Abstract: A mechanism is provided for implementing a bias detection mechanism that mitigates unintended bias in a conversational agent by leveraging conversational agent definitions, a conversational agent chat logs, and user satisfaction statistics. One or more protected attributes are identified within an utterance from the conversational agent chat logs. Using the identified protected attributes, a replacement utterance with a replacement term is generated for at least one of the identified protected attributes in the utterance. A score is generated for the utterance and the replacement utterance using utterance level relative term importance for protected attributes and regular terms in the utterance and the replacement utterance. Utilizing the scoring, a determination is made as to whether unintended bias exists within the utterance. Responsive to unintended bias being detected, an action is implemented that causes a change to a machine learning model used by the conversational agent.

Abstract: An approach to determine the quality of encodings assigned to a word by a word embedding model. The approach may include determining the asymmetry of two embeddings associated with two words from a word embedding model. The asymmetry of the two words from a preexisting evocation dataset may be determined. The asymmetry of the two embeddings may be compared to the asymmetry from the evocation dataset to generate an encoding quality score.

Abstract: A method for manufacturing a semiconductor structure is disclosed. The method includes: providing a semiconductor substrate having a plurality of dies thereon; dispensing an underfill material and a molding compound to fill spaces beneath and between the dies; disposing a temporary carrier over the dies; thinning a thickness of the semiconductor substrate; performing back side metallization upon the thinned semiconductor substrate; removing the temporary carrier; and attaching a plate over the dies. An associated semiconductor structure is also disclosed.

Abstract: A pixel circuit includes a driving sub-circuit, a data writing sub-circuit, a compensation sub-circuit and a storage sub-circuit. The data writing sub-circuit is configured to write a data signal into the second node in response to a first scan signal. The driving sub-circuit is configured to transmit the data signal and a compensation signal to the third node. The compensation sub-circuit is configured to write the data signal and the compensation signal into the first node in response to a second scan signal. The driving sub-circuit is further configured to output a driving signal to a light-emitting device at least according to a voltage of the first node. The storage sub-circuit is configured to store the data signal and the compensation signal, and includes first capacitors and switching element(s). Each switching element is configured to be turned on in response to a control signal, so that at least two first capacitors are connected in parallel.

Abstract: A detection structure, a display panel, a detection device and a detection system, the detection structure is located in display panel, and detection structure includes test units and groups of first connecting lines; each test units includes chip bonding part, test pin parts located at chip bonding part, and groups of second connecting lines; in each test units, first end of each group in groups of second connecting lines is connected with test pin parts, and second end of each group of second connecting lines is connected with chip bonding part; in any two adjacent test units in test units, two adjacent test pin parts are respectively connected with first end and second end of one group of first connecting lines; and when detection structure is used in lightening test, at least two test units arranged at intervals are configured for being electrically connected with detection device respectively.

Abstract: A semiconductor package including a first semiconductor device, a second semiconductor device, an insulating encapsulant, a redistribution structure and a supporting element is provided. The insulating encapsulant encapsulates the first semiconductor device and the second semiconductor device. The redistribution structure is over the first semiconductor device, the second semiconductor device and the insulating encapsulant. The redistribution structure is electrically connected to the first semiconductor device and the second semiconductor device. The supporting element is embedded in one of the insulating encapsulant and the redistribution structure.

Abstract: A plug connector includes: an insulative housing having a front mating end and a rear terminating end; a conductive module received in the insulative housing and exposed to the front mating end; a rod mounted to the insulative housing; a latch mounted to the insulative housing, the latch including a securing portion, a latching portion movable between a latched position and a released position, and an operating portion operable by the rod to move the latching portion to the released position; and a pulling tab connected to the rod, wherein the insulative housing includes a pair of grooves and the rod is slidable in the pair of grooves by the pulling tab to operate the operating portion of the latch to move the latching portion to the released position.

Abstract: A method for systematically controlling rapid proliferation of cyanobacteria cells in lakes in spring is described. The method applies physical, biological and chemical methods in an integrated and synergistic manner and includes using flexible enclosures to close a control water area and using modified soil to adsorb and settle algal cells; coupling four technologies including algae control by fish, algae control by biological floating islands, algae inhibition by plants and algae control by microorganisms; and using H2O2 to kW algae and using 40% FeCl3+CaCO3 and 50% Ca(NO3)2 to treat sediment. The method establishes a complete system for physical, chemical and biological control of algae, forms joint control of exogenous contaminants, endogenous contaminants and cyanobacteria, integrates physical, biological and chemical methods and optimizes systematic methods that are used synergistically.

Abstract: The present disclosure provides a tidal simulation test device and a method of use thereof. The tidal simulation test device includes a water supply pool, a water level adjustment pool, a subtidal zone simulation pool, an intertidal zone simulation pool, a supratidal zone simulation pool and a two-way water flow adjustment control box. The two-way water flow adjustment control box is provided therein with an electromagnetic flow control meter, a forward frequency conversion self-priming pump, a reverse frequency conversion self-priming pump and an intelligent time-controlled three-way controller. In the method, the start and stop of the forward frequency conversion self-priming pump and the reverse frequency conversion self-priming pump are controlled through an intelligent switch, and a water flow is adjusted through the electromagnetic flow control meter, thereby realizing periodic changes in a water level to simulate different types of tides and coastal wetlands.

Abstract: A well log is scanned for one or more dimensions that describe one or more features of a well. Each dimension includes a plurality of values in a numerical format that represents each dimension. A missing value is detected in a first plurality of values of a first dimension of the well log. The first dimension of the well log is transformed, in response to the missing value, into a first image that visually depicts the first dimension including the first plurality of values and the missing value. Based on the first image and based on an image analysis algorithm a second image is created that visually depicts the first plurality of values and includes a found depiction visually depicting a found value in place of the missing value. The found depiction is converted, based on the second image, into a first value in the numerical format.

