Abstract: Example implementations relate to systems and methods for detecting disengagement. In an example, a system determines, by a risk identifier, predetermined risk labels based on the user data. The system determines, using a risk evaluator that receives input features, a risk of disengagement for a user. The system determines, using a risk interpreter that receives the input features and the predetermined risk labels, a disengagement reason for the user. The system also determines, based on the risk of disengagement and the disengagement reason, a disengagement prevention incentive for the user. The system further presents, at a computing device associated with the user, a user interface element for interacting with the disengagement prevention incentive.

Abstract: This application relates to apparatus and methods for automatically determining and providing digital customer insights based on historical customer data. In some examples, a computing device obtains user data for a user. In response, the computing device receives a plurality of product types relevant to the user and their corresponding relevance scores. For each product type, the computing device then receives a set of attributes, where each attribute is associated with an affinity score for the user. The computing device determines, for each product type, an overall score for each attribute and product type pair based on the relevance score for the product type and the affinity score for the corresponding attribute. At least one attribute and product type pair is presented to the user based on the corresponding overall score.

Abstract: A pallet return assembly includes at least one support member, a bracket member, a base member including a stopper member, and a lever that, when actuated, moves the base member and its stopper member from a pallet catch position to a pallet release position. In the pallet catch position, the stopper member catches a bottom of a pallet positioned on the roller tracks of the pallet bay and restricts the pallet from moving along the roller tracks. In the pallet release position, the stopper member no longer catches the bottom of the pallet positioned on the roller tracks and no longer restricts the pallet from moving along the roller tracks.

Abstract: In some embodiments, apparatuses and methods are provided herein useful to wheel-based power generation on a cart for transporting commercial products. In some embodiments, a system may include at least one wheel that rotates in multiple directions; a first shaft coupled to the at least one wheel; a rotation transmitter coupled to the at least one wheel and the first shaft; a first gear coupled to the first shaft; a second gear coupled to the first gear; a second shaft coupled to the second gear; and a generator coupled to the second shaft to generate power by converting rotation of the at least one wheel into an alternating current power output. The system may include a control unit coupled to the generator to convert the alternating current power output into a direct current power, and a power storage component coupled to the generator, to store the direct current power output.

Abstract: A computer-implemented method including detecting respective one or more text boxes in each of multiple partial images of a text-bearing area. The method also can include determining respective one or more edge text boxes of the respective one or more text boxes in each of overlapping partial images of the multiple partial images, wherein each of the respective one or more edge text boxes comprise a respective incomplete text. The method additionally can include matching one or more pairs of corresponding edge text boxes from the respective one or more edge text boxes of two adjacent images of the overlapping partial images of the multiple partial images. The method also can include determining cross-image texts in the one or more pairs of the corresponding edge text boxes.

Abstract: Systems and methods of anomaly detection using a generalized time-series forecasting model are disclosed. A time-series dataset is received and a forecast is generated for the time-series dataset utilizing a trained generalized time-series forecasting model configured to utilize a fixed set of characteristic features. A first anomaly determination for the time-series dataset is generated based on a comparison of the forecast for the time-series dataset to the time-series dataset. In response to the first anomaly determination identifying an anomaly, an anomaly identification notification is transmitted to an additional system associated with the anomaly identification.

Abstract: Examples relate to a computer-implemented method of system including a processor that can perform certain operations. The operations can include obtaining campaign inputs for a campaign associated with product specifications for an advertiser. The operations also can include generating, using an approximate nearest neighbor similarity search, targeting cuts for a targeting tactic based at least on the campaign inputs. The targeting cuts can be personalized to the advertiser and the product specifications. The operations additionally can include generating an optimized ad group plan including one or more groups of the targeting cuts and, for each of the one or more groups of the targeting cuts, a respective recommended maximum bid, a respective expected amount of winning impressions, and a respective budget allocation. The operations further can include outputting the optimized ad group plan for the campaign. Other embodiments are described.

Abstract: A computer-implemented method is provided that can smooth a truck flow to a destination node. Input data can be received by a truck flow planning (TFP) system, including truck demand data, truck capacity constraints, and node capacity constraints. The TFP system can include truck demand projection (TDP), truck flow optimization (TFO), and receding horizon control (RHC) components. Truck demand can be projected over a horizon. A plan for the truck flow can be generated that reduces variability while at least maintaining inventory health and can include formulating the plan as a Mixed Integer Programming (MIP) problem. The plan can be adjusted when one or more of new input data features, outputs of the RHC component, outputs of a truck planning optimization (TPO) component, or outputs of a truck load optimization (TLO) component are within thresholds. The plan, as generated and adjusted, can be based on a rolling time window.

Abstract: A method includes receiving candidate recommended items based on an item included in routine reorders of a user. For each candidate recommended item, dimension features for each dimension are embedded as embedding vectors. The dimensions include a user preference, a department affinity, a model suitability, and an item conversion potential dimension. The embedding vectors are combined into a feature vector input to a deep neural network (DNN) model to generate a ranking score representing a likelihood that the user engages with the candidate recommended item. Dimension-specific scores are obtained from intermediate layers of the DNN model including a user preference, a department affinity, a model suitability, and an item conversion potential score, which are output by a multilayer perceptron (MLP) network applied to respective embedding vectors. A final score is derived by incorporating the ranking score with respective weighted contributions. The candidate recommended items are ranked by the final scores.

