2018 04 19 NLP meetup - WordPress.com · @adrien_ball, Senior Machine Learning Engineer. Wakeword...
Transcript of 2018 04 19 NLP meetup - WordPress.com · @adrien_ball, Senior Machine Learning Engineer. Wakeword...
Snips NLU An Open Source, Private by Design alternative to NLU cloud services
@adrien_ball, Senior Machine Learning Engineer
Wakeword Speech to text Natural Language Understanding
Speech to text Natural Language Understanding
“What will be the weather in Paris
tomorrow evening?”
{ "intent": { "intentName": "searchWeatherForecast", "probability": 0.95 }, "slots": [ { "value": “Paris", "entity": "locality", "slotName": "forecastLocality" }, { "value": { "kind": "InstantTime", "value": "2018-04-19 19:00:00 +00:00" }, "entity": "snips/datetime", "slotName": "forecastDatetime" } ] }
Natural Language Understanding
Why Snips NLU?
Data Privacy
Q. Are voice and text inputs processed by Amazon Lex stored, and how are they used by AWS?
Amazon Lex may store and use voice and text inputs processed by the service solely to provide and maintain the service and to improve and develop the quality of Amazon Lex and other Amazon machine-learning/artificial-intelligence technologies. Use of your content is necessary for continuous improvement of your Amazon Lex customer experience, including the development and training of related technologies.(…)
https://aws.amazon.com/lex/faqs/#data-privacy
AI logic developper code
NLU cloud
NLU cloud
AI logic developper code
NLU cloud as backendNLU cloud as frontend
NLU CLOUD SOLUTIONS + ALTERNATIVES
AI logic developper code
On premises NLU
AI logic developper code
NLU cloud
NLU cloud
AI logic developper code
NLU cloud as backendNLU cloud as frontend
NLU CLOUD SOLUTIONS + ALTERNATIVES
AI logic developper code
On premises NLU Embedded NLU
snips-nluPython library
Training + Inference
Apache 2
snips-nlu-rs Snips consoleRust library
Inference
Apache 2 / MIT
https://console.snips.ai
Training
Self service / Professional services
Snips NLU ecosystem
• deterministic behavior
• great generalization power
Objectives & Constraints
• rapid prototyping
• fast & lightweight
• portability
• extensibility
snips-nlu• Rapid prototyping
• Rapid debugging
• ML ecosystem
snips-nlu-rs
• Speed
• Portability
Python / Rust complementarity
snips-nlu• Rapid prototyping
• Rapid debugging
• ML ecosystem
snips-nlu-rs
• Speed
• Portability
Python / Rust complementarity
JSON serialization
🤔 Speed?
Accuracy?
Snips NLU inference time
🤔 Speed?
🤔 Accuracy?
Extension of Evaluating Natural Language Understanding Services for Conversational Question Answering Systems
NLU cloud frontend
NLU cloud backend
On premises NLU
Embedded NLU
Privacy
Speed 2 roundtrips 2 roundtrips 1 roundtrip Few milliseconds
Accuracy State of the art State of the art State of the art State of the art
https://github.com/snipsco/snips-nlu
Willing to integrate Snips NLU?
Want to join the Snips team?
Questions?
[email protected]@adrien_ball
https://medium.com/snips-ai