SOURCES SOUGHT
A -- Big Data Learning Platform development. - Capability Statement
- Notice Date
- 7/10/2017
- Notice Type
- Sources Sought
- NAICS
- 541712
— Research and Development in the Physical, Engineering, and Life Sciences (except Biotechnology)
- Contracting Office
- NASA Shared Services Center (NSSC), Building 1111, Jerry Hlass Road, Stennis Space Center, Mississippi, 39529, United States
- ZIP Code
- 39529
- Solicitation Number
- RFQ4200621357
- Archive Date
- 7/28/2017
- Point of Contact
- Joseph H Reynolds, Phone: 2288136609
- E-Mail Address
-
joseph.h.reynolds@nasa.gov
(joseph.h.reynolds@nasa.gov)
- Small Business Set-Aside
- N/A
- Description
- Capability Statement for Big Data Learning Platform development. PROJECT NAME: Big Data Learning Platform PROPOSAL: 1. DESCRIPTION: Big Spatiotemporal Data has emerged with new opportunities for research, development, innovation and business. However, the transformation of Big Data into value poses grand challenges for: 1) big data management: the volume and velocity of Big Data require the extensible and scalable data management and access strategy, but also the on-demand computing resource; the variety of Big Data increases the difficulty to design search engines, parallelize queries, and optimize indices; 2) spatiotemporal data modeling: the complex spatiotemporal content and relationship make it difficult to design a data model for better representing and organizing heterogeneous data; 3) spatiotemporal data mining: most existing analytical algorithms require structured homogeneous data and have difficulties in processing the heterogeneity of Big Data (Bertino et al. 2011); more robust models are required to discover the implicit spatiotemporal topological relationship patterns; sophisticated scalable and interoperable algorithms are needed to train and evaluate models involving large volume of data in a reasonable time. To enable such transformation, we propose to develop a deep learning platform based on ongoing center projects. We will investigate how deep learning can help mine value from different geospatial domains, which will accumulate the valuable experience for geospatial data mining. In addition, advanced data management, high performance computing and cloud computing technologies will be utilized to empower deep learning on big spatiotemporal data. 2. OBJECTIVE: This project is to develop a deep learning platform to discover the hidden values in Big Spatiotemporal Data based on the spatiotemporal innovation current projects. The platform will have advanced data management and computing technologies to mine valuable knowledge from Big Spatiotemporal Data. More robust models will be built to discover the implicit spatiotemporal dynamic patterns in climate, dust storm, and weather with remote sensing and model simulation data to solve the concerned environment and health issues. Meanwhile, user- generated data, such as log files and social media, will be mined to improve geospatial data discovering and form a knowledge base for spatiotemporal data. In addition, high performance computing (e.g. GPU and parallel computing) and cloud computing technologies will be utilized to accelerate the knowledge discovering process. The proposed deep learning platform for Big Spatiotemporal Data will be developed/integrated with a suite of software for big spatiotemporal data mining, and contribute a core to spatiotemporal innovation. 3. EXPERIMENTAL PLAN: • Leverage the data containers (GMU-16-06) to manage different spatiotemporal data: • Raster data: SciDB, Rasdaman, HDFS, MongoDB (Li et al. 2016; Yang et al. 2016) • Vector data: MongoDB, HDFS • Text files: MongoDB, Cassandra, HDFS • Utilize high performance data search engine to access big data in real time • Raster data: SciDB, Rasdaman, ClimateSpark (Hu et al. 2016) • Vector data: Hive, SpatialHadoop • Text files: ElasticSearch, Spark, Hive (Jiang et al. 2016) • Apply deep learning technologies to mine the hidden knowledge in geospatial domains • Remote sensing: 1) Prepare the training and test datasets; 2) Training the convolution neutral networks to extract the surface objects (Yue et al. 2016); 3) Build the models to simulate/predict urban land cover change. • Climate: 1) detect the climate anomaly events and patterns (Li et al. 2016); 2) identify the relationship between climate factors and disease vectors (e.g. Dengue and Zika) using deep learning technologies; 3) simulate/model how climate factors affect the disease transmission. • Dust storm: 1) use feature identification and tracking algorithms to automatically detect dust events (Yu and Yang, 2016); 2) study how dust transport relates to atmospheric cycles, rainfall, and other phenomena in the long-term; 3) study how dust events impact on long- term climate, ecosystem, ocean, etc. • Weather: 1) detect extreme weather from big weather datasets; 2) use deep convolutional neural networks to detect automatic features for multiple types of weather events: hurricane, dust storm, flood, volcano (ash plume); 3) bridge the semantic gap between satellite imagery with semantic concepts perceived by human beings • Log: 1) extract user behavior, and detect crawler from usage log; 2) identify the spatiotemporal web usage pattern to Improve crawler detection performance and processing speed (Jiang et al. 2016) • Log mining: 1) provide search ranking and recommendation; 2) discover the semantic relationship among geospatial vocab through click-through and metadata to improve ranking and recommendation modelling. (Jiang et al. 2016) • Social media: 1) detect the human mobility with anomaly events; 2) detect unusual patterns and distinguish active/inactive individuals in archived datasets to improve data quality (Qin and Rice, 2016); 3) detect locations and impact areas of specific events (e.g. disasters, parades, or presidential election) • Utilize high performance computing (e.g. GPU and parallel computing) and cloud computing technologies to accelerate the knowledge discovering process
- Web Link
-
FBO.gov Permalink
(https://www.fbo.gov/notices/b76481d2078fb53695407b3c19d9bc4a)
- Place of Performance
- Address: George Mason University, 4400 University Drive, Fairfax, Virginia, 22030, United States
- Zip Code: 22030
- Zip Code: 22030
- Record
- SN04573413-W 20170712/170710235908-b76481d2078fb53695407b3c19d9bc4a (fbodaily.com)
- Source
-
FedBizOpps Link to This Notice
(may not be valid after Archive Date)
| FSG Index | This Issue's Index | Today's FBO Daily Index Page |