Abstract: Systems, methods, and non-transitory computer-readable media can be configured to determine a page embedding for each page in a sequence of pages visited by a user. A pooled page embedding can be determined based on the page embeddings for the sequence of pages visited by the user. One or more page recommendations for the user can be determined based at least in part on the pooled page embedding.

Abstract: Systems and methods of the present disclosure are directed to a method for training a machine-learned visual attention model. The method can include obtaining image data that depicts a head of a person and an additional entity. The method can include processing the image data with an encoder portion of the visual attention model to obtain latent head and entity encodings. The method can include processing the latent encodings with the visual attention model to obtain a visual attention value and processing the latent encodings with a machine-learned visual location model to obtain a visual location estimation. The method can include training the models by evaluating a loss function that evaluates differences between the visual location estimation and a pseudo visual location label derived from the image data and between the visual attention value and a ground truth visual attention label.

Abstract: In one embodiment, a method involves accessing training data, where the training data contains an ordered sequence of data associated with a plurality of entities, training one or more deep learning models to determine, from the ordered sequence of data, a first set of embeddings for each entity of the plurality of entities, where each entity has a plurality of entity attributes, determining, for each of the plurality of entity attributes, a corresponding initial embedding, training the one or more deep-learning models to refine the initial embeddings according to one or more criterion, generating one or more updated embeddings for each of the plurality of entities based on the refined initial embeddings of the plurality of entity attributes, and modifying the first set of embeddings based on the one or more updated embeddings.

Abstract: Systems and methods for training a restoration model can leverage training for two sub-tasks to train the restoration model to generate realistic and identity-preserved outputs. The systems and methods can balance the training of the generation task and the reconstruction task to ensure the generated outputs preserve the identity of the original subject while generating realistic outputs. The systems and methods can further leverage a feature quantization model and skip connections to improve the model output and overall training.

Abstract: At training time, a base neural network can be trained to perform each of a plurality of basis subtasks included in a total set of basis subtasks (e.g., individually or some combination thereof). Next, a description of a desired combined subtask can be obtained. Based on the description of the combined subtask, a mask generator can produce a pruning mask which is used to prune the base neural network into a smaller combined-subtask-specific network that performs only the two or more basis subtasks included in the combined subtask.

Abstract: Systems, methods, and non-transitory computer readable media are configured to apply a spectral clustering technique to at least a portion of a similarity graph to generate clusters of geographic sub-regions constituting geographic regions. A tf-idf technique is performed to determine pages of a social networking system associated with a geographic region as potential local suggestions for a user associated with a geographic sub-region in the geographic region. References to at least a portion of the pages are presented as local suggestions to the user.

Abstract: Systems, methods, and non-transitory computer readable media are configured to train a machine learning model. The training can be based on a training set of embeddings of a first type and a training set of embeddings of a second type. The machine learning model can be trained to receive an embedding of a second type and to output a corresponding embedding of the first type. A given embedding of the second type can be provided as input to the machine learning model. An embedding of the first type can be obtained from the machine learning model. The embedding of the first type can correspond to the given embedding of the second type.

Abstract: Systems, methods, and non-transitory computer readable media are configured to determine an interaction between a first entity and a first item. A second entity can be determined. The first entity can have formed a connection with the second entity on a social networking system. A belief that the second entity will interact with the first item can then be generated.

Abstract: The present disclosure provides for the generation of embeddings within a machine learning framework, such as, for example, a federated learning framework in which a high-quality centralized model is trained on training data distributed over a large number of clients each with unreliable network connections and low computational power. In an example federated learning setting, in each of a plurality of rounds, each client independently updates the model based on its local data and communicates the updated model back to the server, where all the client-side updates are used to update a global model. The present disclosure provides systems and methods that may generate embeddings with local training data while preserving the privacy of a user of the client device.

Abstract: Systems, methods, and non-transitory computer readable media are configured to determine an interaction between a first entity and a first item. A second entity can be determined. The first entity can have formed a connection with the second entity on a social networking system. A belief that the second entity will interact with the first item can then be generated.

