A
Andrew Howard
Researcher at Google
Publications - 58
Citations - 46496
Andrew Howard is an academic researcher from Google. The author has contributed to research in topics: Artificial neural network & Object detection. The author has an hindex of 25, co-authored 53 publications receiving 26745 citations. Previous affiliations of Andrew Howard include Columbia University.
Papers
More filters
Patent
Efficient Convolutional Neural Networks and Techniques to Reduce Associated Computational Costs
Andrew Howard,Bo Chen,Dmitry Kalenichenko,Tobias Weyand,Menglong Zhu,M. Andreetto,Weijun Wang +6 more
TL;DR: MobileNets as mentioned in this paper is based on a straight-forward architecture that uses depthwise separable convolutions to build light weight deep neural networks and provides two global hyper-parameters that efficiently trade-off between latency and accuracy.
Proceedings ArticleDOI
Dynamical systems trees
Andrew Howard,Tony Jebara +1 more
TL;DR: DSTs extend Kalman filters, hidden Markov models and nonlinear dynamical systems to an interactive group scenario to accommodate nonlinear temporal activity and provide tractable inference and learning algorithms for arbitrary DST topologies via an efficient structured mean-field algorithm.
Proceedings Article
K for the price of 1. Parameter efficient multi-task and transfer learning
TL;DR: In this paper, a model patch is learned to specialize to each task, instead of fine-tuning the last layer or the entire network, which enables parameter-efficient transfer and multi-task learning with deep neural networks.
Proceedings ArticleDOI
Non-Discriminative Data or Weak Model? On the Relative Importance of Data and Model Resolution
TL;DR: It is shown that up to a point, the input resolution alone plays little role in the network performance, and it is the internal resolution that is the critical driver of model quality.