Abstract: An interconnect structure is provided. The interconnect structure includes a transistor on a substrate, a first dielectric layer over the transistor, a first metal line through the first dielectric layer, a second dielectric layer over the first dielectric layer, and a via through the second dielectric layer and on the first metal line. A first side surface of the first dielectric layer includes a first portion in direct contact with the first metal line and a second portion in direct contact with the via, and the first portion of the first side surface of the first dielectric layer is aligned with the second portion of the first side surface of the first dielectric layer.

Abstract: A device comprises a first metal structure, a dielectric structure, a dielectric residue, and a second metal structure. The dielectric structure is over the first metal structure. The dielectric structure has a stepped sidewall structure. The stepped sidewall structure comprises a lower sidewall and an upper sidewall laterally set back from the lower sidewall. The dielectric residue is embedded in a recessed region in the lower sidewall of the stepped sidewall structure of the dielectric structure. The second metal structure extends through the dielectric structure to the first metal structure.

Abstract: A lever-operated latch device includes an assembly of a case body, an actuation body mounted on the case body, a linking member and a slide body. The actuation body has a free end and a pivoted end pivotally connected with the case body in cooperation with elastic members. The free end of the actuation body is formed with two protruding arms and an opening section positioned between the protruding arms. An operation section is disposed in the opening section. The linking member has a first end pivotally connected with the free end of the actuation body (or the operation section) and a second end connected with the slide body. When an operator presses the operation section, the actuation body is permitted to move from a closed position to an opened position so as to drive the linking member and the slide body to move.

Abstract: A magnetoresistive random access memory (MRAM) device includes a first array region and a second array region on a substrate, a first magnetic tunneling junction (MTJ) on the first array region, a first top electrode on the first MTJ, a second MTJ on the second array region, and a second top electrode on the second MTJ. Preferably, the first top electrode and the second top electrode include different nitrogen to titanium (N/Ti) ratios.

Abstract: A method includes, in a first etching step, etching a semiconductor substrate to form first recesses in a first device region and second recesses in a second device regions simultaneously. A first semiconductor strip is formed between the first recesses. A second semiconductor strip is formed between the second recesses. In a second etching step, the semiconductor substrate in the second device region is etched to extend the second recesses. The first recesses and the second recesses are filled with a dielectric material to form first and second isolation regions in the first and second recesses, respectively. The first isolation regions and the second isolation regions are recessed. Portions of the semiconductor substrate in the first and the second device regions protrude higher than top surfaces of the respective first and second isolation regions to form a first and a second semiconductor fin, respectively.

Abstract: A magnetoresistive random access memory (MRAM) structure, including a substrate and multiple MRAM cells on the substrate, wherein the MRAM cells are arranged in a memory region adjacent to a logic region. An ultra low-k (ULK) layer covers the MRAM cells, wherein the surface portion of ultra low-k layer is doped with fluorine, and dents are formed on the surface of ultra low-k layer at the boundaries between the memory region and the logic region.

Abstract: The disclosure provides a method for training generative adversarial network (GAN), a method for generating images by using GAN, and a computer readable storage medium. The method may train the first generator of the GAN with available training samples belonging to the first type category and share the knowledge learnt by the first generator to the second generator. Accordingly, the second generator may learn to generate (fake) images belonging to the second type category even if there are no available training data during training the second generator.

Abstract: The disclosure provides a method for verifying training data, a training system, and a computer program produce. The method includes: receiving a labelled result with a plurality of bounding regions, wherein the labelled result corresponds to an image, the bounding regions are labelled by a plurality of annotators, the annotators comprises a first annotator and a second annotator, and the bounding region comprises a first bounding region labelled by the first annotator and a second bounding region labelled by the second annotator; and determining the first bounding region and the second bounding region respectively corresponds to different two target objects or corresponds to one target object according to a similarity between the first bounding region and the second bounding region.

