A5073057301- Topic Modeling
- Natural Language Processing Techniques
- Data Quality and Management
- Semantic Web and Ontologies
- Domain Adaptation and Few-Shot Learning
- Advanced Neural Network Applications
- Adversarial Robustness in Machine Learning
- Video Surveillance and Tracking Methods
- Online Learning and Analytics
- Time Series Analysis and Forecasting
- Robotic Path Planning Algorithms
- Complex Network Analysis Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Database Systems and Queries
- Video Analysis and Summarization
- SARS-CoV-2 and COVID-19 Research
- Speech and dialogue systems
- Intelligent Tutoring Systems and Adaptive Learning
- Image and Video Quality Assessment
- Image Retrieval and Classification Techniques
- Machine Learning and Data Classification
- Machine Learning and Algorithms
- Speech Recognition and Synthesis
- Bayesian Modeling and Causal Inference
- Multimodal Machine Learning Applications
Johns Hopkins University Applied Physics Laboratory
2002-2022
Johns Hopkins University
2004-2022
U.S. National Science Foundation
2021
Expedition (United Kingdom)
2021
University of Virginia
2021
Virginia Department of Health
2021
After a period of rapidly declining U.S. COVID-19 incidence during January-March 2021, increases occurred in several jurisdictions (1,2) despite the rapid rollout large-scale vaccination program. This increase coincided with spread more transmissible variants SARS-CoV-2, virus that causes COVID-19, including B.1.1.7 (1,3) and relaxation prevention strategies such as those for businesses, gatherings, educational activities. To provide long-term projections potential trends cases,...
Remote identification of people is an important capability for security systems. Automatically controlling a pan-tilt-zoom camera effective way to collect high resolution video or images in unconstrained environment. Often there will be more area than cameras available. The must then divide their time among the order view everyone. In this paper, we discuss challenges involved scheduling active observe multiple people. We present some candidate policies address these and evaluate...
Adam Poliak, Max Fleming, Cash Costello, Kenton Murray, Mahsa Yarmohammadi, Shivani Pandya, Darius Irani, Milind Agarwal, Udit Sharma, Shuo Sun, Nicola Ivanov, Lingxi Shang, Kaushik Srinivasan, Seolhwa Lee, Xu Han, Smisha João Sedoc. Proceedings of the 1st Workshop on NLP for COVID-19 (Part 2) at EMNLP 2020.
A challenge to scaling a video surveillance system is the amount of human supervision required for control cameras. In this paper we consider problem coordinating network cameras purpose identifying people. We pose as machine scheduling where each person job that should be scheduled before deadline. To ensure scalability, propose distributed algorithm only depends on neighbor communication. compare performance localized approach.
We demonstrate two annotation platforms that allow an English speaker to annotate names for any language without knowing the language. These provided high-quality ’‘silver standard” annotations low-resource name taggers (Zhang et al., 2017) achieved state-of-the-art performance on surprise languages (Oromo and Tigrinya) at LoreHLT20171 ten TAC-KBP EDL2017 (Ji 2017). discuss strengths limitations compare other methods of creating silver- gold-standard using native speakers. will make our...
The 2017 shared task at the Balto-Slavic NLP workshop requires identifying coarse-grained named entities in seven languages, each entity’s base form, and clustering name mentions across multilingual set of documents. fact that no training data is provided to systems for building supervised classifiers further adds complexity. To complete we first use publicly available parallel texts project entity recognition capability from English evaluation language. We ignore entirely subtask...
Training reinforcement learning agents that continually learn across multiple environments is a challenging problem. This made more difficult by lack of reproducible experiments and standard metrics for comparing different continual approaches. To address this, we present TELLA, tool the Test Evaluation Lifelong Learning Agents. TELLA provides specified, curricula to lifelong while logging detailed data evaluation standardized analysis. Researchers can define share their own over various or...
Self-supervised learning (SSL) methods have resulted in broad improvements to neural network performance by leveraging large, untapped collections of unlabeled data learn generalized underlying structure. In this work, we harness unsupervised augmentation (UDA), an SSL technique, mitigate backdoor or Trojan attacks on deep networks. We show that UDA is more effective at removing trojans than current state-of-the-art for both feature space and point triggers, over a range model architectures,...
While there are high-quality software frameworks for information retrieval experimentation, they do not explicitly support cross-language (CLIR). To fill this gap, we have created Patapsco, a Python CLIR framework. This framework specifically addresses the complexity that comes with running experiments in multiple languages. Patapsco is designed to be extensible many language pairs, scalable large document collections, and reproducible driven by configuration file. We include results on...
Self-supervised learning (SSL) methods have resulted in broad improvements to neural network performance by leveraging large, untapped collections of unlabeled data learn generalized underlying structure. In this work, we harness unsupervised augmentation (UDA), an SSL technique, mitigate backdoor or Trojan attacks on deep networks. We show that UDA is more effective at removing trojans than current state-of-the-art for both feature space and point triggers, over a range model architectures,...