Abstract: A method includes obtaining, using at least one processing device, noisy speech signals and extracting, using the at least one processing device, acoustic features from the noisy speech signals. The method also includes receiving, using the at least one processing device, a predicted speech mask from a speech mask prediction model based on a first acoustic feature subset and receiving, using the at least one processing device, a predicted noise mask from a noise mask prediction model based on a second acoustic feature subset. The method further includes providing, using the at least one processing device, predicted speech features determined using the predicted speech mask and predicted noise features determined using the predicted noise mask to a filtering mask prediction model. In addition, the method includes generating, using the at least one processing device, a clean speech signal using a predicted filtering mask output by the filtering mask prediction model.

Abstract: A method for generating a customized speech enhancement model includes obtaining noisy-clean speech data from a source domain, obtaining noisy speech data from a target domain; obtaining raw speech data, using the noisy-clean speech data, the noisy speech data, and the raw speech data, training the customized SE model based on at least one of self-supervised representation-based adaptation (SSRA), ensemble mapping, or self-supervised adaptation loss, generating the customized SE model by denoising the noisy speech data using the trained customized SE model, and providing the customized SE model to a user device to use the denoised noisy speech data.

Abstract: A method for performing cuffless blood pressure (BP) measurement, including: obtaining a first physiological signal and a second physiological signal associated with a user; providing the first physiological signal as an input to a first transformer model; providing the second physiological signal as an input to a second transformer model; providing an output of the first transformer model and an output of the second transformer model as inputs to a third transformer model; providing an output of the third transformer model to at least one BP estimation model; and generating an estimated BP value corresponding to the first physiological signal and the second physiological signal based on an output of the at least one BP estimation model

Abstract: A method includes: receiving one or more training text sentences; generating one or more training vectors based on inputting the one or more training sentences input into a text encoder, the one or more training vectors corresponding to one or more operations that an electronic device is configured to perform; generating one or more speech vectors based on one or more speech utterances input into a speech encoder; generating a similarity matrix that compares each of the one or more training vectors with each of the one or more speech vectors; and updating at least one of the text encoder and the speech encoder based on the similarity matrix.

Abstract: A method for performing multi-token prediction by an apparatus includes receiving, from an artificial intelligence (AI) assistance device, a request for an output token sequence that is subsequent to an input token sequence indicated by the request, predicting, by a trained machine learning model, a plurality of candidate output tokens, estimating joint probability distributions of one or more combinations of the plurality of candidate output tokens, calculating joint probabilities of the one or more combinations by masking the joint probability distributions with a co-occurrence weighted mask, determining, based on the joint probabilities, whether to reduce the number of candidate output tokens included in each combination of the one or more combinations, identifying, based on the joint probabilities, a combination of the one or more combinations as the output token sequence, and outputting, to the AI assistance device, a response to the request, the response comprising the output token sequence.

Abstract: A method for enabling an improved device control capability of a small language model (SLM) transferrable to a hub device configured to be operable by a user in an environment, is disclosed. The method includes performing a fine-tuning the SLM based on a data set including base plans and contrastive plans; generating computer codes corresponding to the fine-tuned SLM; and transferring the generated computer codes to the hub device to be connected with a group of the electronic devices in the environment.

Abstract: Embodiments of the present disclosure are directed to a system for monitoring moving and stepping silicon wafers and a method adopts the system. The system includes a first sensing element to monitor whether a gap between adjacent silicon wafers exists, and a second sensing element to monitor whether two silicon wafers are laminated. The first sensing element and the second sensing element are both disposed in a slot where the silicon wafers are turned over, and suspended on a front side of a sorting wheel in the slot and obliquely arranged towards one side of the sorting wheel. The system installed in the narrow space between the cleaning tank and the sorting tank has simple structure and is capable of quickly and accurately monitoring the silicon wafer turnover, and accurately determining whether there is a continuous or laminated silicon wafer during the silicon wafer turnover process.

Abstract: A silicon wafer transfer auxiliary device includes blowpipes disposed on a side of a sorting wheel. Nozzles of the blowpipes are inclined towards the side of the sorting wheel. The blowpipes are capable of spraying liquid or gas towards a silicon wafer during transfer to increase an adsorption force between the silicon wafer and the sorting wheel and to successfully complete a turning of the silicon wafer from a vertical direction to a side close to the sorting wheel.

Abstract: A liquid storage tank may include a body with an inner cavity, wherein a first liquid storage area and a second liquid storage area are provided in the body, and a partition is provided between the first liquid storage area and the second liquid storage area. A heater for heating solution is provided in the first liquid storage area. A liquid outlet is provided in the second liquid storage area.

Abstract: A method of interpreting a verbal input, may include: assigning a meaning classification to the verbal input, and a confidence score to the meaning classification; and based on the confidence score corresponding to the meaning classification of the verbal input being less than or equal to a threshold, generating at least one paraphrase of the verbal input using at least one large language model (LLM); assigning the meaning classification to the at least one paraphrase, and the confidence score to the meaning classification; and concatenating the verbal input, the at least one paraphrase, the meaning classification, and the confidence score to generate a concatenated input; inputting the concatenated input into the at least one LLM.

