no code implementations • 19 Jan 2023 • Abhijit Suprem, Joao Eduardo Ferreira, Calton Pu
Toxic misinformation campaigns have caused significant societal harm, e. g., affecting elections and COVID-19 information awareness.
no code implementations • 22 Nov 2022 • Abhijit Suprem, Sanjyot Vaidya, Joao Eduardo Ferreira, Calton Pu
Recent advances in text classification and knowledge capture in language models have relied on availability of large-scale text datasets.
no code implementations • 16 Nov 2022 • Abhijit Suprem, Purva Singh, Suma Cherkadi, Sanjyot Vaidya, Joao Eduardo Ferreira, Calton Pu
We evaluate ATEAM and KID for vehicle recognition problems and show that our integrated dataset can help off-the-shelf models achieve excellent accuracy on VMMR and vehicle re-id with no changes to model architectures.
no code implementations • 13 Nov 2022 • Abhijit Suprem, Sanjyot Vaidya, Avinash Venugopal, Joao Eduardo Ferreira, Calton Pu
We present several examples of ML pipelines with EdnaML, including a large-scale fake news labeling and classification system with six sub-pipelines managed by EdnaML.
no code implementations • 20 May 2022 • Abhijit Suprem, Sanjyot Vaidya, Suma Cherkadi, Purva Singh, Joao Eduardo Ferreira, Calton Pu
CoLabel performs VMMR with a composite of interpretable features such as vehicle color, type, and make, all based on interpretable annotations of the ground truth labels.
no code implementations • 19 May 2022 • Abhijit Suprem, Calton Pu
Given a set of fake news models trained on multiple domains, we propose an adaptive decision module to select the best-fit model for a new sample.
1 code implementation • 15 May 2022 • Abhijit Suprem, Calton Pu
However, on some subsets of unseen data that overlap with training data, models have higher accuracy.
no code implementations • 9 Nov 2020 • Calton Pu, Abhijit Suprem, Rodrigo Alves Lima
We describe the EDNA/LITMUS tools that implement EBKA, integrating social media such as Twitter and Facebook with authoritative sources such as WHO and CDC, creating and updating live knowledge on the COVID-19 pandemic.
1 code implementation • 6 Oct 2020 • Abhijit Suprem, Calton Pu
To address this, we present (i) the EDNA streaming toolkit for consuming and processing streaming data, and (ii) EDNA-Covid, a multilingual, large-scale dataset of coronavirus-related tweets collected with EDNA since January 25, 2020.
Social and Information Networks
no code implementations • 9 Sep 2020 • Abhijit Suprem, Joy Arulraj, Calton Pu, Joao Ferreira
In this paper, we present a visual data analytics system, called ODIN, that automatically detects and recovers from drift.
no code implementations • 6 Feb 2020 • Abhijit Suprem, Calton Pu
Vehicle re-identification (re-id) is a fundamental problem for modern surveillance camera networks.
no code implementations • 24 Jan 2020 • Abhijit Suprem, Calton Pu, Joao Eduardo Ferreira
We propose a Small, Accurate, and Fast Re-ID (SAFR) design for flexible vehicle re-id under a variety of compute environments such as cloud, mobile, edge, or embedded devices by only changing the re-id model backbone.
1 code implementation • 23 Jan 2020 • Abhijit Suprem, Calton Pu
EventMapper integrates high-latency, high-accuracy corroborative sources such as physical sensors with low-latency, noisy probabilistic sources such as social media streams to deliver real-time, global event recognition.
no code implementations • 9 Dec 2019 • Abhijit Suprem, Rodrigo Alves Lima, Bruno Padilha, Joao Eduardo Ferreira, Calton Pu
Current frameworks for management are designed for multi-camera networks in a closed dataset environment where there is limited variability in cameras and characteristics of the surveillance environment are well known.
no code implementations • 21 Nov 2019 • Abhijit Suprem, Calton Pu
The ubiquity of social networks and human sensors in the field provides a tremendous amount of real-time, live data about true physical events from around the world.
no code implementations • 17 Sep 2019 • Abhijit Suprem, Aibek Musaev, Calton Pu
Our application has high performance: using classifiers trained in 2014, achieving event detection accuracy of 0. 988, compared to 0. 762 for static approaches.
no code implementations • 17 Sep 2019 • Abhijit Suprem, Calton Pu
Specifically, ASSED is a framework to support continuous filter generation and updates with machine learning using streaming data from high-confidence sources (physical and annotated sensors) and social networks.
1 code implementation • 18 Aug 2019 • Yanzhao Wu, Ling Liu, Juhyun Bae, Ka-Ho Chow, Arun Iyengar, Calton Pu, Wenqi Wei, Lei Yu, Qi Zhang
Learning Rate (LR) is an important hyper-parameter to tune for effective training of deep neural networks (DNNs).
no code implementations • 3 Apr 2019 • Lei Yu, Ling Liu, Calton Pu, Mehmet Emre Gursoy, Stacey Truex
However, when the training datasets are crowdsourced from individuals and contain sensitive information, the model parameters may encode private information and bear the risks of privacy leakage.
1 code implementation • 29 Oct 2018 • Yanzhao Wu, Ling Liu, Calton Pu, Wenqi Cao, Semih Sahin, Wenqi Wei, Qi Zhang
Big data powered Deep Learning (DL) and its applications have blossomed in recent years, fueled by three technological trends: a large amount of digitized data openly accessible, a growing number of DL software frameworks in open source and commercial markets, and a selection of affordable parallel computing hardware devices.