Abstract: A computing system is provided for generating a synthesized image using subject-driven text-to-image generation. The computing system includes a processing circuitry and memory storing instructions that, when executed, cause the processing circuitry to receive one or more input images of a subject and a user prompt, generate a mosaic of two or more component images based on the one or more input images and the user prompt, generate a mask to mark a placeholder area in the mosaic to be inpainted, use a generative model to inpaint the placeholder area, extract the inpainted placeholder area to generate an extracted image of the subject, and output the extracted image of the subject as the synthesized image.

Abstract: A method, an apparatus, an electronic device and a storage medium for generating media content are provided. The method comprises: obtaining a prompt; and processing the prompt using a media generation model, to generate media content associated with a first effect and a second effect, where the media generation model is constructedbased on: processing a training prompt using a first model associated with the first effect, to generate a first feature representation; processing the training prompt using a second model associated with the second effect, to generate a second feature representation; processing the first feature representation and the second feature representation using a discriminator to determine an adversarial loss; and adjusting, based on the adversarial loss, parameters of the first model to construct a media generation model.

Abstract: A computing system including one or more processing devices configured to receive a weight tensor of a neural network. The one or more processing devices are further configured to execute a gradient descent optimizer that updates the weight tensor over a plurality of projection matrix update intervals. Each of the projection matrix update intervals includes computing a gradient over the weight tensor in each of a plurality of gradient descent iterations. Each of the gradient descent iterations further includes projecting the gradient into a reduced-rank subspace using a projection matrix and updating the weight tensor by performing gradient descent using the projected gradient. Each of the projection matrix update intervals further includes computing a projection matrix error value associated with the projection matrix and updating the projection matrix based at least in part on the projection matrix error value.

Abstract: A computing system receives an input prompt and input images, generates identification images based on the input images, and generates identification patches based on the identification images, respectively. The system further generates a pose-patch image based on the identification patches and a pose image, and generates word tokens based on the identification images, respectively. Token embeddings are generated based on the input prompt, and the word tokens and the token embeddings are concatenated to generate concatenated token embeddings. The system inputs the pose-patch image and the concatenated token embeddings into a control network to generate features. Then the features, latent noise, and the concatenated token embeddings are inputted into a diffusion model to generate the synthesized image, and an output is generated based on the synthesized image.

Abstract: The present disclosure describes techniques for fast personalized image generation using a machine learning model. Contextualized features are generated by a first sub-model of the machine learning model. The contextualized features are generated based on an object image and a text prompt. The object image comprises a target object. The text prompt depicts a target image and indicates at least one of changing textures of the target object or placing the target object in a particular context. Self-attention features are generated from the object image. The self-attention features contain detailed information about the target object. A target image is generated based on the contextualized features and the self-attention features by a second sub-model of the machine learning model. The second sub-model comprises context cross attention layers and object cross attention layers. The target image preserves an identity of the target object.

Abstract: The present disclosure provides systems and methods for generating a synthesized image of a user with a trained machine learning diffusion model. In one example, a computing system includes one or more processors configured to execute instructions stored in memory to execute a trained machine learning diffusion model including an image encoder, a text encoder, and a diffusion model. The image encoder is configured to receive an image of a user and generate a set of embeddings that semantically describe visual features of the user based at least on the image of the user. The text encoder is configured to receive the set of embeddings and generate an input feature vector based at least on the set of embeddings. The diffusion model is configured to receive the input feature vector and generate a synthesized image of the user based at least on the input feature vector.

Abstract: Systems and methods are provided that include a processor executing a program to process an initial image through a first diffusion stage to generate a final first stage image, wherein the first diffusion stage includes using a diffusion model, a gradient estimator model smaller than the diffusion model, and a text-image match gradient calculator. The processor further executes the program to process the final first stage image through a second diffusion stage to generate a final second stage image. The second diffusion stage includes, for a second predetermined number of iterations, inputting the final first stage image to through the diffusion model, back-propagate the image through the text-image match gradient calculator to calculate a second stage gradient against the input text, and update the final first stage image by applying the second stage gradient to the final first stage image.

Abstract: The present disclosure provides systems and methods for generating a synthesized image of a user with a trained machine learning diffusion model. In one example, a computing system includes one or more processors configured to execute instructions stored in memory to execute a trained machine learning diffusion model including an image encoder, a text encoder, and a diffusion model. The image encoder is configured to receive an image of a user and generate a set of embeddings that semantically describe visual features of the user based at least on the image of the user. The text encoder is configured to receive the set of embeddings and generate an input feature vector based at least on the set of embeddings. The diffusion model is configured to receive the input feature vector and generate a synthesized image of the user based at least on the input feature vector.

Abstract: Systems and methods are provided that include a processor executing a program to process an initial image through a first diffusion stage to generate a final first stage image, wherein the first diffusion stage includes using a diffusion model, a gradient estimator model smaller than the diffusion model, and a text-image match gradient calculator. The processor further executes the program to process the final first stage image through a second diffusion stage to generate a final second stage image. The second diffusion stage includes, for a second predetermined number of iterations, inputting the final first stage image to through the diffusion model, back-propagate the image through the text-image match gradient calculator to calculate a second stage gradient against the input text, and update the final first stage image by applying the second stage gradient to the final first stage image.

