Abstract: There are provided system and method of segmentation a fabrication process (FP) image obtained in a fabrication of a semiconductor specimen. The method comprises: upon obtaining a Deep Neural Network (DNN) trained to provide segmentation-related data, processing a fabrication process (FP) sample using the obtained trained DNN and, resulting from the processing, obtaining by the computer segments-related data characterizing the FP image to be segmented, the obtained segments-related data usable for automated examination of the semiconductor specimen. The DNN is trained using a segmentation training set comprising a plurality of first training samples and ground truth data associated therewith, each first training sample comprises a training image; FP sample comprises the FP image to be segmented.

Abstract: A method for receiving training data for training a neural network to perform a machine learning task and for searching for, using the training data, an optimized neural network architecture for performing the machine learning task is described. Searching for the optimized neural network architecture includes: maintaining population data; maintaining threshold data; and repeatedly performing the following operations: selecting one or more candidate architectures from the population data; generating a new architecture from the one or more selected candidate architectures; for the new architecture: training a neural network having the new architecture until termination criteria for the training are satisfied; and determining a final measure of fitness of the neural network having the new architecture after the training; and adding data defining the new architecture and the final measure of fitness for the neural network having the new architecture to the population data.

Abstract: A processor may generate initial subspaces by reducing data of a plurality of domains to a same dimensionality, cluster data within each initial subspace into clusters, create an input data matrix by projecting data from each cluster to an associated one of a plurality of generative subspaces associated with the plurality of clusters, perform a parallel transport between first and second sets of generative subspaces to identify a set of directions between pairs of subspaces of the first and second sets, sample points comprising a plurality of intermediate representations along the directions via an exponential mapping, create an expanded set of prior information by projecting the data from the first and second sets to the plurality of intermediate representations, perform a sparse coding with the input data matrix and the expanded set of prior information, and perform a cross domain inference using the output of the sparse coding.

Abstract: A method of training and using a machine learning model that controls for consideration of undesired factors which might otherwise be considered by the trained model during its subsequent analyses of new data. For example, the model may be a neural network trained on a set of training images to evaluate an insurance applicant based upon an image or audio data of the insurance applicant as part of an underwriting process to determine an appropriate life or health insurance premium. The model is trained to probabilistically correlate an aspect of the applicant's appearance with a personal and/or health-related characteristic. Any undesired factors, such as age, sex, ethnicity, and/or race, are identified for exclusion. The trained model receives the image (e.g., a “selfie”) of the insurance applicant, analyzes the image without considering the identified undesired factors, and suggests the appropriate insurance premium based only on the remaining desired factors.

Abstract: A decode unit to decode an instruction that indicates a source packed data that includes data elements, and indicates a source mask that includes mask elements. Each of the mask elements corresponds to a different one of the data elements. Each of the mask elements is one of a masked mask element and an unmasked mask element. The processor also includes an execution unit coupled with the decode unit. The execution unit, in response to the instruction, is to store a result packed data. When the source packed data includes one or more masked data elements disposed within unmasked data elements, the result packed data includes, the unmasked data elements consolidated together without the one or more masked data elements disposed within them. The execution unit, is to store a result in a second destination storage location that reflects a number of the unmasked data elements consolidated together.

Abstract: Training prediction models and applying machine learning prediction to data is illustrated herein. A prediction instance comprising a set of data and metadata associated with the set of data identifying a prediction type is obtained. The data and metadata are used to determine an entity to train a prediction model using the prediction type. A trained prediction model is obtained from the entity. A notification system may be configured to react to monitor contextual information and apply the prediction. A workflow system may automatically perform a function in a workflow based on prediction.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for selecting actions from large discrete action sets. One of the methods includes receiving a particular observation representing a particular state of an environment; and selecting an action from a discrete set of actions to be performed by an agent interacting with the environment, comprising: processing the particular observation using an actor policy network to generate an ideal point; determining, from the points that represent actions in the set, the k nearest points to the ideal point; for each nearest point of the k nearest points: processing the nearest point and the particular observation using a Q network to generate a respective Q value for the action represented by the nearest point; and selecting the action to be performed by the agent from the k actions represented by the k nearest points based on the Q values.

Abstract: An interactive recommendation system and an interactive recommendation method are provided. The interactive recommendation method includes the following steps. Firstly, a server receives a user request from a user device through the Internet. Then, the server analyzes the user request and generates at least one subsidiary request corresponding to the user request. Then, the server issues an interactive request to the user device according to the at least one subsidiary request. The server generates a recommended application program list according to the at least one subsidiary request. After the user device responds to the interactive request, the server recommends at least one application program in the recommended application program list to the user device.

Abstract: A neural device to which a conditioned response function is imparted and a driving method thereof are disclosed. Quantum dots and a polymer insulating layer are formed between upper and lower electrodes. Conductive filaments are formed at interfaces between the quantum dots and the polymer insulating layer. When a positive pulse, which is an unconditioned stimulus signal, is applied, the conductive filaments are formed, and a low resistance state is implemented. As the number of applications of a negative pulse, which is a conditioned stimulus signal, increases, the neural device is switched from a high resistance state to the low resistance state. Through this, the neural device having learning ability for the conditioned stimulus signal may be implemented and driven.