Abstract: A method including generating, using a semantic embedding generation machine learning model, one or more respective shelf embedding vector representations for each of one or more browse shelves based on a respective shelf name for the each of the one or more browse shelves. The method also can include obtaining a keyword. The method additionally can include generating, using the semantic embedding generation machine learning model, a keyword embedding vector representation based on the keyword. The method further can include determining a respective similarity score between the keyword embedding vector representation and each of the one or more respective shelf embedding vector representations for each of the one or more browse shelves. The method additionally can include determining whether any of the respective similarity scores for the one or more respective shelf embedding vector representations across the one or more browse shelves exceed a predetermined threshold value.

Abstract: Examples may be related to entity-aware multi-task machine learning. An example may involve: obtaining a search query; retrieving, using an entity retrieval model, at least one entity based on the search query; generating, using a neural network, query embedding data based on the search query and the at least one entity; generating, using a plurality of task-specific networks, task prediction data for a plurality of tasks based on the query embedding data, wherein each task of the plurality of tasks captures a different aspect of a user intent associated with the search query; and generating at least one search result for the search query based on the task prediction data for the plurality of tasks.

Abstract: Systems, apparatuses, and methods are provided herein for processing a merchandise return transaction. A system for processing a merchandise return transaction comprising: a user interface for allowing direct entry of information by a customer without assistance from a worker; a database; and a control circuit coupled to the user interface and the database, the control circuit configured to execute the return transaction by: obtaining from the customer identifier relating to a purchase transaction in which at least one item was previously purchased; identifying a product item from the purchase transaction to be returned; obtaining from the customer a reason for the product item to be returned; obtaining from the customer a preferred tender for the return transaction; authorizing the return; and providing instructions to the customer for returning the product item to a retail facility.

Abstract: In some embodiments, apparatuses and methods are provided herein useful to providing customized in-store navigation. In some embodiments, there is provided a system for providing customized in-store navigation including shelving units; shelf electronic shelf labels; product electronic shelf labels; a navigation services control circuit configured to transmit location information associated with the product ESLs at a retail store to an in-store navigation interface operable on a portable electronic device when the portable electronic device is inside the retail store; and the in-store navigation interface configured to: identify a target retail product; receive location information; determine a location of the target retail product; and display directions to the location of the target retail product on a display device.

Abstract: A method can include capturing, via a camera, an image for upload. The method further can include determining a polygon mask layer having one or more polygon masks for a user body in the image, comprising: (a) determining, via a body pose landmarking algorithm, key joint points for the user body based on joint landmarks defined in a configuration set; and (b) determining the polygon masks based on the key joint points and one or more regions of interest defined in the configuration set. The method additionally can include determining a skin mask layer with a skin mask for the image. The method also can include blending the one or more polygon masks and the skin mask into a final mask layer comprising a final mask for the image. The method further can include applying the final mask to the image. Other embodiments are disclosed.

Abstract: A method can include upon receiving, from a policy update engine, one or more hold-time recommendations, selectively determining, based on one or more selection rules, one or more selected hold-time values of the one or more hold-time recommendations. The method further can include implementing the one or more selected hold-time values, as determined. The method additionally can include after implementing the one or more selected hold-time values, determining one or more effects associated with the one or more selected hold-time values. The method also can include transmitting the one or more selected hold-time values and the one or more effects to the policy update engine for retraining. Other embodiments are disclosed.

Abstract: Systems and methods including one or more processors and one or more non-transitory storage devices storing computing instructions configured to run on the one or more processors and perform: receiving a first set of facet information corresponding to a first marketplace; receiving a second set of facet information corresponding to a second marketplace; analyzing the first set of facet information to reduce a number of attributes to process; building a machine learning model to map candidate facets from the first set of facet information to facets from the second set of facet information; training the machine learning model based on the first set of facet information, the second set of facet information, and the mapping of candidate facets from the first set of facet information to facets from the second set of facet information; and displaying a selected number of re-ranked facets based on an output from the machine learning model. Other embodiments are disclosed herein.

Abstract: Systems and methods for multi-item searching for limited-channel interfaces are disclosed. In some embodiments, a disclosed method includes: receiving an input message from a requesting device via a limited-communication channel; identifying, via a first trained natural language model, a plurality of queries in the input message; implementing a plurality of search threads, wherein each of the plurality of search threads is configured to generate search results for a corresponding one of the plurality of queries; generating a responsive message comprising at least a portion of the search results for each of the corresponding one of the plurality of queries; and transmitting the responsive message to the requesting device.

Abstract: Example implementations relate to order prediction and automated cart creation. In an example, basket features, item features, and order features associated with a profile are inputted into a predictive model and daily order predictions are outputted by the predictive model. The daily order predictions are inputted into a grouping algorithm to output an order predictability. Profiles are segmented into respective order predictability cohorts of a plurality of order predictability cohorts based on respective order predictabilities. A cart may be automatically created for at least one daily order prediction associated with at least one profile when a respective order predictability cohort belongs to a predetermined order predictability cohort.