network modified according to such principles can be implemented in various devices including mobile vision systems such as smartphones, drones, and head-mounted augmented reality devices.

Abstract: The present invention provides processes of producing the following compound: and methods of using the compound in targeted PET and SPECT imaging.

Abstract: Embodiments of a process are provided. In the process, a rare earth (RE) element is added to an aqueous feedstock including a first salt of the RE element and a second salt of a metal to form a mixture. The mixture is heated to a temperature of at least 50° C. to cause hydrolysis of the second salt to form a precipitate of a hydroxide comprising the metal and to cause an increase in a concentration of the first salt or formation of a third salt comprising the RE element. The precipitate is separated from the mixture, and the first salt or the third salt in the mixture is reacted with an oxalate to precipitate a RE oxalate. The RE oxalate is separated from the mixture, and the RE oxalate is calcined to form the RE element.

Abstract: An anchorage system for prestressing a non-metallic (e.g., FRP) tendon against a support member may include a dead-end assembly and a live-end assembly configured to be placed on opposite sides of the support member, each assembly having a support sleeve affixed to a support plate. The anchorage system may also include a tensioning rod coupled to an end of the support sleeve of the live-end assembly that is remote from the respective support plate, and a removable support box extending around the support sleeve of the live-end assembly, the tensioning rod extending through an opening defined in the removable support box, the support box configured to be an abutment structure against which a jack may be placed. The anchorage system may also include one or more slotted shim plates interposed between the support plate of the live-end assembly and a first confronting surface of the support member.

Abstract: An apparatus is described for robotic control of an ultrasound imaging device. The apparatus comprises a fixed base and a movable platform with an end-effector. A plurality of passive soft links connects the platform to the base. A series of cables connect the platform to a set of motors adjacent the base. When the platform is at rest, the tension in the cables counter-balances the passive soft links pushing the platform away from the base plate. Control electronics move the motors so as to control the location and the orientation of the platform. By increasing or decreasing tension in the cables, the motors smoothly control the location and the orientation of the platform and thus the end effector. Such a robotic apparatus can be used to provide expert handling of an ultrasound imaging device remotely.

Abstract: Provided are ceramic foams. The ceramic foams may have a hierarchical pore gradient. The ceramic foams may be silica aerogels. The ceramic foams may be made by reaction of one or more precursors in the presence of an inert gas generated by a pore-forming gas-forming additive. The ceramic foams may be used as insulating materials.

Abstract: Provided herein are compositions and methods for treating Pitt Hopkins Syndrome (PTHS) by restoring PTHS patient's gut microbiota. These methods can be used with PTHS patient with or without ongoing gastrointestinal symptoms. The methods comprise administering a therapeutic composition comprising a fecal microbe or a fecal microbe preparation to the subject.

Abstract: The invention is directed generally to oligonucleotide compositions for the treatment of DNA repeat expansion diseases. The invention also relates to oligonucleotides directed to subunits of the DNA mismatch repair system.

Abstract: A computer-implemented system implements a named entity recognition (NER) model trained for automatic dataset labeling and corpus generation for cybersecurity entities. The NER model includes a semantic similarity measure to determine which category an unclassified/unlabeled word such as an ambiguous keyword with more than one meaning should belong to based on the semantic similarity of an entire sentence.

Abstract: Described herein are means for implementing annotation-efficient deep learning models utilizing sparsely-annotated or annotation-free training, in which trained models are then utilized for the processing of medical imaging. An exemplary system includes at least a processor and a memory to execute instructions for learning anatomical embeddings by forcing embeddings learned from multiple modalities; initiating a training sequence of an AI model by learning dense anatomical embeddings from unlabeled data, then deriving application-specific models to diagnose diseases with a small number of examples; executing collaborative learning to generate pretrained multimodal models; training the AI model using zero-shot or few-shot learning; embedding physiological and anatomical knowledge; embedding known physical principles refining the AI model; and outputting a trained AI model for use in diagnosing diseases and abnormal conditions in medical imaging. Other related embodiments are disclosed.

Abstract: Modified ribosomes that were selected using a dipeptidyl-puromycin aminonucleoside are used to mediate site-specific incorporation of one or more peptides and peptidomimetics into protein in a cell free translation system. In addition, new fluorescent dipeptidomimetics have been synthesized and incorporated into proteins, as well as modified proteins containing one or more non-naturally occurring dipeptides.

Abstract: Systems for piezoelectric composite materials are described. Various piezoelectric composite material structures and configurations can be incorporated into piezoelectric panels. Various methods are described to recycle and heal various piezoelectric composite materials.

