Abstract: A Video Semantic Segmentation System (VSSS) is disclosed that performs accurate and fast semantic segmentation of videos using a set of temporally distributed neural networks. The VSSS receives as input a video signal comprising a contiguous sequence of temporally-related video frames. The VSSS extracts features from the video frames in the contiguous sequence and based upon the extracted features, selects, from a set of labels, a label to be associated with each pixel of each video frame in the video signal. In certain embodiments, a set of multiple neural networks are used to extract the features to be used for video segmentation and the extraction of features is distributed among the multiple neural networks in the set. A strong feature representation representing the entirety of the features is produced for each video frame in the sequence of video frames by aggregating the output features extracted by the multiple neural networks.

Abstract: Embodiments of bonded semiconductor structures and fabrication methods thereof are disclosed. In an example, a bonded structure includes a first bonding layer including a first bonding contact and a first bonding alignment mark, a second bonding layer including a second bonding contact and a second bonding alignment mark, and a bonding interface between the first bonding layer and the second bonding layer. The first bonding alignment mark is aligned with the second bonding alignment mark at the bonding interface, such that the first bonding contact is aligned with the second bonding contact at the bonding interface. The first bonding alignment mark includes a plurality of first repetitive patterns. The second bonding alignment mark includes a plurality of second repetitive patterns different from the plurality of first repetitive patterns.

Abstract: Digital imaging and learning systems and methods are described for analyzing pixel data of an image of a hair region of a user's head to generate one or more user-specific recommendations. A digital image of a user is received at an imaging application (app) and comprises pixel data of at least a portion of a hair region of the user's head. A hair based learning model, trained with pixel data of a plurality of training images depicting hair regions of heads of respective individuals, analyzes the image to determine an image classification of the user's hair region. The imaging app generates, based on the image classification, a user-specific recommendation designed to address at least one feature identifiable within the pixel data comprising the at least the portion of a hair region of the user's head. The imaging app renders on displays screen the user-specific recommendation.

Abstract: Methods and systems for advanced and augmented training of deep neural networks (DNNs) using synthetic data and innovative generative networks. A method includes training a DNN using synthetic data, training a plurality of DNNs using context data, associating features of the DNNs trained using context data with features of the DNN trained with synthetic data, and generating an augmented DNN using the associated features.

Abstract: Disclosed are a protein molecular marker Dkk-3 protein associated with age-related muscle atrophy and the use thereof in the diagnosis of age-related muscle atrophy. The expression level of the Dkk-3 protein in amyotrophic cells is significantly higher than that in normal myocytes or tissues, and thus the Dkk-3 protein can be used as an effective marker for the detection of age-related muscle atrophy.

Abstract: Embodiments of hybrid-bonded semiconductor structures and methods for forming a hybrid-bonded semiconductor structure are disclosed. The method can include providing a substrate and forming a base dielectric layer on the substrate. The method also includes forming first and second conductive structures in the base dielectric layer and disposing an alternating dielectric layer stack. Disposing alternating dielectric layer stack includes disposing a first dielectric layer on the base dielectric layer and the first and second conductive structures and sequentially disposing second, third, and fourth dielectric layers. The method further includes planarizing the disposed alternating dielectric layer stack and etching the alternating dielectric layer stack to form first and second openings using preset etching rates for each of the first, second, third, and fourth dielectric layers. The etching continues until at least portions of the first and second conductive structures are exposed.

Abstract: Embodiments of bonded semiconductor structures and fabrication methods thereof are disclosed. In an example, a semiconductor device includes a first semiconductor structure, a second semiconductor structure, and a bonding interface between the first semiconductor structure and the second semiconductor structure. The first semiconductor structure includes a substrate, a first device layer disposed on the substrate, and a first bonding layer disposed above the first device layer and including a first bonding contact and a first bonding alignment mark. The second semiconductor structure includes a second device layer, and a second bonding layer disposed below the second device layer and including a second bonding contact and a second bonding alignment mark. The first bonding alignment mark is aligned with the second bonding alignment mark at the bonding interface, such that the first bonding contact is aligned with the second bonding contact at the bonding interface.

Abstract: Embodiments of methods and structures for forming a 3D integrated wiring structure are disclosed. The method can include forming an insulating layer on a front side of a first substrate; forming a semiconductor layer on a front side of the insulating layer; patterning the semiconductor layer to expose at least a portion of a surface of the insulating layer; forming a plurality of semiconductor structures over the front side of the first substrate, wherein the semiconductor structures include a plurality of conductive contacts and a first conductive layer; joining a second substrate with the semiconductor structures; performing a thinning process on a backside of the first substrate to expose the insulating layer and one end of the plurality of conductive contacts; and forming a conductive wiring layer on the exposed insulating layer.

Abstract: Methods and systems for budgeted and simplified training of deep neural networks (DNNs) are disclosed. In one example, a trainer is to train a DNN using a plurality of training sub-images derived from a down-sampled training image. A tester is to test the trained DNN using a plurality of testing sub-images derived from a down-sampled testing image. In another example, in a recurrent deep Q-network (RDQN) having a local attention mechanism located between a convolutional neural network (CNN) and a long-short time memory (LSTM), a plurality of feature maps are generated by the CNN from an input image. Hard-attention is applied by the local attention mechanism to the generated plurality of feature maps by selecting a subset of the generated feature maps. Soft attention is applied by the local attention mechanism to the selected subset of generated feature maps by providing weights to the selected subset of generated feature maps in obtaining weighted feature maps.

