Local AI will help develop new applications

[zakiyatasnim]

With the help of miniature artificial intelligence, existing services, such as voice assistants and auto-correction, as well as computational photography technologies, will become better and faster without the need to turn to the cloud to run a deep learning model.

such as mobile analysis of medical images, improve the responsiveness of self-driving cars, and provide greater data privacy because data does not have to be sent to cloud storage for processing.

But it could also lead to new problems, such as making it harder luxembourg number data to combat fake videos and surveillance systems. Researchers, engineers, and regulators will need to work together to develop measures to prevent potential harm.

Differential privacy
Using a method that adds random data when collecting users' confidential information to protect them.

Why it matters: The U.S. Census Bureau is finding it increasingly difficult to keep the data it collects confidential. The method could solve the problem, improve public trust, and set an example for other countries.
Key players: US Census Bureau, Apple, Facebook.
Availability: The largest application is in the 2020 US Census.
The 2020 U.S. census will task the government with collecting data on 330 million residents, keeping their identities secret. The data is published in statistical tables that scientists and policymakers analyze and take into account when developing laws or conducting research.

By law, the Census Bureau must keep information about respondents confidential, but there are ways to de-anonymize individuals, especially when census data is combined with other publicly available statistics.

So the Bureau introduces “noise” into the data, for example, changing a person’s age or skin color, while maintaining the general indicators of the age or ethnic group. The more “noise”, the more difficult it is to de-anonymize.

But too much noise can make data useless: for example, in the 2010 census, after using “noise,” households with 90 people appeared.

So the Bureau is using differential privacy, a mathematical method that makes the process more precise by measuring how much privacy increases when noise is added. It is already used by Google and Facebook to collect data without identifying specific users.

If the census is successful and the data remains secure, differential privacy will be used by other federal agencies, and some countries, such as Canada and the United Kingdom, are considering implementing it.

Justification of climate change
Scientists can now determine the role of climate change in extreme weather conditions.

Why it's important: This will provide a more accurate picture of how climate change is worsening weather, and what needs to be done to prepare for dangerous weather events.
Key players: World Weather Attribution, Netherlands Meteorological Institute, Red Cross Red Crescent Climate Centre, University of Oxford.
Availability: now.