Abstract: Systems and methods for training physics-informed neural network (PINN) surrogate models to model physical problems are provided. The method may comprise coupling, using a processor, one or more convolutional neural networks (CNNs) with a finite element method (FEM). The coupling may comprise calculating, for a finite element mesh comprising a plurality of finite elements, an internal force vector, P, a force vector, F, and a tangent stiffness matrix, KT, using the FEM. Each finite element may comprise one or more finite element nodes. The coupling may further comprise applying a CNN to the finite element mesh to obtain a solution to a physical problem. The CNN may be trained on a loss function, and the loss function may incorporate the internal force vector, P, and the force vector, F.

Abstract: In examples, provided are leadless power couplers that include (1) a thermal insulating system having an outer wall and an inner wall, (2) a first electrically conductive winding located outside the thermal insulating system, where the first electrically conductive winding is configured to create a varying magnetic field, (3) a plurality of second electrically conductive windings located inside the thermal insulating system and configured to couple to the varying magnetic field, the plurality of second electrically conductive windings being superconductors, (4) a plurality of cryogenic rectifiers, each cryogenic rectifier being coupled to a respective second electrically conductive winding in the plurality of second electrically conductive windings, and (5) a plurality of cryogenic cables coupled between respective outputs of the plurality of cryogenic rectifiers and respective loads.

Abstract: An apparatus and system are described to provide an in-memory computing non-volatile flash memory cell array used in a neural network. Each cell includes a Resistive RAM memory (RRAM) and a physical resistor formed from a high resistive material. The RRAM is programmed to either an on or off state in which the resistance is respectively significantly less or more than the resistor to permit the RRAM to act as a switch and allow for in-situ training. Multi-bit RRAM cells contain multiple RRAMs, each of which is connected to a resistor having a different resistance and read using the same input line. The resistors are formed from the same material as the resistor in the analog-to-digital converter used to read the array.

Abstract: A cancer vaccine technology is provided which knocks out expression of cell surface immune checkpoint proteins, to facilitate their processing by immune cells, and optionally by knocking-in the expression of cytokines to boost immune response. Non-replicating tumor cells lacking cell surface CD47 are highly effective immunizing agents against subcutaneous mouse melanoma. Whole-cell vaccines inhibited tumor growth, and immunophenotyping showed a dramatic increase in activated effector cell subsets and M1-type macrophages aided by a significant reduction in the tumor-associated macrophage and myeloid derived suppressor cell compartments. A remarkable downregulation of cell surface CD47 was observed in the tumors that did escape after vaccination with genetically modified cells, suggesting the intricate involvement of CD47 in a prophylactic situation. An effective vaccination strategy to increase tumor-specific immune response in solid tumors is provided to improve the outcome of cancer immunotherapy.

Abstract: A laser system and method include a self-referencing shaper. A self-referencing pulse shaper is provided in an embodiment. Another aspect of a laser system includes at least one beam splitter splitting a reference beam from a working beam and a test beam, a delay optic delaying a reference laser beam, an active shaper, an interferometer, and a programmable controller. In another aspect, a method includes splitting an input laser pulse into a reference pulse and a shaping pulse, controlling phase and amplitude of the shaping pulse with an adjustable pulse shaper, creating an optical delay of the reference pulse, comparing a test pulse and the reference pulse after the controlling and delay, the laser system characterizing the input laser pulse and monitoring the laser system's own dispersion in a self-referenced manner, and correcting an output working laser pulse by adjusting the pulse shaper based on the comparing step.

Abstract: A recombinant strain of F. diplosiphon was made by transforming wild type F. diplosiphon with a pGEM-7Zf (+) plasmid containing the photolyase gene (phrA) via electroporation. The recombinant strain was designated B481-ViAnSa and overexpressed the phrA gene to result in enhanced UV tolerance compared to wild type F. diplosiphon.

Abstract: Provided herein are photochemical separations. The methods herein can include exposing a first metal complex and a second metal complex to light to facilitate an irreversible chemical reaction to form a modified first metal complex. The modified first metal complex then may be separated from the second metal complex. Compositions also are provided.

Abstract: An unmanned aerial vehicle includes a plurality of arms coupled along a body, each arm of the plurality of arms including: a linkage structure configurable in a first linkage state, a second linkage state, and a third linkage state. The vehicle can further include an actuation mechanism including: a ratchet gear having a contact surface that applies a lateral force against a leading link of the linkage structure to transition the linkage structure from the first linkage state to the second linkage state or the third linkage state; a bistable spring that transitions from a first spring state to a second spring state upon application of the lateral force against the leading link of the linkage structure; and a cable for straightening the linkage structure.