Abstract: A data processing device is provided, which includes a storage device and a data quality assessment measurement calculation module. The storage device is configured to obtain change data of data tables from a data center. The data quality assessment measurement calculation module is configured to calculate an updated data quality assessment measurement value and an updated data quality characteristic data according to the change data and a data quality assessment measurement reference value and perform a notification function according to the updated data quality assessment measurement value and a data quality threshold value. The data quality assessment measurement calculation module is configured to compare the updated data quality assessment measurement value with the data quality threshold value to generate a comparison result.

Abstract: A data mining method includes extracting a main character column of a data table of a data sheet; determining an evaluation measurement value of the data table according to a variance and an outlier ratio of the main character column of the data table; and determining a data mining value of the data table according to the evaluation measurement value of the data table.

Abstract: A search system and method for endoscopic images is proposed. In this method, a feature extraction model generates a target feature value, along with multiple first and second feature values based on a target image and multiple first and second source images from different areas of a human organ. Next, the similarity between each first and second feature value and the target feature value is calculated, followed by the computation of first and second reference values based on these similarities. If the first reference value is greater than the second, the source image with the highest similarity among the first values is selected as the search result; otherwise, the source image with the highest similarity among the second values is chosen as the result.

Abstract: A reinforcement learning-based agent policy generation method and a non-transitory computer-readable media are proposed. The method includes: obtaining a first state, an action network and a value network of an agent, and a reward function of an environment in which the agent is located; generating a first action for the agent to execute according to the action network and the first state, and generating a first value according to the value network and the first state; obtaining a second state of the agent generated by the environment and a reward generated by the reward function; storing the first state, the first action, the first value, the second state, and the reward into a buffer; and training the value network and the action network according to the buffer; wherein a loss function of the action network includes a policy gradient loss and a regularization loss.

Abstract: An automatic selection method and system for endoscopic images are proposed. This method includes several steps performed by a computing device, which involve: obtaining multiple endoscopic images and their corresponding position markers, with each marker indicating a specific part of a human organ captured by the endoscope. Based on these position markers, multiple candidate images belonging to the same organ part are selected from the endoscopic images. An image segmentation is performed based on an image color or a predefined symptom to divide each candidate image into a first and a second non-overlapping region. A ratio is calculated based on the areas of the first and second regions, and at least one candidate image with a ratio greater than a threshold is outputted as the selection result.

Abstract: A method and system for reducing hallucinations generated by a Large Vision-Language Model (LVLM) are provided. The method includes a plurality of steps performed by a computing device, and these steps include: obtaining a test image, inputting the test image and a prompt into the LVLM to generate a test embedding, where the prompt instructs the LVLM to describe the test image, identifying a candidate embedding closest to the test embedding among a plurality of reference embeddings, replacing data of the test embedding in a salient dimension with data of the candidate embedding in the salient dimension, and generating a test result by the LVLM according to the test embedding with replaced data.

Abstract: The present disclosure provides a method including: generating a first private key and a public key according to a parameter set of full homomorphic encryption; encrypting test data and label by the public key to generate test data ciphertext and label ciphertext; generating a smart contract executed by a blockchain system, and transferring control of an amount of cryptocurrency from a first cryptocurrency account to the blockchain; receiving a result of a verification to a model ciphertext; when the result indicates that the model ciphertext does not pass the verification, retrieving the control of the amount of cryptocurrency; and when the result indicates that the model ciphertext passes the verification, receiving the model ciphertext and a second private key from the blockchain system, and decrypting, according to the first and second private keys, the model ciphertext to generate a model to infer the test data.

Abstract: An inference method for encrypted deep neural network model is executed by a computing device and includes: encoding a message according to a quantization parameter to generate a plaintext, encrypting the plaintext according to a private key to generate a ciphertext, sending the ciphertext to a deep neural network model to generate a ciphertext result, decrypting the ciphertext result according to the private key to generate a plaintext result, and decoding the plaintext result according to the quantization parameter to generate an inference result.

Abstract: A method of preparing an ionic liquid includes the following steps. A halogen-containing compound is reacted with a first compound to form a second compound. The halogen-containing compound includes a halohydrocarbon, a sulfonyl halide, or a combination thereof. The first compound includes an amine compound having a tertiary amine group, a phosphine compound, or a combination thereof. The second compound includes a first quaternary ammonium salt, a first quaternary phosphonium salt, or a combination thereof. The second compound and a lithium salt are reacted in a microwave device to form a third compound. The third compound includes a second quaternary ammonium salt, a second quaternary phosphonium salt, or a combination thereof. Anions of the second compound and the third compound are different. A microwave power of the microwave device is 700 watts to 1400 watts.

Abstract: A method for preparing an organic metal skeleton material embedded with nanometal, comprising: S1: making a precursor comprising a mixed solution of a metal salt and an organic ligand; S2: providing an adjusting solution, mixing and causing reactions the adjusting solution with the precursor so that the adjusting solution makes the precursor mixed solution have a pH value ranged from 3 to 7, thereby forming an organic metal skeleton material; and S3: dispersing the organic metal skeleton material in a hydrophobic solvent, then adding a metal salt solution, drying the organic metal skeleton material and the metal salt after adsorption, and causing a metal reduction reaction in a gas environment to form an internal organic metal skeleton material embedded with nanometals.