Abstract: A mobile computing device is configured to dynamically aggregate wireless communications on a common antenna. The device determines that a WiFi communication in a first frequency band is associated with an antenna, the WiFi communication having a number of assigned transmission time slots. The device determines that a second wireless communication in a second frequency band also is associated with the antenna, and determines a periodicity and/or a media quality of the second wireless communication. Based on the periodicity and/or media quality of the second wireless communication, the device aggregates the WiFi communication and the second wireless communication, the aggregation comprising assigning a number of packets of the second wireless communication to an aggregation frame of a plurality of frames associated with the transmission time slots assigned to the WiFi communication.

Abstract: A mechanism is provided to implement a text classifier training augmentation mechanism for incorporating unlabeled data into the generation of a text classifier. For each term of a plurality of terms in each document of a plurality of documents in a set of unlabeled data, a term frequency value is determined. The term is normalized by dividing the term frequency value by a total number of terms in the document. An inverse document frequency (idf) value is determined for each term based on the term frequency value. A subset of terms is filtered from the plurality of terms based the determined idf values. The idf values for the remaining terms are transformed into feature weights. Terms from a set of labeled data are re-weighted based on the feature weights determined from the set of unlabeled data. The text classifier is then generated using the re-weighted labeled data.

Abstract: An artificial intelligence (AI) platform to support random action replay for natural language (NL) learning. A NL conversation is explored to train a neural network. One or more tuples are leverage for the training, with each tuple representing an input action, a vector, an output action, and a reward value. An action is sampled from the vector, with the sampling including assessment of a corresponding first gradient. The first gradient is applied to selectively adjust the neural network. As NL input is received and applied to the selectively adjusted neural network, an output corresponding to the NL input is identified and a corresponding action is executed.

Abstract: An intelligent pesticide spraying robot self-adaptive to terrain of mountain land is provided. Walking mechanisms are respectively connected to both side ends of a chassis. A pesticide spraying mechanism includes rotary drive mechanisms respectively connected to both sides of the rear end of the chassis, telescopic spray boom mechanisms correspondingly connected to the rotary drive mechanisms, rotary nozzle mechanisms correspondingly connected to the telescopic spray boom mechanisms, a liquid pesticide storage box fixed to the top of the chassis, a first water pump fixed in the liquid pesticide storage box, and first water pipes in communication with the first water pump and the rotary nozzle mechanisms. A control chip of a control identification mechanism is communicatively connected to the walking mechanisms, the rotary drive mechanisms, the telescopic spray boom mechanisms, the rotary nozzle mechanisms, the first water pump, a front-end camera, side-end cameras and ultrasonic sensors respectively.

Abstract: Disclosed are a pulverized coal preprocessing method and a pulverized coal gasifying method. The pulverized coal preprocessing method comprises the following steps: (1) performing pore broadening on pulverized coal to obtain preprocessed pulverized coal; (2) loading alkali metal ions into the preprocessed pulverized coal under an ion exchange condition to obtain alkali metal loaded pulverized coal. The method further comprises loading a chrome complex into the alkali metal loaded pulverized coal obtained in described step (2). In gasification, the pulverized coal loaded with alkali metal potassium and chrome catalysts obtained by the method has the advantages of high sulphur removal rate, high carbon conversion rate, short gasifying reaction time and high methane production.

Abstract: A test method for an eye tracking algorithm creatively defines indicators for eye tracking algorithm performance test, and proposes a general environment and an overall process for the eye tracking algorithm performance test to realize the versatility of the eye tracking algorithm test, improve the algorithm development efficiency, shorten the development cycle, and provide a direction and reference for the algorithm development and iteration. The test method decouples the performance test of an eye tracking algorithm from development of the eye tracking algorithm itself, so that a test environment is not limited to a specific eye tracking algorithm and a general performance test environment for the eye tracking algorithm is established. The test method includes: defining indicators for the eye tracking algorithm performance test, defining an overall architecture of a performance test environment of the eye tracking algorithm, and proposing steps for performing the eye tracking algorithm performance test.

Abstract: A method for producing an abrasion-resistant coating on the inner wall of an aluminum alloy workpiece is provide. The steps include mixing a graphene powder and Al powder to obtain a mixed powder; combining and heating the mixed power with a polyvinyl alcohol (PVA) liquid, and performing spray granulation to obtain a low-temperature self-propagating composite; stirring a slurry comprising the low-temperature self-propagating composite and sodium silicate; injecting the slurry into a cylindrical inner cavity of an aluminum alloy workpiece mounted on a horizontal rotary table for rotation, the aluminum alloy workpiece is heated with the rotation at a second temperature of 80-100° C. so that the slurry is uniformly solidified on the cylindrical inner surface of the cylindrical inner cavity; and burning the slurry, after the slurry is uniformly solidified and while the rotation is maintained, with an oxyacetylene flame to form the wear-resistant coating.

Abstract: A system and method for processing mammographic images of target breast tissue is provided. The mammographic images are processed to generate modified images. A deep learning algorithm, having a tailored Convolutional Neural Networks (CNN) model, is applied to the modified images to generate a first output and a second output. Global features associated with the entirety of the mammographic images are extracted by using the first output. Local features associated with Regions of Interest (ROIs) of the mammographic images are extracted by using the second output. The global features and the local features are combined and fuse to generate an indicator representative of likelihood of malignancy of the target breast tissue.

Abstract: Provided is a method for expanding a human DC cell. The method includes the step of contacting a cell sample of a DC cell to be expanded with a viral transactivator protein sourcing from simian-T-lymphotropic virus (STLV). Also provided are an expanded DC cell prepared by the method, and a DC cell and data repository constructed by the method.