-
DreamBooth3D: Subject-Driven Text-to-3D Generation
Authors:
Amit Raj,
Srinivas Kaza,
Ben Poole,
Michael Niemeyer,
Nataniel Ruiz,
Ben Mildenhall,
Shiran Zada,
Kfir Aberman,
Michael Rubinstein,
Jonathan Barron,
Yuanzhen Li,
Varun Jampani
Abstract:
We present DreamBooth3D, an approach to personalize text-to-3D generative models from as few as 3-6 casually captured images of a subject. Our approach combines recent advances in personalizing text-to-image models (DreamBooth) with text-to-3D generation (DreamFusion). We find that naively combining these methods fails to yield satisfactory subject-specific 3D assets due to personalized text-to-im…
▽ More
We present DreamBooth3D, an approach to personalize text-to-3D generative models from as few as 3-6 casually captured images of a subject. Our approach combines recent advances in personalizing text-to-image models (DreamBooth) with text-to-3D generation (DreamFusion). We find that naively combining these methods fails to yield satisfactory subject-specific 3D assets due to personalized text-to-image models overfitting to the input viewpoints of the subject. We overcome this through a 3-stage optimization strategy where we jointly leverage the 3D consistency of neural radiance fields together with the personalization capability of text-to-image models. Our method can produce high-quality, subject-specific 3D assets with text-driven modifications such as novel poses, colors and attributes that are not seen in any of the input images of the subject.
△ Less
Submitted 27 March, 2023; v1 submitted 23 March, 2023;
originally announced March 2023.
-
DER Forecast using Privacy Preserving Federated Learning
Authors:
Venkatesh Venkataramanan,
Sridevi Kaza,
Anuradha M. Annaswamy
Abstract:
With increasing penetration of Distributed Energy Resources (DERs) in grid edge including renewable generation, flexible loads, and storage, accurate prediction of distributed generation and consumption at the consumer level becomes important. However, DER prediction based on the transmission of customer level data, either repeatedly or in large amounts, is not feasible due to privacy concerns. In…
▽ More
With increasing penetration of Distributed Energy Resources (DERs) in grid edge including renewable generation, flexible loads, and storage, accurate prediction of distributed generation and consumption at the consumer level becomes important. However, DER prediction based on the transmission of customer level data, either repeatedly or in large amounts, is not feasible due to privacy concerns. In this paper, a distributed machine learning approach, Federated Learning, is proposed to carry out DER forecasting using a network of IoT nodes, each of which transmits a model of the consumption and generation patterns without revealing consumer data. We consider a simulation study which includes 1000 DERs, and show that our method leads to an accurate prediction of preserve consumer privacy, while still leading to an accurate forecast. We also evaluate grid-specific performance metrics such as load swings and load curtailment and show that our FL algorithm leads to satisfactory performance. Simulations are also performed on the Pecan street dataset to demonstrate the validity of the proposed approach on real data.
△ Less
Submitted 7 July, 2021;
originally announced July 2021.
-
Evaluating Link-Based Techniques for Detecting Fake Pharmacy Websites
Authors:
Ahmed Abbasi,
Siddharth Kaza,
F. Mariam Zahedi
Abstract:
Fake online pharmacies have become increasingly pervasive, constituting over 90% of online pharmacy websites. There is a need for fake website detection techniques capable of identifying fake online pharmacy websites with a high degree of accuracy. In this study, we compared several well-known link-based detection techniques on a large-scale test bed with the hyperlink graph encompassing over 80 m…
▽ More
Fake online pharmacies have become increasingly pervasive, constituting over 90% of online pharmacy websites. There is a need for fake website detection techniques capable of identifying fake online pharmacy websites with a high degree of accuracy. In this study, we compared several well-known link-based detection techniques on a large-scale test bed with the hyperlink graph encompassing over 80 million links between 15.5 million web pages, including 1.2 million known legitimate and fake pharmacy pages. We found that the QoC and QoL class propagation algorithms achieved an accuracy of over 90% on our dataset. The results revealed that algorithms that incorporate dual class propagation as well as inlink and outlink information, on page-level or site-level graphs, are better suited for detecting fake pharmacy websites. In addition, site-level analysis yielded significantly better results than page-level analysis for most algorithms evaluated.
△ Less
Submitted 27 September, 2013;
originally announced September 2013.
