ML News Monthly – June 2021

Welcome to the eighth edition of ML News Monthly – Jun 2021!!

Here are the key happenings this month in the Machine Learning field that I think are worth knowing about. 🕸


AI Politics

US state, Maine’s legislature passed one of the strictest state laws governing facial recognition in the U.S

The law, which takes effect later this year, bars police from using the techology unless they have probable cause that a suspect committed a serious crime.

  • Under the law, police will no longer have direct access to facial recognition services, which can match images of a suspect to photos in a database.
  • Rather, police will have to ask the FBI or Maine Bureau of Motor Vehicles (BMV) to conduct the biometric searches.

Read more: https://slate.com/technology/2021/07/maine-facial-recognition-government-use-law.html

Experts Doubt Ethical AI Design Will Be Broadly Adopted as the Norm Within the Next Decade

A survey by Pew Research Center and Elon University asked 602 software developers, business leaders, policymakers, researchers, and activists: “By 2030, will most of the AI systems being used by organisations of all sorts employ ethical principles focused primarily on the public good?” 68 percent said no.

They also cite the difficulty of achieving consensus about ethics. Many who expect progress say it is not likely within the next decade.

Read More : https://www.pewresearch.org/internet/2021/06/16/experts-doubt-ethical-ai-design-will-be-broadly-adopted-as-the-norm-within-the-next-decade/

Related:

US military – DOD Adopts Ethical Principles for Artificial Intelligence – https://www.defense.gov/Newsroom/Releases/Release/Article/2091996/dod-adopts-ethical-principles-for-artificial-intelligence/

EU – Ethics guidelines for trustworthy AI – https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai

Microsoft -Responsible AI – https://www.microsoft.com/en-us/ai/responsible-ai

Google – Artificial Intelligence at Google:Our Principles – https://ai.google/principles/

It’s 2021. Do You Know What Your AI Is Doing?

New “State of Responsible AI” report from Corinium and FICO finds that most companies don’t—and are deploying artificial intelligence at significant risk

Read More: https://www.fico.com/blogs/its-2021-do-you-know-what-your-ai-doing

AI in the News

How has AI Contributed to Dealing with the COVID-19 Pandemic

A new report assessed how AI has helped address Covid-19 and where it has fallen short.

The past year has been an annus horribilis in many ways. However, the past year has also been a year of astonishing scientific accomplishments. Numerous research groups, in combination with industry and government, have reoriented their resources to help contribute in the quest to contain the pandemic. The daily number of articles uploaded to arXiv that mention COVID-19 and their cumulative sum is shown below. It gives an idea of how people around the world, including the AI community, contributed to tackling the pandemic…

Read more: https://thegradient.pub/how-has-ai-contributed-to-dealing-with-the-covid-19-pandemic/

Five Industries Reaping The Benefits Of Artificial Intelligence

Read more: https://www.forbes.com/sites/forbestechcouncil/2021/06/02/five-industries-reaping-the-benefits-of-artificial-intelligence/

AI Applications

GitHub and OpenAI release Copilot: an AI pair programmer

Copilot autocompletes code snippets, suggests new lines of code, and can even write whole functions based on the description provided. According to the GitHub blog, the tool is not just a language-generating algorithm based on user input — it is a virtual pair programmer.

Read More :

https://copilot.github.com

Related

Will Artificial Intelligence Replace Developers? – https://catalins.tech/github-copilot-will-artificial-intelligence-replace-developers

‘Robot Manicure’ Impresses Users As Creators Insist Jobs Aren’t at Risk

How it works: Users select a color and place a hand or finger into a slot in a toaster-sized machine. The system scans the fingertips, and an automated paint dispenser — in some cases, a mechanical arm tipped by a brush — coats each nail. These machines update earlier nail-decorating gadgets that, say, applied decals without using AI.

  • Clockwork aims to install its machines in offices and retail stores. The company recently opened a storefront in San Francisco.
  • Nimble and Coral aim their devices at home users.

Read more: https://www.newsweek.com/robot-manicure-impresses-users-creators-insist-jobs-arent-risk-1597517

Sonoma County, California experiments with artificial intelligence to spot wildfires

The software is developed by a South Korean company called Alchera, which uses similar AI technology in Asia, to monitor everything from airports to major industrial systems and utilities. In this case, they taught the algorithm to quickly recognize smoke.

Read more:

https://abc7news.com/wildfire-ai-artificial-intelligence-sonoma-county/10763475/

http://alcherainc.com/en/

From admissions to teaching to grading, AI is infiltrating higher education

In a bid to cut costs, many schools are adopting chatbots, personality-assessment tools, and tutoring systems

New map created by AI reveals hidden links between Milky Way and Andromeda galaxies

A new cosmic map revealed hidden structures connecting galaxies, which could help scientists model a future collision between the Milky Way and Andromeda, our galaxy’s nearest neighbor.

Read more: https://www.space.com/artificial-intelligence-models-milky-way-andromeda-collision

AI Research

[Paper] assessing the test error of a neural network on unlabeled data

Authors empirically show that the test error of deep networks can be estimated by simply training the same architecture on the same training set but with a different run of Stochastic Gradient Descent (SGD), and measuring the disagreement rate between the two networks on unlabeled test data.

Read More : https://arxiv.org/abs/2106.13799

[Paper] Learning a Universal Template for Few-shot Dataset Generalization

Getting high accuracy out of a classifier trained on a small number of examples is tricky. You might train the model on several large-scale datasets prior to few-shot training, but what if the few-shot dataset includes novel classes.

Team at Google and Vector Institute, along with colleagues at both organizations, designed Few-shot Learning with a Universal Template (FLUTE). Training some layers on several tasks while training others on only one reduces the number of parameters that need to be trained for a new task. Since fewer parameters need training, the network can achieve better performance with fewer training examples.

Read More : https://arxiv.org/abs/2105.07029


That’s it !!

Let me know if I missed anything or if there’s anything you think should be included in a future post.