Abstract: A method includes identifying multiple tokens contained in an input utterance. The method also includes generating slot labels for at least some of the tokens contained in the input utterance using a trained machine learning model. The method further includes determining at least one action to be performed in response to the input utterance based on at least one of the slot labels. The trained machine learning model is trained to use attention distributions generated such that (i) the attention distributions associated with tokens having dissimilar slot labels are forced to be different and (ii) the attention distribution associated with each token is forced to not focus primarily on that token itself.

Abstract: In one embodiment, a method includes accessing an image and a natural-language question regarding the image and extracting, from the image, a first set of image features at a first level of granularity and a second set of image features at a second level of granularity. The method further includes extracting, from the question, a first set of text features at the first level of granularity and a second set of text features at the second level of granularity; generating a first output representing an alignment between the first set of image features and the first set of text features; generating a second output representing an alignment between the second set of image features and the second set of text features; and determining an answer to the question based on the first output and the second output.

Abstract: A method includes accessing a plurality of weight matrices of a machine learning model. The method also includes, for each weight matrix, decomposing the weight matrix into a U matrix, an S matrix, and a V matrix using singular value decomposition. The S matrix is a diagonal matrix, and a singular group corresponds to each element in the S matrix. The method further includes, for each weight matrix, determining an importance score of each singular group. The importance score of the singular group represents a change in loss if the singular group is removed from the machine learning model. The method also includes, for each weight matrix, ranking the singular groups across the plurality of weight matrices based on the importance scores. In addition, the method includes, for each weight matrix, identifying one or more of the singular groups to prune based on the ranking of the singular groups.

Abstract: In one embodiment, a method includes accessing at least a portion of a training dataset for a trained neural network that includes multiple layers, where each layer includes a number of parameters, and where the training dataset includes multiple training samples that each include an input and a ground-truth output used to train the trained neural network. The method further includes training a hypernetwork to generate a layer-specific compression mask for each of one or more of the multiple layers of the trained neural network. The method further includes generating, by the trained hypernetwork, a final layer-specific compression mask for the trained neural network and compressing the trained neural network by reducing, for each of the one or more layers of the neural network, the number of parameters of that layer according to the final layer-specific compression mask.

Abstract: An apparatus for supplying cleaning liquid may include: a stock solution unit configured with a plurality of stock solution tanks; a cleaning unit configured with a plurality of cleaning tanks; and a transfer unit disposed between the stock solution unit and the cleaning unit. The transfer unit is configured with a plurality of independent transfer tanks. Each of the stock solution tanks communicates with at least one of the transfer tanks, each of the transfer tanks communicates with one of the cleaning tanks, and each of the cleaning tanks communicates with at least one of the transfer tanks.

Abstract: A silicon wafer sorting mechanism includes a wafer distribution unit, a sorting unit, and a transmission unit disposed sequentially. The sorting unit includes a sorting table with one terminal hinged to the wafer distribution unit. The sorting table rotates and tilts with a point where the sorting table is hinged to the wafer distribution unit. The sorting table and the transmission unit are arranged up and down in dislocation or the sorting table is placed horizontally against the transmission unit, thereby forming two transmission lines in different directions to sort silicon wafers of different types.

Abstract: A lifting control apparatus may include: a base for holding a basket containing silicon wafers; brackets provided on an outside of both ends of the base; and a driver for driving the brackets to move up and down. The driver is provided with a lead screw. The lead screw is connected with the bracket on a same side. The lead screw on one side is driven to rotate the lead screw on another side synchronously, and the lead screws on both sides are driven to move the brackets on both sides and the base between the brackets up and down together. Or the lead screws on both sides are driven to rotate synchronously, so that the lead screws on both sides drive the base between the brackets through the correspondingly brackets to move up and down together.

Abstract: A method includes obtaining a batch of training data including multiple paired image-text pairs and multiple unpaired image-text pairs, where each paired image-text pair and each unpaired image-text pair includes an image and a text. The method also includes training a machine learning model using the training data based on an optimization of a combination of losses. The losses include, for each paired image-text pair, (i) a first multi-modal representation loss based on the paired image-text pair and (ii) a second multi-modal representation loss based on two or more unpaired image-text pairs, selected from among the multiple unpaired image-text pairs, wherein each of the two or more unpaired image-text pairs includes either the image or the text of the paired image-text pair.

Abstract: An inpainting method includes obtaining an image including an object having a delicate shape and identifying a target region within the image, where the target region is adjacent to the object. The method also includes using a first mask to separate the image into a number of semantic categories and aggregating neighboring contexts for the target region based on the semantic categories. The method further includes restoring, based on the aggregated contexts, textures in the target region without affecting the delicate shape of the object. In addition, the method includes displaying a refined image including the restored textures in the target region and the object.

Abstract: Provided is a U-shaped network for image restoration. The U-shaped network is lightweight based on a transformer block and is suitable to be deployed on-device, such as in a smartphone. The U-shaped network uses the transformer block to implement encoder, decoder and bottleneck functions. Decoders are connected to respective encoders using skip connections based on element-wise addition, which avoids dimension expansion of concatenation. The transformer block uses a configuration of scaling and pool mixing to process input image data without the need for self-attention computations which permits reduction in memory, reduction in latency, reduction in computational demand, all while maintaining good output image quality.