Identifying and Reducing Human Interaction with Text – Interpersonal communication is a fundamental activity in human social interactions. As a consequence of the human-computer interaction, multiple users interacting on different levels of social interaction have a common goal to learn a new communication technique. We propose a collaborative, online method to build a deep neural network to model interpersonal behavior using collaborative filtering over the user interactions. A particular learning algorithm is proposed, which utilizes the data collected from a person’s daily activities in order to learn the underlying state of the user. We evaluate the learning algorithm on several well-known human-computer interactions and show that it has significant performance gain compared to state-of-the-art approaches.
We present a new method for learning and applying a Bayesian network to extract useful information from a large dataset of unlabeled data sets, often collected from the social network data. We demonstrate that such a network can successfully extract useful information from the dataset and, by using it, extract useful information in the form of a probabilistic belief graph constructed from social media posts. The network structure and the content of the posts are analyzed with both qualitative and quantitative metrics to obtain a Bayesian model that outperforms other model models and achieves the best performance in the state-of-the-art task.
Distributed Convex Optimization for Graphs with Strong Convexity
A Hierarchical Clustering Model for Knowledge Base Completion
Identifying and Reducing Human Interaction with Text
Feature-Augmented Visuomotor Learning for Accurate Identification of Manipulating Objects
Efficient Batch Sufficient Verification to Train Large-Scale Bayesian Networks on True ConditionsWe present a new method for learning and applying a Bayesian network to extract useful information from a large dataset of unlabeled data sets, often collected from the social network data. We demonstrate that such a network can successfully extract useful information from the dataset and, by using it, extract useful information in the form of a probabilistic belief graph constructed from social media posts. The network structure and the content of the posts are analyzed with both qualitative and quantitative metrics to obtain a Bayesian model that outperforms other model models and achieves the best performance in the state-of-the-art task.