Google announces Simple ML machine learning add-on for Google Sheets

By

-

December 9, 2022

This week, Google LLC announced a brand-new tool that will let users create machine learning models in Google Sheets. Currently available as a beta, the tool is dubbed Simple ML. Users are able to get it for free as a Google Sheets add-on.

Simple ML was developed by one of the Google teams responsible for TensorFlow, a prominent open-source AI technology published by the search giant in 2015. Due to the extensible nature of Google Sheets, users can take advantage of add-ons that enhance the application’s built-in features. Users won’t need to use a specialized TensorFlow service in this instance because Google has designed Simple ML for Sheets to be as user-friendly as possible.

With just a few clicks, anyone, including those without programming or ML expertise, can experiment and apply some of the power of machine learning to their data in Google Sheets. It is important to note users must first create a Google Sheets spreadsheet with a collection of data points arranged in rows and columns in order to train AI models using Simple ML.

After configuring Simple ML for Sheets from the Google Workspace Marketplace, open the spreadsheet and run the add-on from the Extensions menu. Simple ML, which launches in a side panel, currently offers two AI use cases: “Predict missing values” and “Spot abnormal values.” You select the “Column with empty cells” in the first example and then hit the blue “Predict” button.

After a few seconds, the predictions and confidence percentages will be loaded into your spreadsheet. Google warns that it is possible that these statistical predictions are inaccurate. However, this may not be the case in future versions.

In the second use scenario, Simple ML can detect abnormal data items. This is possible since Simple ML creates no less than 10 AI models that automatically assess the accuracy of the data in a spreadsheet to identify data abnormalities. For instance, the add-on can assess whether a text string has been unintentionally entered into a spreadsheet field that is supposed to hold a number value.

A handful of advanced user features are also included in Simple ML. Google claims that the tool allows users to assess the quality of the AI models it creates and offers technical information about them. Additionally, Simple ML lets users transfer AI models to Colab, a Google cloud-based code editor that is used for projects involving machine learning and data science.

Simple ML may be used in conjunction with Connected Sheets, another powerful data processing tool from Google. It gives users access to the Google Sheets interface for data analysis hosted in the search engine giant’s BigQuery cloud data repository. With the help of the program, users can handle billions of spreadsheet data entries without having to create SQL queries.

Google also points out that because Simple ML runs in your browser, your data remains secure in your Google Sheets spreadsheet. For convenient sharing with the rest of your team, the models are instantly saved to Google Drive. Additionally, you can export models trained in Simple ML to the TensorFlow ecosystem because Simple ML is built on top of TensorFlow Decision Forests.

Meta to employ AI to verify user age on Facebook Dating

By

Preetipadma K

-

December 9, 2022

facebook dating ai age verification meta , meta facebook dating

In the past, Meta has employed AI facial scanning tools to confirm Instagram users’ ages. The company has now declared that it is testing the technology on Facebook Dating in order to make the product safer.

Recently, Meta stated in a blog post that it will begin requesting Facebook Dating users to verify their age if the platform suspected a user is underage. Users may upload a video selfie by following a few simple guidelines. Meta then shares a still picture from the video with Yoti, a third-party age-verification company, which calculates an age estimate based on facial features. The image is removed when Meta receives the result.

You could also upload an ID that shows your age, like a driving license. While your age is being confirmed, the ID is encrypted and saved. The ID is never made public on your Facebook page, and Meta then allows users to choose how long it should be kept. According to Meta, the verification procedure takes a few days, but its own help documentation reveals that it can store IDs for up to 1 year.

Following the company’s Cambridge Analytica data privacy breach, Facebook Dating debuted in 2019, however, it had fallen behind rivals in the dating industry. Meta has been experimenting with various artificial intelligence-based age identification systems as governments put more pressure on online companies to provide minimal protections to protect minors. Although Meta hasn’t completely described the markers it searches for to determine a person’s age, it has previously stated that it analyzes things like a user’s birthday postings since friends frequently respond with the user’s genuine age in their comments. With this most recent action, Meta is ahead of legislative frameworks that are increasingly centered on the online safety of kids and teenagers, with federal legislation underway both nationally and internationally.

According to Meta, testing since June has shown that the service was able to deter 96% of teenagers who attempted to modify their birthdays from doing so. The company said that the new age verification technologies will aid in preventing youngsters from using adult-only services. The age verification test will be implemented across all other Meta products that demand users to be at least 18 years old as well as in other nations where Facebook Dating is active.

For adults using Facebook Dating, Meta did not list any requirements (such as ensuring a 45-year-old is not pretending to be 18). The Verge claims that the technology is not equally accurate for all users: Yoti’s data reveals that the algorithm’s accuracy is worse for “feminine” faces and individuals with darker skin tones.

Howie Mandel Gets a Digital Twin from DeepBrain AI

By

Disha Chopra

-

December 9, 2022

howie mandel digital twin from deepbrain ai

Howie Mandel enters the metaverse as he gets a hyperrealistic digital twin from DeepBrain AI. Calling it “AI Howie,” DeepBrain AI developed it to be an interactive virtual human.