Abstract: An exemplary embodiment of the present disclosure provides an MRI-compatible robot comprising one or more fiducial markers, a first planar stage comprising a first joint configured to receive a surgical tool and a first mechanism configured to move the surgical tool, a second planar stage comprising a second joint configured to receive the surgical tool and a second mechanism configured to move the surgical tool, and wherein the second planar stage is generally parallel with the first planar stage.

Abstract: The invention provides a gait evaluating system and a gait evaluating method. The gait evaluation system includes a gait evaluating device configured to: obtain, from a pressure detection device, a plurality of pressure values of a user walking on the pressure detection device; obtain a plurality of step feature values of the user based on the pressure values; obtain a plurality of walking limb feature values when the user walks on the pressure detection device based on a sensing data provided by a limb sensing device; and evaluate a gait of the user based on the step feature values and the walking limb feature values.

Abstract: Embodiments of bonded semiconductor structures and fabrication methods thereof are disclosed. In an example, a semiconductor device includes a first and a second semiconductor structures. The first semiconductor structure includes a first interconnect layer including first interconnects. At least one first interconnect is a first dummy interconnect. The first semiconductor structure further includes a first bonding layer including first bonding contacts. Each first interconnect is in contact with a respective first bonding contact. The second semiconductor structure includes a second interconnect layer including second interconnects. At least one second interconnect is a second dummy interconnect. The second semiconductor structure further includes a second bonding layer including second bonding contacts. Each second interconnect is in contact with a respective second bonding contact. The semiconductor device further includes a bonding interface between the first and second bonding layers.

Abstract: Embodiments of bonded semiconductor structures and fabrication methods thereof are disclosed. In an example, a semiconductor device includes a first semiconductor structure, a second semiconductor structure, and a bonding interface between the first and second semiconductor structures. The first semiconductor structure includes a substrate, a first device layer disposed on the substrate, and a first bonding layer disposed above the first device layer and including a first bonding contact. The second semiconductor structure includes a second device layer, and a second bonding layer disposed below the second device layer and including a second bonding contact. The first bonding contact is in contact with the second bonding contact at the bonding interface. At least one of the first bonding contact and the second bonding contact includes a capping layer at the bonding interface and having a conductive material different from a remainder of the respective first or second bonding contact.

Abstract: Described herein are advanced artificial intelligence agents for modeling physical interactions. An apparatus to provide an active artificial intelligence (AI) agent includes at least one database to store physical interaction data and compute cluster coupled to the at least one database. The compute cluster automatically obtains physical interaction data from a data collection module without manual interaction, stores the physical interaction data in the at least one database, and automatically trains diverse sets of machine learning program units to simulate physical interactions with each individual program unit having a different model based on the applied physical interaction data.

Abstract: Techniques for forking and merging of electronically presented content items, such as MMO and other video games, are described herein. In particular, a first content item session may be forked to generate a second (forked) content item session. The first and forked content item sessions may then continue to execute and evolve in separate directions from one another. In some examples, a user of the forked content item session may serve as an administrator of the forked content item session and may receive compensation in exchange for serving as an administrator. Additionally, one or more forked or other separate content item sessions may be merged together to form a merged content item session. In some examples, conflicts between merged content item sessions may be identified and resolved based, at least in part, on input from administrators.

Abstract: Techniques are provided for neural network based, human attribute recognition, guided by anatomical key-points and statistic correlation models. Attributes include characteristics that can be visibly identified or inferred from an image, such as gender, hairstyle, clothing style, etc. A methodology implementing the techniques according to an embodiment includes applying an attribute feature extraction (AFE) convolutional neural network (CNN) to an image of a human to generate attribute feature maps based on the image. The method further includes applying a key-point guided proposal (KPG) CNN to the image of the human to generate proposed hierarchical regions of the image based on associated anatomical key-points.

Abstract: An apparatus for dynamic emotion recognition in unconstrained scenarios is described herein. The apparatus comprises a controller to pre-process image data and a phase-convolution mechanism to build lower levels of a CNN such that the filters form pairs in phase. The apparatus also comprises a phase-residual mechanism configured to build middle layers of the CNN via plurality of residual functions and an inception-residual mechanism to build top layers of the CNN by introducing multi-scale feature extraction. Further, the apparatus comprises a fully connected mechanism to classify extracted features.

Abstract: A Video Semantic Segmentation System (VSSS) is disclosed that performs accurate and fast semantic segmentation of videos using a set of temporally distributed neural networks. The VSSS receives as input a video signal comprising a contiguous sequence of temporally-related video frames. The VSSS extracts features from the video frames in the contiguous sequence and based upon the extracted features, selects, from a set of labels, a label to be associated with each pixel of each video frame in the video signal. In certain embodiments, a set of multiple neural networks are used to extract the features to be used for video segmentation and the extraction of features is distributed among the multiple neural networks in the set. A strong feature representation representing the entirety of the features is produced for each video frame in the sequence of video frames by aggregating the output features extracted by the multiple neural networks.

Abstract: A semiconductor structure is disclosed. The semiconductor structure includes a staircase structure disposed over a substrate. The staircase structure includes a plurality of layer stacks, where each layer stack is made of a first material layer over a portion of a second material layer. The staircase structure further includes a plurality of landing pads, where each landing pad is disposed over another portion of the second material layer of a respective layer stack.

Abstract: Methods and apparatus for multi-task recognition using neural networks are disclosed. An example apparatus includes a filter engine to generate a facial identifier feature map based on image data, the facial identifier feature map to identify a face within the image data. The example apparatus also includes a sibling semantic engine to process the facial identifier feature map to generate an attribute feature map associated with a facial attribute. The example apparatus also includes a task loss engine to calculate a probability factor for the attribute, the probability factor identifying the facial attribute. The example apparatus also includes a report generator to generate a report indicative of a classification of the facial attribute.