[INT-2]USA Showcase Keynote Session, Cutting edge US technology: AI / IoT Collaboration with TECH CRUNCH and EN GADGET
10/05 13:00-15:00 Simultaneous Japanese-English Interpretation available Convention Hall B , International Conference Hall
- US Embassy, Japan Commercial Attache Ms. Brittany Banta
- Google Japan Inc.Head of Public Policy and Government Relations Mr. Yoshitaka Sugihara
《Talk Session #1》
- DataRobot JapanChief Data Scientist Akira Shibata, PhD
- TECH CRUNCHEditor in chief Mr. Ken Nishimura
《Talk Session #2》The World as Seen Through AI-integrated Smart Glasses
- Recently, various Ai-equipped, voice-enabled devices have been introduced to the market and some vendor’s strategy is to have one such device in each room (of a user’s house or building). When smart glasses that combine technology and fashion and that can be worn every day, such as Vuzix is proposing with the Blade series, we believe that they will become so commonplace that everyone will be using them.
The future of UI: from manual to voice control.
- Vuzix Corporation Director of Operations Japan Mr. Keiichiro Fujii
- EngadgetEditor in chief Mr. Asuka Yazawa
《Presentation》How solving the data problem is changing the future of AI
- Fortune 500 companies are looking for ways to implement Artificial Intelligence to reduce operating costs and increase market share. However, with this rapid evolution, the expectation of what AI can do is set high, often to unrealistic levels.
A big part of the problem has to do with data quality, type and size. In the last 10 years, a few technology shifts happened that are the genesis of the AI revolution: 1) Machine learning switched from rule based to statistically-driven models, 2) Cloud computing emerged allowing unlimited computing power and 3) Connectivity has enabled people to work together on demand, anywhere, online.
Machine learning models now require high quality training data at scale. Obtaining high-quality training data is a complex, multidisciplinary process. The procedure is time consuming and often tedious, forcing data scientists to spend 80% of their time cleaning and structuring data. Deprioritizing data quality, however, ends up compromising the model quality by leading to poor results and high dissatisfaction by users. Bad quality data leads to terrible customer experiences, which can even sometimes taint the brand of a company. So where exactly should a data scientist invest time & resources? Building models or handling data?
In this talk, we will show how data scientists can build better models by accessing higher quality training data and use a platform that combines built-in data workflows, people-in-the-loop and machine learning.
- DefinedCrowdFounder and CEO Dr. Daniela Braga