Abstract: A machine learning method includes steps of: (a) obtaining initial values of hyperparameters and hypernetwork parameters; (b) generating first classification model parameters according to the hyperparameters and the hypernetwork parameters, and updating the hypernetwork parameters according to a classification result based on the first classification model parameters relative to a training sample; (c) generating second classification model parameters according to the hyperparameters and the updated hypernetwork parameters, and updating the hyperparameters according to another classification result based on the second classification model parameters relative to a verification sample; and (d) repeating the steps (b) and (c) for updating the hypernetwork parameters and the hyperparameters.

Abstract: The disclosure provides a method for training generative adversarial network (GAN), a method for generating images by using GAN, and a computer readable storage medium. The method may train the first generator of the GAN with available training samples belonging to the first type category and share the knowledge learnt by the first generator to the second generator. Accordingly, the second generator may learn to generate (fake) images belonging to the second type category even if there are no available training data during training the second generator.

Abstract: A method includes acquiring a plurality of training images through a capturing component, acquiring a plurality of training camera poses of the capturing component corresponding to the training images through a pose sensor disposed corresponding to the capturing component, and training a camera pose estimation model according to the training images and the training camera poses of the capturing component.

Abstract: The disclosure provides a method for verifying training data, a training system, and a computer program produce. The method includes: receiving a labelled result with a plurality of bounding regions, wherein the labelled result corresponds to an image, the bounding regions are labelled by a plurality of annotators, the annotators comprises a first annotator and a second annotator, and the bounding region comprises a first bounding region labelled by the first annotator and a second bounding region labelled by the second annotator; and determining the first bounding region and the second bounding region respectively corresponds to different two target objects or corresponds to one target object according to a similarity between the first bounding region and the second bounding region.

Abstract: The disclosure provides a method for verifying training data, a training system, and a computer program produce. The method includes: providing a plurality of raw data to a plurality of annotators; retrieving a plurality of labelled results, wherein the labelled results includes a plurality of labelled data, and the labelled data are generated by the annotators via labelling the raw data; determining a plurality of consistencies by comparing the labelled results, and accordingly determining whether the labelled results are valid for training an artificial intelligence machine; in response to determining that the labelled results are valid, determining at least a specific part of the labelled results are valid for training the artificial intelligence machine.

Abstract: An image processing method includes generating, by a processing component, a first input feature map based on an input image using a first convolutional neural network; generating, by the processing component, a first template feature map based on a template image using the first convolutional neural network; generating, by the processing component, a first estimated motion parameter based on an initial motion parameter, the first input feature map and the first template feature map using an iterative Lucas-Kanade network; and performing, by the processing component, image alignment between the input image and the template image based on the first estimated motion parameter.

Abstract: An image processing method includes generating, by a processing component, a first input feature map based on an input image using a first convolutional neural network; generating, by the processing component, a first template feature map based on a template image using the first convolutional neural network; generating, by the processing component, a first estimated motion parameter based on an initial motion parameter, the first input feature map and the first template feature map using an iterative Lucas-Kanade network; and performing, by the processing component, image alignment between the input image and the template image based on the first estimated motion parameter.

Abstract: A method includes acquiring a plurality of training images through a capturing component, acquiring a plurality of training camera poses of the capturing component corresponding to the training images through a pose sensor disposed corresponding to the capturing component, and training a camera pose estimation model according to the training images and the training camera poses of the capturing component.

Abstract: Content-aware display adaptation methods for stereoscopic images and editing interfaces and methods are provided. First, the saliency maps are estimated for stereoscopic images. Each image is represented as a grid mesh, and a per-quad importance is measured based on the saliency maps. Then, features are detected matched between the images. An energy function is defined according to the saliency maps, the matched features, data for the grid mesh, and specification of a target display. The energy function consists of at least a disparity consistency energy and/or an alignment energy. The energy function is minimized to obtain two sets of deformed vertex positions for the images. The control indicator displayed in an operational interface is tuned via a touch-sensitive display unit, such that the energy function is accordingly modified, and corresponding deformed stereoscopic images are displayed.