ADA Lab @ UCSD

 

Project Genisys

Overview

Genisys is a new kind of data system that enables ADA applications to easily deploy ML models in environments ranging from the cloud to personal devices. Genisys exploits deep learning-based ML models to see, hear, and understand unstructured data and query sources such as speech, images, video, time series, and text. We call this vision of type-agnostic data analytics database perception. Watch this space for more details.

Active Component Projects

 

Krypton
Enabling fast interactive diagnosis of the internals of visual perception systems.

 

Panorama
Enabling unbounded vocabulary querying over video.

 

SpeakQL
Enabling speech-driven multimodal querying of structured data with regular SQL and more.

 

Vista
Enabling data systems to truly see image and video data for efficient multimodal analytics.

Publications

  • Query Optimization for Faster Deep CNN Explanations
    Supun Nakandala, Arun Kumar, and Yannis Papakonstantinou
    ACM SIGMOD Record 2020 | Paper PDF | Code and Data coming soon
    ACM SIGMOD Research Highlights Award

  • Incremental and Approximate Computations for Accelerating Deep CNN Inference
    Supun Nakandala, Kabir Nagrecha, Arun Kumar, and Yannis Papakonstantinou
    ACM TODS 2020 | Paper PDF | Code and Data coming soon
    Invited Paper

  • Incremental and Approximate Inference for Faster Occlusion-based Deep CNN Explanations
    Supun Nakandala, Arun Kumar, and Yannis Papakonstantinou
    ACM SIGMOD 2019 | Paper PDF | TechReport | Blog post | Talk Video | Code coming soon
    Honorable Mention for Best Paper Award

  • Demonstration of SpeakQL: Speech-driven Multimodal Querying of Structured Data
    Vraj Shah, Side Li, Kevin Yang, Arun Kumar, and Lawrence Saul
    ACM SIGMOD 2019 Demo | Paper PDF | Video

  • Demonstration of Krypton: Optimized CNN Inference for Occlusion-based Deep CNN Explanations
    Allen Ordookhanians, Xin Li, Supun Nakandala, and Arun Kumar
    VLDB 2019 | Paper PDF | Video | Code coming soon

  • Demonstration of Krypton: Incremental and Approximate Inference for Faster Occlusion-based Deep CNN Explanations
    Supun Nakandala, Arun Kumar, and Yannis Papakonstantinou
    SysML 2019 Demo | Paper PDF | Video

  • Materialization Trade-offs for Feature Transfer from Deep CNNs for Multimodal Data Analytics
    Supun Nakandala and Arun Kumar
    SysML 2018 (Short paper/poster) | Paper PDF

  • SpeakQL: Towards Speech-driven Multi-modal Querying
    Dharmil Chandarana, Vraj Shah, Arun Kumar, and Lawrence Saul
    ACM SIGMOD 2017 HILDA Workshop | Paper PDF