Keith Campbell

ORCID: 0000-0003-0974-0788
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Anomaly Detection Techniques and Applications
  • Traffic Prediction and Management Techniques
  • Air Traffic Management and Optimization
  • Forecasting Techniques and Applications
  • Parallel Computing and Optimization Techniques
  • Stochastic Gradient Optimization Techniques
  • Human-Automation Interaction and Safety
  • Image and Object Detection Techniques
  • Time Series Analysis and Forecasting
  • Safety Warnings and Signage
  • ERP Systems Implementation and Impact
  • Tensor decomposition and applications
  • Traffic and Road Safety
  • Traffic control and management
  • Vehicular Ad Hoc Networks (VANETs)
  • Advanced Data Storage Technologies
  • Operations Management Techniques
  • Urban and Freight Transport Logistics
  • Image and Signal Denoising Methods

Mitre (United States)
2020-2021

IBM (Canada)
2015

IBM (United States)
2015

Exploitation of parallel architectures has become critical to scalable machine learning (ML). Since a wide range ML algorithms employ linear algebraic operators, GPUs with BLAS libraries are natural choice for such an exploitation. Two approaches commonly pursued: (i) developing specific GPU accelerated implementations complete algorithms; and (ii) kernels primitive operators like matrix-vector multiplication, which then used in algorithms. This paper extends the latter approach by fused...

10.1145/2688500.2688521 article EN 2015-01-24

Exploitation of parallel architectures has become critical to scalable machine learning (ML). Since a wide range ML algorithms employ linear algebraic operators, GPUs with BLAS libraries are natural choice for such an exploitation. Two approaches commonly pursued: (i) developing specific GPU accelerated implementations complete algorithms; and (ii) kernels primitive operators like matrix-vector multiplication, which then used in algorithms. This paper extends the latter approach by fused...

10.1145/2858788.2688521 article EN ACM SIGPLAN Notices 2015-01-24

View Video Presentation: https://doi.org/10.2514/6.2021-2388.vid The current practice of manually processing features for high-dimensional and heterogeneous aviation data is labor-intensive, does not scale well to new problems, prone information loss, affecting the effectiveness maintainability machine learning (ML) procedures. This research explored an unsupervised method, autoencoder, extract effective problems. study variants autoencoders with aim forcing learned representations input...

10.2514/6.2021-2388 article EN AIAA Aviation 2019 Forum 2021-07-28

Computing power, big data, and advancement of algorithms have led to a renewed interest in artificial intelligence (AI), especially deep learning (DL). The success DL largely lies on data representation because different representations can indicate degree the explanatory factors variation behind data. In last few year, most successful story is supervised learning. However, apply learning, one challenge that labels are expensive get, noisy, or only partially available. With consideration we...

10.1109/icns52807.2021.9441606 article EN 2019 Integrated Communications, Navigation and Surveillance Conference (ICNS) 2021-04-20

The current practice of manually processing features for high-dimensional and heterogeneous aviation data is labor-intensive, does not scale well to new problems, prone information loss, affecting the effectiveness maintainability machine learning (ML) procedures. This research explored an unsupervised method, autoencoder, extract effective problems. study variants autoencoders with aim forcing learned representations input assume useful properties. A flight track anomaly detection...

10.48550/arxiv.2011.01464 preprint EN other-oa arXiv (Cornell University) 2020-01-01

Computing power, big data, and advancement of algorithms have led to a renewed interest in artificial intelligence (AI), especially deep learning (DL). The success DL largely lies on data representation because different representations can indicate degree the explanatory factors variation behind data. In last few year, most successful story is supervised learning. However, apply learning, one challenge that labels are expensive get, noisy, or only partially available. With consideration we...

10.48550/arxiv.2103.04768 preprint EN cc-by arXiv (Cornell University) 2021-01-01

Logs of vehicle controller area network (CAN) bus traffic supplemented by accelerometer and GPS data can provide valuable information about the use operation advanced driver assistance systems (ADAS) to broader safety research community. Although CAN message codes are often manufacturer-specific, third-party libraries partial decoding messages from many models, which be augmented reverse-engineering additional signals. This study explored value bus, accelerometer, that were logged on a...

10.1177/03611981211039159 article EN Transportation Research Record Journal of the Transportation Research Board 2021-09-21
Coming Soon ...