DeepBrain AI is a South Korean and California-based company providing virtual human services like recreating late family members’ personas in the virtual worlds. With the identical appearance, voice, gestures, and subtle mannerisms as the real person, these virtual characters are their digital twins.

The DeepBrain AI team also produced digital twins of South Korean president Yoon Suk-yeol, several news anchors from Asia, and Premier League soccer star Son Heung-Min before working with Howie Mandel.

Joe Murphy, business development manager at DeepBrain AI, said that the company also develops completely synthetic people called “digital people” and Roblox avatars. However, these technologies are not very advanced, and many companies can do it. But when it comes to the digital twins of real-life people, DeepBrain AI goes through an extensive process of deep learning to clone the voice, face, mannerisms, and even the way their eyes move.

The famous comedian, actor, host, and computer enthusiast Howie Mandel worked with DeepBrain AI to develop a virtual human AI digital twin. Mandel said in a statement, “I am equally thrilled, excited and terrified to finally have the ability of showing up and doing things without going anywhere or doing anything.”

AWS Machine Learning University to offer a free AI Educator enablement program

By

Sahil Pawar

-

December 9, 2022

Amazon Web Services (AWS) Machine Learning University is going to offer a free AI educator enablement program from January 2023 to help institutions prioritize teaching databases, machine learning, and artificial intelligence concepts to historically underserved students.

AWS will facilitate a free hands-on experience for students in a cloud-based sandbox to apply ML concepts and experiment with a variety of AWS services, ML cloud computing, and data analytics tools.

This program will prioritize minority-serving institutions (MSIs), community universities, and historically Black colleges and universities (HBCUs) in the US. The program will have six educator boot camps in 2023, using the same content Amazon uses to train their own developers and data scientists.

The boot camps will introduce learners to the program and will consist of lecture slides, exams, instructor handbooks, and hands-on coding exercises based on feedback from school systems that helped pilot the early program.

Educators who finish the program can get an AWS stipend and continuing education credits. They will also have year-round professional development through tech talks, Slack study groups, regional events, and virtual study sessions moderated by AWS instructors.

FTC tries to block Microsoft’s plan to buy Activation Blizzard

By

Sahil Pawar

-

December 9, 2022

FTC block Microsoft's plan buy Activation Blizzard

The Federal Trade Commission (FTC) has filed a legal case against Microsoft to try and block the company’s plan to buy Activision Blizzard for about $68.7 billion.

The lawsuit was filed today, according to a press release from the regulator, after weeks of back and forth between Sony, Microsoft, and other regulators over concerns about competition and the future of the game Call of Duty.

The FTC believes that the acquisition would allow Microsoft to suppress competitors to the company’s Xbox gaming consoles and its rapidly growing cloud-gaming business and subscription content.”

Today’s vote from the FTC commissioners will cause Microsoft to face significant hurdles in securing its Activision Blizzard deal. Even after Microsoft’s repeated attempts to appease regulators, regulators in the EU and UK are also scrutinizing the agreement closely.

“We continue to believe that the deal will expand competition and generate more opportunities for game developers and gamers,” Brad Smith, Microsoft’s vice chair and president. In a letter to Activision Blizzard employees, CEO Bobby Kotick assured the staff that he wanted to reinforce his confidence in the fact that the acquisition would close.

Thiruvananthapuram to be Kerala’s artificial intelligence hub

By

Sahil Pawar

-

December 9, 2022

Thiruvananthapuram Kerala's artificial intelligence hub

Thiruvananthapuram will be made Kerala’s artificial intelligence hub, the Chief Minister, Pinarayi Vijayan, said on Thursday. He said that various projects are being formulated under the leadership of the Kerala Startup Mission and higher education institutions.

On Thursday, he inaugurated a project for putting in place about 9,000 robotic kits in around 2,000 high schools in the public sector via their Little KITEs units.

The Chief Minister said the government had established a policy for promoting start-ups. However, an effective workforce is needed to realize it. He said that the government is implementing several schemes to link the higher education sector to the industry for achieving this objective.

The project will help children keep up with the changing world by upscaling their technical knowledge. He said they would get opportunities to become acquainted with novel technologies and gain practical knowledge.

Teachers will be provided special training to implement the robotics kits project. Around 60,000 students of the Little KITEs IT clubs will be trained under those teachers’ leadership. They, in turn, would prepare other students. The Chief Minister said that 12 lakh students would be provided the training.

“Interventions, right from the school level, are also needed. This is what prompted the formulation of the project to prepare schools with robotic kits,” the CM said.

Bangalore Traffic Police Use AI to Prevent and Detect Traffic Violation

By

Disha Chopra

-

December 9, 2022

bangalore traffic police use AI to prevent violation

In an attempt to automate traffic violation challans and fine payments, Bangalore traffic police use AI to prevent and detect traffic violations in the city. The Bangalore traffic police (BTP) department has installed over 8 red light cameras and 250 AI-integrated ANPR (automatic number-plate recognition) cameras at several junctions.