Abstract: An intermediate policy training method for agent is provided. This method includes: selecting a source and a target policy from a plurality of policies. Each policy is configured to drive an agent to perform a plurality of actions to be in a plurality of states. Each state includes a plurality of physical properties. The method further includes: respectively selecting one from the source policy and the target policy as a source state and a target state; training an intermediate policy by a reinforcement learning to transition the agent from the source state to the target state over an episode, where the reinforcement learning includes an annealing function for setting a plurality of tolerance boundaries of the plurality of physical properties, and the tolerance boundaries gradually shrink during the episode.

Abstract: A panoramic camera configuration method for configuring a vehicle with a plurality of cameras includes obtaining a plurality of configuration parameters of the vehicle and a plurality of camera parameters of the plurality of cameras; utilizing a three-dimensional simulation software to establish a virtual environment image according to a plurality of environmental parameters of the environment around the vehicle; and obtaining a plurality of environment images corresponding to the plurality of cameras according to the virtual environment image, the plurality of environmental parameters, the plurality of configuration parameters and the plurality of camera parameters; obtaining a panoramic image according to the plurality of environmental images; and updating the plurality of configuration parameters and the plurality of camera parameters according to the virtual environment image and the panoramic image.

Abstract: A system for detecting an event occurrence period of cyclical event includes an image monitoring device and a period judgment device. The image monitoring device captures monitoring images respective in image capturing times. The period judgment device extracts feature vectors from the monitoring images to calculate vector values to accordingly generate a vector value time domain signal, executes a short-time Fourier transform (STFT) with a window width time to transform the vector value time domain signal to a spectrogram, makes the spectrogram be denoised to generate a representative frequency time domain signal, and calculates an event-cycle achievement rate and a product term accumulation number after each window width time, so as to accordingly calculate event occurrence time and the event occurrence period.

Abstract: A training method for continual learning model and a non-transitory computer-readable medium are proposed. The method includes: training the encoder and self-attention layer in the essence generation procedure according to the raw data of a task when the current training process is the first task in continual learning; otherwise, freezing the parameters of the encoder and self-attention layer, performing the essence generation procedure to convert the raw data into a data essence, and adding the data essence into the essence memory. The training process is repeated until the continual learning model converges. The training process includes: obtaining a training batch from the raw data, updating the replay memory according to the training batch, training the continual learning model according to the replay memory and the essence memory, and updating the data essence in the essence memory when the current training process is the first task.

Abstract: An augmented reality collaboration system includes an assembly platform, a first projector, a camera and a computing circuit. The first projector projects a first projection image to the assembly platform. The camera shoots the assembly platform and the first projection image projected on the assembly platform to obtain a first shooting image. The computing circuit creates a three-dimensional model in a virtual three-dimensional space, arranges a standard operating procedure, sets a mark pattern on the three-dimensional model according to the standard operating procedure, analyzes the first shooting picture, and performs a calibration operation on the first projector to obtain a first calibration parameter. The computing circuit enables the first projector to generate a first corrected pattern according to the first calibration parameter and the mark pattern, and projects the first corrected pattern to the object-to-be-assembled according to the standard operating procedure.

Abstract: A data pagination automation method, for processing a plurality of data items includes determining a sorting basis for sorting the plurality of data items according to an ordered unique key of the plurality of data items; sorting and paginating the plurality of data items according to the sorting basis; and updating a pagination information table of the plurality of data items and collecting corresponding data according to the ordered unique key and at least a filtering condition; wherein the pagination information table is updated with to a partial dynamic update method.

Abstract: A generative artificial intelligence (AI) information security management method, comprising following steps: receiving a question to a local generative AI model; providing a security risk analysis information of the question, wherein the security risk analysis information includes a Boolean value of an information security concern; providing a response to the question through the local generative AI model; after the local generative AI model provides the response to the question, in response to receiving a request for answering with an external generative AI model, determining whether the Boolean value of the information security concern is true; and in response to the Boolean value of the information security concern being true, providing a message that the question has the information security concern.

Abstract: A data processing device is provided, which includes a storage device and a data quality assessment measurement calculation module. The storage device is configured to obtain change data of data tables from a data center. The data quality assessment measurement calculation module is configured to calculate an updated data quality assessment measurement value and an updated data quality characteristic data according to the change data and a data quality assessment measurement reference value and perform a notification function according to the updated data quality assessment measurement value and a data quality threshold value. The data quality assessment measurement calculation module is configured to compare the updated data quality assessment measurement value with the data quality threshold value to generate a comparison result.

Abstract: The accelerated bootstrapping fully homomorphic encryption calculator includes a memory, a controller, a task scheduler, a plurality of processing units, and a bus. The memory stores ciphertext structured as learning with errors over rings. The controller manages storing the ciphertext and generates instructions for bootstrapping. The task scheduler organizes tasks based on the controller's instructions, and the processing units decompose the ciphertext and perform computations to generate intermediate or final results. The bus connects all components and handles the transmission of ciphertext, instructions, and results.

Abstract: This anomaly detection method, based on out-of-distribution techniques, is executed by a computing device. It starts by obtaining a training dataset containing various images, including a first image and multiple second images. The method segments objects and contexts in each image, calculating the similarity between the object in the first image and those in the second images. A candidate image is selected if its similarity exceeds a predefined threshold. The object from the first image is blended with the context of the candidate image to produce a blended image. A detection model is then trained using this dataset. Subsequently, in-distribution embeddings are generated, and a test embedding is created. The test sample is classified as an anomaly when the minimum distance between the in-distribution embeddings and the test embedding exceeds a default value.