Trustworthy Machine Learning
When digital meetings are becoming the norm, it is of utter most importance that our video service are working in a reliable and fair manner despite the users ethnicity, persona and environment.
Help us improve our service by working on some of the key challenges below, using our conjured data.
- Background blur – regardless of complexion or color of skin
- Video meeting – every voice character is heard
- Voice assistant – every voice type understood
We are looking for a research person / group that can take a leading role in this, with the assistance from our engineers and data team in Sweden.
Data Set description:
- 102 hours of annotated video meeting data with 54 classes.
- 67 hours of 1542 unique voices with descriptive persona and recording environment.
Expected Field of Expertise:
- Algorithm development.
- The Startup get improved products and you and your research group get to publish method and result.
Challenge contact owner
If you are interested in getting in contact to discuss this further, send an email to firstname.lastname@example.org and answer the following questions:
- Name and organisation (if applicable).
- Previous solutions and resarch (if applicable) (eg. link to academic page, Linkedin or repo).
- Describe why you are a great fit.