Supercharging user experience with machine learning
summary
Periscope is an open source smart search that pushes content and actions closer to users: ResNet-50 pretrained on ImageNet indexes searchable images (millions of labeled images), voice input uses the browser Web Speech API (Chrome/Firefox) to return transcripts, and an adaptive ranking neural net (128 ASCII code points × max 15 chars input) reorders results on the fly with a high learning rate so selections move from fourth to first after one interaction; Magnolia integration and pluggable result suppliers included.
AI-written summary · about
chapters (11)
- 0:00Start
- 2:05Pluggable results supplier API
- 3:03Searchable Images
- 3:59ImageNet database
- 4:33Magnolia Implementation
- 4:57Speech Recognition
- 5:16Web Speech API
- 6:17Adaptive result order
- 7:15Initial Convolution
- 7:43Output Layer
- 8:17Setup
related talks
26mAdvancing Multi-Modal Search Capabilities in Search Pipeline
40mVespa.ai’s Personalized Search: Advanced Ranking & Tensor framework
10mForsense: Accelerating online research through sensemaking integration and machine research support.
30mSemantic Search and Neural Information Retrieval
46mNatural Language & Search, Improve your Search w/ Vector capabilities
25mSemantic Search
39mMachine Learning for Relevance and Serendipity
55mDive into the Future: .NET and AI!