Abstract: A computer-implemented method of training an autoencoder to recover missing data is provided. The autoencoder includes an encoder for encoding its inputs into a latent space and a decoder for decoding the encodings from the latent space. The method comprises creating a first training set including a valid data set of multiple dimensions, and training the encoder and the decoder in a first training stage using the first training set to reduce a difference between the valid data set provided to the encoder and a data set decoded by the decoder. The method further comprises creating a second training set comprising an invalid data set, and training the encoder in a second training stage using the second training set to reduce a difference between encodings of valid data instances and encodings of their corresponding invalid data instances.

Abstract: Embodiments of the disclosure include methods and compositions for the treatment of neovascularization- and ischemic retinopathy-related disorders. In some embodiments, a composition comprising an effective amount of an apoA-I binding protein or its agonist in combination with anti-VEGF reagents is administered to an individual in need thereof to treat, prevent, reverse, and/or meliorate conditions associated with macular degeneration or cancer. In some embodiments, a composition comprising an effective mount of an AIBP-inhibitor is administered to an individual in need thereof to stimulate revascularization in the eye to treat, prevent, reverse, and/or ameliorate conditions associated with ischemic retinopathies.

Abstract: A system may be disclosed in the present disclosure, comprising: an interface service unit configured to perform at least one of sending data to or receiving data from one or more users via a user interface; and a transmission unit including: a blockchain adaptor configured to transmit data to a plurality of different types of blockchains, via a blockchain interface, according to data format and communication mode requirements of each of the plurality of different types of blockchains and a transmission controller configured to trigger a smart contract running on a blockchain of the plurality different types of blockchains via the blockchain interface, and control a data transmission under instructions from the smart contract, wherein the data transmission is based on data received from the user interface.

Abstract: A data transmission platform for data transmission using a blockchain is configured to receives ciphertext data from a data provider, wherein the ciphertext data include transmitted data authenticated by the data transmission platform and encrypted by the data provider; identifies a target data requestor; creates a smart contract according to a data transmission requirement of the target data requestor; uploads the smart contract to a blockchain; sends a data transmission request to the data provider; re-encrypts the ciphertext data to obtain re-encrypted data using a re-encryption key generated by the data provider; triggers the smart contract to send the data transmission demand of the data requestor to the data provider; and notifies the data requester to extract the decrypted data and perform decryption to obtain the transmitted data.

Abstract: Provided is a composition comprising the following components: a) bendamustine or a salt thereof, b) a cyclodextrin, and c) an optional stabilizer, wherein the stabilizer is selected from an inorganic or organic substance containing chlorine element, lipoic acid, thioglycerol, polyvinylpyrrolidone and any combination thereof, and wherein the mass ratio of the bendamustine or the salt thereof to the cyclodextrin is 1:5 to 1:100. Further provided is a method for preventing or treating a cancer using the composition and the use of the composition in the preparation a medicament for treating a cancer.

Abstract: A computer-implemented method of training an autoencoder to recover missing data is provided. The autoencoder includes an encoder for encoding its inputs into a latent space and a decoder for decoding the encodings from the latent space. The method comprises creating a first training set including a valid data set of multiple dimensions, and training the encoder and the decoder in a first training stage using the first training set to reduce a difference between the valid data set provided to the encoder and a data set decoded by the decoder. The method further comprises creating a second training set comprising an invalid data set, and training the encoder in a second training stage using the second training set to reduce a difference between encodings of valid data instances and encodings of their corresponding invalid data instances.

Abstract: A data transmission platform for data transmission using a blockchain is configured to receives ciphertext data from a data provider, wherein the ciphertext data include transmitted data authenticated by the data transmission platform and encrypted by the data provider; identifies a target data requestor; creates a smart contract according to a data transmission requirement of the target data requestor; uploads the smart contract to a blockchain; sends a data transmission request to the data provider; re-encrypts the ciphertext data to obtain re-encrypted data using a re-encryption key generated by the data provider; triggers the smart contract to send the data transmission demand of the data requestor to the data provider; and notifies the data requester to extract the decrypted data and perform decryption to obtain the transmitted data.

Abstract: A cloud water resource detecting system and method, the system includes computing device, and Raman lidar, microwave radiometer, millimeter wave cloud radar, wind profile radar and Doppler wind lidar in communication connection with the computing device. The Raman lidar and the microwave radiometer are configured to obtain temperature and humidity data of a cloud system. The millimeter wave cloud radar and the microwave radiometer are configured to obtain cloud water content profiles. The wind profile radar is configured to obtain a horizontal wind speed of the cloud system. The Doppler wind lidar is configured to obtain the vertical motion speed of a cloud bottom. The computing device is configured to calculate the total amount of cloud water according to the temperature and humidity data of the cloud system, the cloud water content profiles, the horizontal wind speed of the cloud system and the vertical motion speed of the cloud bottom.

Abstract: A system may be disclosed in the present disclosure, comprising: an interface service unit configured to perform at least one of sending data to or receiving data from one or more users via a user interface; and a transmission unit including: a blockchain adaptor configured to transmit data to a plurality of different types of blockchains, via a blockchain interface, according to data format and communication mode requirements of each of the plurality of different types of blockchains and a transmission controller configured to trigger a smart contract running on a blockchain of the plurality different types of blockchains via the blockchain interface, and control a data transmission under instructions from the smart contract, wherein the data transmission is based on data received from the user interface.