Abstract: A neural network processing system includes one source node having a source memory and a source core, and one destination node having a destination memory and a destination core, the source core and the destination core being von-Neumann cores, the destination memory including weight data storage areas for storing weight data corresponding to each node, an accumulation memory for accumulating the weight data, and an event address memory, the destination core identifying the weight data storage area and accumulating the weight data to store the accumulated weight data in the accumulation memory, the source memory including a data set having first information for identifying the destination node and second information for identifying the weight data storage area, and the source core reading the data set and sending the second information in the data set to the destination node to conduct remote memory write.

Abstract: A computer-implemented method, computerized apparatus and computer program product for activity recognition using adaptive window size segmentation of sensor data stream. A data stream generated by one or more sensors is obtained. A frequency analysis of the data in a first segment of the data stream is performed. A size of a second segment is determined based on the frequency analysis. Activity recognition is performed for the second segment by extracting one or more features of the data therein and applying a machine learning process on the extracted features to obtain a classification of the data into an activity class.

Abstract: In an approach for dynamically determining a suitable SME for a particular critical situation, a processor receives an initial problem statement. A processor constructs a collaboration channel associated with the initial problem statement. A processor identifies an SME based on a rating that associates the SME with the initial problem statement. A processor sends an invitation to the SME to join the collaboration channel.

Abstract: Various implementations of a method, system, and computer program product for pattern matching using a reconfigurable array processor are disclosed. In one embodiment, a processor array manager of the reconfigurable array processor receives an input data stream for pattern matching and generates a tokenized input data stream from the input data stream. A different portion of the tokenized input data stream is provided to each of a plurality of processing elements of the reconfigurable array processor. Each processing element can compare the received portion of the tokenized input data stream against one or more reference patterns to generate an intermediate result that indicates whether the portion of the tokenized input data stream matches a reference pattern. The processor array manager can combine the intermediate results received from each processing element to yield a final result that indicates whether the input data stream includes a reference pattern.

Abstract: Various implementations of a method, system, and computer program product for pattern matching using a reconfigurable array processor are disclosed. In one embodiment, a processor array manager of the reconfigurable array processor receives an input data stream for pattern matching and generates a tokenized input data stream from the input data stream. A different portion of the tokenized input data stream is provided to each of a plurality of processing elements of the reconfigurable array processor. Each processing element can compare the received portion of the tokenized input data stream against one or more reference patterns to generate an intermediate result that indicates whether the portion of the tokenized input data stream matches a reference pattern. The processor array manager can combine the intermediate results received from each processing element to yield a final result that indicates whether the input data stream includes a reference pattern.

Abstract: Systems and methods for guided user actions are described, including gathering information associated with a detected first action; applying a predictive model to suggest a second action based on an application of machine learning to the first action and the information; calculating a first confidence level associated with the application of the machine learning to the first action and the information; for the first confidence level exceeding a threshold, providing the second action, and performing the second action or a third action based on user input.

Abstract: Systems and methods are disclosed for saving and restoring the search state of a pattern-recognition processor. Embodiments include a pattern-recognition processor having a state variable array and a state variable storage array stored in on-chip memory (on-silicon memory with the processor). State variable storage control logic of the pattern-recognition processor may control the saving of state variables from the state variable array to the state variable storage array. The state variable storage control logic may also control restoring of the state variables from the state variable storage array to restore a search state.

Abstract: A processor includes a binary translator an a decoder. The binary translator includes logic to analyze a stream of atomic instructions, identify words by boundary bits in the atomic instructions, generate a mask to identify the words, and load the mask and the plurality of words into an instruction cache line. The words include atomic instructions. At least one word includes more than one atomic instruction. The decoder includes logic to apply the mask to identify a first word from the instruction cache line and decode the first word based upon the applied mask.

Abstract: Techniques for bi-modal learning in a financial context are described. These techniques are configured to improve a user's financial acumen and bring the user into an optimal financial state. Some of these techniques are embodied in a device that operates financial education lessons specifically configured for the improving the user's current financial state. These techniques may implement rewards/penalties (in tokens) for correct/incorrect user responses to financial decisions being presented in these lessons for user to make. By exploiting the user's desire for rewards and tokens and the desire to improve the user's current financial state, these techniques may leverage machine learning techniques to identify an appropriate financial education lesson that is most likely to have a positive effect on the user. Over time, administrating the financial education lessons builds customer loyalty to the device that implements these techniques. Other embodiments are described and claimed.

Abstract: Preventing hazard flushes in an instruction sequencing unit of a multi-slice processor including receiving a load instruction in a load reorder queue, wherein the load instruction is an instruction to load data from a memory location; subsequent to receiving the load instruction, receiving a store instruction in a store reorder queue, wherein the store instruction is an instruction to store data in the memory location; determining that the store instruction causes a hazard against the load instruction; preventing a flush of the load reorder queue based on a state of the load instruction; and re-executing the load instruction.

Abstract: A system and method for the detection of pathogens and other microbes using nucleotide analysis is described. Aligned and unaligned nucleotide sequences are utilized to predict the presence or absence of pathogens and other microbes.