Since deep neural networks (DNNs) became successful, domain adaptation methods also leveraged them for representation learning.

In this paper, we capture the dependence among the labels using a learned tree-structured hierarchy.

First announced in March last year, Amman-based Mawdoo3 has finally launched its Arabic personal voice assistant, Salma. The personal assistant was recently launched at TechWadi Annual Forum 2019 last month in California. After the launch, Salma has been made available as a standalone iOS and Android app. According to its (her?) website, the personal assistant can share weather forecasts, currency exchange rates, and prayer times. It (she?) can also help users set an alarm on their phone, play music from their favorite apps, or call anyone from their address book – all with a quick voice command.

Historically, most of the data businesses have analyzed for decision-making has been of the structured variety--easily entered, stored, and queried. In the digital age, that universe of potentially valuable data keeps expanding exponentially. Most of it is unstructured data, coming from a wide variety of sources, from websites to wearable devices. As a recent McKinsey Global Institute report noted: "Much of this newly available data is in the form of clicks, images, text, or signals of various sorts, which is very different than the structured data that can be cleanly placed in rows and columns." At the same time, we have entered an era when machine learning can theoretically find patterns in vast amounts of data to enable enterprises to uncover insights that may not have been visible before.

Michael B. Jordan scored with a breakthrough role in a football series and took a swing at movie stardom in a boxing flick. The "Creed" star has transitioned to basketball for his latest -- though virtual -- role in the "NBA 2K17" video game. Jordan performs in the game's MyCAREER storyline in the latest edition of the hoops hit. "NBA 2K11" and "NBA 2K16" are widely considered among the best basketball video games. Jordan hoped another role in a continuation of a popular sports anthology would add this year's version to the list of best basketball games.

We propose a new Bayesian model for reliable aggregatio of crowdsourced estimates of real-valued quantities in participatory sensing applications. Existing approaches focus on probabilistic modelling of user’s reliability as the key to accurate aggregation. However, these are either limited to estimating discrete quantities, or require a significant number of reports from each user to accurately model their reliability. To mitigate these issues, we adopt a community-based approach, which reduces the data required to reliably aggregate real-valued estimates, by leveraging correlations between the reporting behaviour of users belonging to different communities. As a result, our method is up to 16.6% more accurate than existing state-of-the-art methods and is up to 49% more effective under data sparsity when used to estimate Wi-Fi hotspot locations in a real-world crowdsourcing application.

While topic models can discover patterns of word usage in large corpora, it is difficult to meld this unsupervised structure with noisy, human-provided labels, especially when the label space is large. In this paper, we present a model-Label to Hierarchy (L2H)-that can induce a hierarchy of user-generated labels and the topics associated with those labels from a set of multi-labeled documents. The model is robust enough to account for missing labels from untrained, disparate annotators and provide an interpretable summary of an otherwise unwieldy label set. We show empirically the effectiveness of L2H in predicting held-out words and labels for unseen documents.

In this note we provide detailed derivations of two versions of small-variance asymptotics for hierarchical Dirichlet process (HDP) mixture models and the HDP hidden Markov model (HDP-HMM, a.k.a. the infinite HMM). We include derivations for the probabilities of certain CRP and CRF partitions, which are of more general interest.

How should statistical procedures be designed so as to be scalable computationally to the massive datasets that are increasingly the norm? When coupled with the requirement that an answer to an inferential question be delivered within a certain time budget, this question has significant repercussions for the field of statistics. With the goal of identifying "time-data tradeoffs," we investigate some of the statistical consequences of computational perspectives on scability, in particular divide-and-conquer methodology and hierarchies of convex relaxations.

We develop stochastic variational inference, a scalable algorithm for approximating posterior distributions. We develop this technique for a large class of probabilistic models and we demonstrate it with two probabilistic topic models, latent Dirichlet allocation and the hierarchical Dirichlet process topic model. Using stochastic variational inference, we analyze several large collections of documents: 300K articles from Nature, 1.8M articles from The New York Times, and 3.8M articles from Wikipedia. Stochastic inference can easily handle data sets of this size and outperforms traditional variational inference, which can only handle a smaller subset. (We also show that the Bayesian nonparametric topic model outperforms its parametric counterpart.) Stochastic variational inference lets us apply complex Bayesian models to massive data sets.