The camera-driven management system is a part of BTP’s plans to integrate technology into the police enforcement system under the plan Intelligent Traffic Management System (ITMS). The plan was initiated by CM Basavaraj Bommai to reduce human intervention in traffic management.

CM Basavaraj also announced a few more traffic police station sanctions in the coming week to expand jurisdiction and cover as many areas with AI-based monitoring stations as possible.

The cameras detect visible traffic violations and categorize them into 7 categories: speed violations, stop lane violations, red light violations, helmet, triple riding, no seat belts, and driving while using mobiles.

As police officers, the cameras are as accurate as 90%, and with sufficient data, the accuracy can be enhanced by up to 99%.

MA Saleem, Special Commissioner of Police (Traffic), said, “These cameras record violations round the clock, and the system is automated, which saves a lot of manpower that can be used for traffic management and regulation.”

DoD at the Pentagon Shares US$9b Cloud Computing Contracts with Four Companies: AWS, Microsoft, Oracle, and Google

By

Disha Chopra

-

December 8, 2022

The US Department of Defense announced its lucrative Pentagon cloud-computing contract with four industry-leading companies: AWS, Microsoft, Google, and Oracles. As per the contract, the Pentagon will split the total to give each as much as US$9b for developing a joint new cloud architecture called ‘Joint Warfighting Cloud Capability.’

The cloud architecture would provide the department with cloud services “across all security domains and classification levels.” The Pentagon cloud computing contract came three years after the previous US$10b cloud-computing contract given to Microsoft for JEDI (Joint Enterprise Defense Infrastructure).

The JEDI contract fell under a legal battle involving Donald Trump’s interference to favor Microsoft over Amazon. The DoD chose not to proceed with the JEDI contract saying that it “was developed at a time when the department’s needs were different and our cloud conversancy less mature.”

Instead, the Pentagon indicated that it would solicit numerous technology firms’ bids for the Joint Warfighting Cloud Capability.

According to market research, the leading companies to suit the requirements were Microsoft and Amazon, but officials said they would also contact IBM, Oracle, and Google. Dan Ives, a tech analyst at Wedbush Securities, said, “It’s good to finally end this chapter (Pentagon cloud computing contract) and get a cloud deal finally done for the Pentagon after years of a roller coaster.”

ByteDance AI introduces Magic Video, a text-to-video generation framework based on latent diffusion models

By

Sahil Pawar

-

December 8, 2022

ByteDance AI researchers have introduced ‘MagicVideo,’ which is an efficient framework for text-to-video generation based on latent diffusion models.

Magic Video generates videos in the latent space with the help of a pre-trained variational autoencoder, which enables significantly less computational requirement for MagicVideo.

MagicVideo makes use of 2D convolution instead of 3D convolutions to overcome getting video-text paired datasets. Temporal computation operators are used along with 2D convolution operations to process the spatial and temporal information present in the video. Moreover, using 2D convolutions allows MagicVideo to use pre-trained weights of text-to-image models.

Although switching from 3D to 2D convolution reduces the computational complexity significantly, the memory cost is still too much. Thus, MagicVideo shares equal weights for each of the 2D convolution operations.

However, doing so can reduce the generation quality since this approach assumes that all the frames are almost identical, although, in reality, the temporal difference is present. To overcome this, MagicVideo uses a custom lightweight adaptor module to modify the frame distribution.

MagicVideo learns the inter-frame relation with the help of a directed self-attention module. Frames are calculated on the basis of the previous ones, similar to the approach used in video encoding. Finally, produced video clips are enhanced using a post-processing module.

Giraffe360, a Real Estate VR Company, Raises $16m in Funding Led by Founder’s Fund

By

Disha Chopra

-

December 8, 2022

A real estate VR company called Giraffe360 raises over US$16m in a Series A funding led by Founder’s Fund, a marquee venture capitalist from San Fransico. Giraffe360 was founded in 2016 as a robotic technology startup to aid real estate agents in creating virtual listings and enhancing buyers’ experience.

Mikus Opelts, CEO of Giraffe360, said, “We are very excited to partner up with Founders Fund. It is one of the strongest brands in the VC industry, with a strong track record of backing category-defining companies.”

After the COVID-19 breakout in 2019, the company was able to surface as it raised US$4.5m in 2020 for its ability to deprecate the need for photographers in the virtual sight-showing process. Giraffe360 offers a robotic camera that enables brokers to generate high-resolution photos and virtual floor plans. The service is available on a subscription basis.

With the recent funding from Founder’s Fund, the startup plans to launch its latest and most upgraded version of the robot camera, Giraffe Go Cam. This new version is reportedly 30% lighter in weight, comes with faster charging, and has 500 GB of storage.

Giraffe360 secured an extra $6 million in long-term loans from the London-based venture lending provider Columbia Lake Partners in conjunction with the most recent equity round. To date, Giraffe360 raised US$22m in equity and US$9m in venture debt.