Jason Mars

TL;DR

Researcher Jason Mars discusses the creation of Serious/Lucida, an open-source voice and vision personal assistant, and how it led to the formation of Clarity Lab Inc.

Transcript

hi so I'm Jason Mars I think the first do is to introduce myself and the research lab that we run at University of Michigan so I'm Jason Mars this is my third year as an assistant professor at Michigan I co run the clarity lab in Michigan with Ling JA tang she's in the audience wave awesome and we have a great team of very very enthusiastic high-en... Read More

Key Insights

• 😮 Intelligent personal assistants are changing the way we interact with technology, and with the rise of wearables, it is crucial to develop natural and intuitive interfaces.
• ❤️‍🩹 Building an end-to-end voice and vision personal assistant like Serious/Lucida presents challenges due to the monopolization of infrastructure by companies like Apple and Google.
• 😃 Serious/Lucida aims to democratize the technology found in big companies by making it available to everyone, opening up opportunities for new companies and applications.
• 🤩 Deep learning is a key focus for companies like Google and Facebook, as it offers superior solutions for many problems and tasks.

Questions & Answers

Q: What are some examples of queries that Serious/Lucida can handle?

Serious can handle voice commands, voice queries, and combined voice and vision queries. For example, you can give a command to your mobile device or object, ask a question like "Who is the lead singer of YouTube?", or snap a picture of a bar and ask, "When does this bar close?".

Q: How did the team at Clarity Lab gather information about the algorithms used by companies like Google and Facebook?

The team did investigative research by talking to leaders of important teams at these companies. They learned about the algorithms used for speech recognition, computer vision, and question-and-answer systems, and incorporated that knowledge into Serious/Lucida.

Q: What is the scalability gap that Mars refers to?

The scalability gap refers to the increase in computation required when all search queries are transformed into intelligent personal assistant queries. Mars explains that this scalability gap would require data centers to grow by orders of magnitude, which is not sustainable.

Q: How did porting the algorithms used in Serious/Lucida to GPUs and FPGAs contribute to reducing the scalability gap?

Mars and his team found that porting the algorithms to GPU and FPGA platforms resulted in significant speed ups. By leveraging these technologies, the scalability gap can be brought down, potentially providing a 10x improvement and making the future of intelligent personal assistants more feasible.

Summary & Key Takeaways

• Jason Mars introduces himself and the research lab he runs at the University of Michigan, highlighting their productivity and success in the field of AI.

• He talks about the rise of intelligent personal assistants like Siri and Cortana and the increasing importance of natural and intuitive ways to interact with technology.

• Mars explains how the team designed Serious/Lucida to be an end-to-end voice and vision personal assistant and walks through the different types of queries it can support.