Abstract: In one embodiment, a method involves accessing training data, where the training data contains an ordered sequence of data associated with a plurality of entities, training one or more deep learning models to determine, from the ordered sequence of data, a first set of embeddings for each entity of the plurality of entities, where each entity has a plurality of entity attributes, determining, for each of the plurality of entity attributes, a corresponding initial embedding, training the one or more deep-learning models to refine the initial embeddings according to one or more criterion, generating one or more updated embeddings for each of the plurality of entities based on the refined initial embeddings of the plurality of entity attributes, and modifying the first set of embeddings based on the one or more updated embeddings.

Abstract: Systems, methods, and non-transitory computer readable media can determine one or more actions that a user is likely to take on a page associated with a social networking system, based on one or more first machine learning models. One or more card types that correspond to the one or more actions can be ranked based on a second machine learning model. One or more cards can be generated based on the ranked card types, and each card can include a recommended action associated with the page.

Abstract: In one embodiment, an embedding is determined for each entity in a set of entities that is selected from a plurality of entities. Each embedding corresponds to a point in an embedding space, which includes points corresponding to embeddings of entities. The embeddings of the entities are determined using a deep-learning model. Embeddings are determined for each entity attribute in a set of entity attributes. Each of the entity attributes in the set is of an entity-attribute type and is associated with at least one entity. The entity-attribute embeddings are refined using the deep-learning model. The embeddings of the entities in the set of entities are modified based on the entity-attribute embeddings that are associated with the respective entity to obtain updated embeddings for each entity in the set. The updated embeddings include information regarding the entity attributes that are associated with the respective entities.

Abstract: Systems, methods, and non-transitory computer-readable media can be configured to determine a page embedding for each page in a sequence of pages visited by a user. A pooled page embedding can be determined based on the page embeddings for the sequence of pages visited by the user. One or more page recommendations for the user can be determined based at least in part on the pooled page embedding.

Abstract: Systems, methods, and non-transitory computer-readable media can determine at least one scenario that applies to a user of a social networking system based at least in part on features associated with the user. One or more groups of content recommendations associated with the at least one scenario can be determined. Each group of content recommendations can include a set of content items that relate to the at least one scenario. The one or more groups of content recommendations can be provided to the user as recommendations.

Abstract: A social networking system recommends objects, such as pages, of the social networking system to users of the social networking system based on the location of the user. The social networking system obtains location information identifying the location of the user. Based on the location of the user, the social networking system identifies levels of geographical partitions encompassing the location of the user. For each level of geographical partitions, the social networking system accesses relevant objects of the social networking system with connections to users located within the level of geographical partitions. The social networking system may have determined a term frequency-inverse document frequency (tf-idf) value for each relevant object. Based on the number of connections and the tf-idf value associated with each relevant object, the social networking system merges the relevant objects accessed at each level into a set of relevant objects to recommend to the user.

Abstract: Systems, methods, and non-transitory computer readable media configured to generate session information based on information regarding items of a plurality of item types associated with interactions performed by active users of a social networking system. A graph is generated based on the session information. At least a first item of the items is assigned to a cluster based on similarity between the item and the cluster. The cluster is provided to a recommender system to facilitate selection of relevant information for potential presentation to a user.

Abstract: Systems, methods, and non-transitory computer-readable media can determine at least one web site that is of interest to a user of the social networking system. One or more pages can be determined based at least in part on the web site, the one or more pages being accessible through the social networking system. At least one page recommendation that references at least one of the one or more pages can be provided to the user.

Abstract: Systems, methods, and non-transitory computer-readable media can select a set of selected pages from a plurality of pages on a social networking system based on page selection criteria. A set of potential stories from the set of selected pages is aggregated. The set of potential stories are ranked based on ranking criteria. An administrator feed associated with a first page is generated, the administrator feed comprising a plurality of stories from the set of potential stories based on the ranking the set of potential stories.

Abstract: Systems, methods, and non-transitory computer-readable media can train a machine learning model to determine predictive search recommendation based on search prediction information. Search prediction information associated with a user is provided to the machine learning model. A predictive search recommendation is presented to the user based on the machine learning model and the search prediction information. A search is performed based on the predictive search recommendation for one or more search results associated with entity pages on a social networking system.