Follow Datanami:
March 4, 2020

On the Radar: COVID-19, Circonus, and Fake FB Accounts

Alex Woodie

(Oleksii Lishchyshyn/Shutterstock)

Welcome to On the Radar, a new Datanami feature about interesting startups and news in the big data space. In this week’s edition, we check out how AI is driving the coronavirus response, the machine data intelligence firm Circonus, and the open sourcing of OpenDS4All.

As the new coronavirus continues to spread around the world, researchers are seeking novel methods of detecting the disease, including the use of AI technologies to interpret sensor data from people who may have been infected.

Baidu has reportedly developed a method of interpreting data from thermal scanners to determine if a person has a fever, which is one sign of COVID-19 infection. The system, which is said to be accurate to within half a degree centigrade, has been implemented in a busy Beijing railway station, where it can scan thousands of people per hour, according to a story in Venture Beat.

Meanwhile, another Chinese tech giant, Alibaba, has developed a computer vision system that can detect COVID-19 infection by analyzing CT scans of people’s lungs. According to Nikkei Asian Review, the system was trained on CT scans of 5,000 confirmed cases of coronavirus, and is able to diagnose the presence or absence of the disease in about 20 seconds with 96% accuracy. By comparison, it would typically take a doctor five to 15 minutes to make a clinical diagnosis.

A CT scan of the lungs of a patient affected with COVID-19 (Image courtesy Radiology)

The Chinese government’s expansive surveillance system is also being employed to help stop the spread of the virus. According to a story in Al Jazeera, it’s using its network of facial recognition-equipped cameras and national ID cards to track the movement of individuals suspected of having the disease, in some cases calling people to remind them to return home to self-quarantine. Its extensive monitoring system is also coming in handy with contact tracing, as the government has rolled out a new mobile app called “Close Contact Detector” that lets users determine if they have come in close proximity to somebody infected with the disease.

Circonus Taps Bob Moul

The machine data intelligence company Circonus has landed on our radar. The Maryland firm specializes in collecting large amounts of data from computer logs, applications, and network devices, and then analyzing the data to assist IT professionals with the management of their systems.

The Circonus Platform brings a host of advanced analytics, visualizations, and alerts to the table, to help IT professionals stay on top of today’s complex, containerized, and hybrid IT architectures. The company boasts that its underlying time-series database, called IRONdb, can scale to 128 nodes, giving customers as much scalability as they need.

The Circonus Platform (Courtesy Circonus)

Circonus was founded in 2010, but just launched its Circonus Platform last October. The company, which counts companies like Joyent, Venmo, and Major League Baseball as customers, also recently brought on serial entrepreneur Bob Moul as CEO in January.

It also raised a $6.8M Series A1 investment led by Osage Venture Partners to accelerate its sales and marketing efforts.

Facebook Detects Fake Accounts

It may not come as news that Facebook is a frequent hangout spot for fraudsters, charlatans, and con-men, who use the site to spread span, malware, and launch phishing attacks. But now the social media giant is using AI to clamp down on the ne’er-do-wells.

According to a recent story in MIT Technology Review, Facebook uses a combination of manually written rules and machine learning to detect fake accounts before they go active. Its deep learning system is able to detect a number of different fake account types, including account takeovers, spammers, phisher-men, and “illegitimate accounts not representative of the person,” the story states.

The goal is to step up identification and elimination of fake accounts this year, which is an election year in the US. The company, which eliminated 2 million fake accounts in 2019, estimates there are still 125 million fake accounts on the site, which accounts for about 5% of the 2.5 billion accounts on the site.

Related Items:

On the Radar: Promethium, StreamNative, Inzata

On the Radar: Cortex Labs, RadicalBit, Qeexo

 

Do NOT follow this link or you will be banned from the site!