Homeland Security backs off on scanning US citizens, Amazon ups AI ante, and more

And why China might not be as big as first thought in AI spending

Roundup Hello, welcome to this week's machine learning musings. We bring you news about the hottest topics in AI: Facial recognition, the so-called AI arms race between the US and China, and erm, GPUs in the cloud.

US citizens won’t be subjected to facial biometric scans as they fly in and out of America, after all: The Department of Homeland Security has withdrawn a proposal asking for everyone - including US citizens - to pass through facial recognition cameras as they travel in and out of the country.

Documents filed to the Office of Information and Regulatory Affairs, a government bureau that reviews policies and regulations, revealed that the DHS wanted to expand its border security rights.

But after the DHS faced public backlash, including from Senator Edward Markey (D-MA) and the American Civil Liberties Union (ACLU), a non-profit group based in New York, it withdrew its decision.

“Customs Border Patrol is committed to keeping the public informed about our use of facial comparison technology,” John Wagner, deputy executive assistant commissioner for the Customs and Border Patrol field operations, said in a statement.

“We are implementing a biometric entry-exit system that protects the privacy of all travelers while making travel more secure and convenient.”

Under current rules, US citizens and permanent residents can refuse to have their faces scanned at airport terminals by talking to a CBP officer or an airline representative. Non-US citizens, however, do not have that choice and must go through a more rigorous process that involves having a photo taken and fingerprints scanned.

Senator Markey, however, said that extending the process to include American citizens was too risky. “This proposal would amount to disturbing government coercion, and as the recent data breach at Customs and Border Protection shows, Homeland Security cannot be trusted to keep our information safe and secure. I will soon introduce legislation to ensure that innocent American citizens are never forced to hand over their facial recognition information.”

Jay Stanley, a senior policy analyst, at the ACLU was more worried about the invasion of privacy.

“Travelers, including US citizens, should not have to submit to invasive biometric scans simply as a condition of exercising their constitutional right to travel. The government's insistence on hurtling forward with a large-scale deployment of this powerful surveillance technology raises profound privacy concerns,” he said in a statement sent to The Register.

AWS is fastest cloud provider if you want to train BERT or Mask R-CNN: Amazon claimed to have achieved “the world’s fastest model training times to date on the cloud” for BERT and Mask R-CNN, popular machine learning models used in natural language and computer vision at its re:Invent conference this week.

It takes just 69 minutes to train BERT written in PyTorch using 1,536 Nvidia V100 GPUs on 192 P3 instances and 62 minutes if you use TensorFlow with 2,048 V100 GPUs on 256 P3 instances. Of course it’s going to fast with that amount of hardware, and although that might beat other cloud platforms there’s probably not many customers willing to splash that much cash spinning up thousands of GPUs. The same goes for Mask R-CNN. You’ll have to fork out for 192 V100 GPUs to shorten the training time from hours to 26 minutes across PyTorch, TensorFlow, and MXNet.

“Over the past several months, AWS has significantly improved the underlying infrastructure, network, machine learning framework, and model code to achieve the best training time for these two popular state-of-the-art models,” it said this week.

In other AI-related re:Invent news, Amazon also announced it was releasing a physical keyboard to help tinkerers automatically compose music using machine learning, a speech-to-text transcriber for medical physicians, and a new cloud instance that uses the company’s own custom-built inference chip.

China isn’t spending as much as it says it is on AI R&D: An academic report from the Center for Security and Emerging Technology at Georgetown University disputed the common belief that China is pouring tens of billions of dollars into AI.

“We assess with low to moderate confidence that China’s public investment in AI R&D was on the order of a few billion dollars in 2018,” the report said. “With higher confidence, we assess that China’s government is not investing tens of billions of dollars annually in AI R&D, as some have suggested.”

China’s pledge to become the world leader in AI by 2030 has sparked the idea of an AI arms race between between the US and China. The efforts made by the Trump Administration to advance AI have often been viewed as lackluster. The US government is often criticized for not having a clear strategy and for not investing enough money and resources.

But the latest finding reveals that China isn’t spending as much money as people believe. “China’s spending in 2018 was on the same order of magnitude as U.S. planned spending for FY 2020,” it said.

Researchers studied public data from China’s Ministry of Finance and two of its biggest science and technology investment programs taken from 2018 to arrive at their conclusion. You can read the report in more detail here. ®

Broader topics

Other stories you might like

  • GPUs aren’t always your best bet, Twitter ML tests suggest
    Graphcore processor outperforms Nvidia rival in team's experiments

    GPUs are a powerful tool for machine-learning workloads, though they’re not necessarily the right tool for every AI job, according to Michael Bronstein, Twitter’s head of graph learning research.

    His team recently showed Graphcore’s AI hardware offered an “order of magnitude speedup when comparing a single IPU processor to an Nvidia A100 GPU,” in temporal graph network (TGN) models.

    “The choice of hardware for implementing Graph ML models is a crucial, yet often overlooked problem,” reads a joint article penned by Bronstein with Emanuele Rossi, an ML researcher at Twitter, and Daniel Justus, a researcher at Graphcore.

    Continue reading
  • US Copyright Office sued for denying AI model authorship of digital image
    What do we want? Robot rights! When do we want them? 01001110 01101111 01110111!

    The US Copyright Office and its director Shira Perlmutter have been sued for rejecting one man's request to register an AI model as the author of an image generated by the software.

    You guessed correct: Stephen Thaler is back. He said the digital artwork, depicting railway tracks and a tunnel in a wall surrounded by multi-colored, pixelated foliage, was produced by machine-learning software he developed. The author of the image, titled A Recent Entrance to Paradise, should be registered to his system, Creativity Machine, and he should be recognized as the owner of the copyrighted work, he argued.

    (Owner and author are two separate things, at least in US law: someone who creates material is the author, and they can let someone else own it.)

    Continue reading
  • AI chatbot trained on posts from web sewer 4chan behaved badly – just like human members
    Bot was booted for being bothersome

    A prankster researcher has trained an AI chatbot on over 134 million posts to notoriously freewheeling internet forum 4chan, then set it live on the site before it was swiftly banned.

    Yannic Kilcher, an AI researcher who posts some of his work to YouTube, called his creation "GPT-4chan" and described it as "the worst AI ever". He trained GPT-J 6B, an open source language model, on a dataset containing 3.5 years' worth of posts scraped from 4chan's imageboard. Kilcher then developed a chatbot that processed 4chan posts as inputs and generated text outputs, automatically commenting in numerous threads.

    Netizens quickly noticed a 4chan account was posting suspiciously frequently, and began speculating whether it was a bot.

    Continue reading
  • AMD touts big datacenter, AI ambitions in CPU-GPU roadmap
    Epyc future ahead, along with Instinct, Ryzen, Radeon and custom chip push

    After taking serious CPU market share from Intel over the last few years, AMD has revealed larger ambitions in AI, datacenters and other areas with an expanded roadmap of CPUs, GPUs and other kinds of chips for the near future.

    These ambitions were laid out at AMD's Financial Analyst Day 2022 event on Thursday, where it signaled intentions to become a tougher competitor for Intel, Nvidia and other chip companies with a renewed focus on building better and faster chips for servers and other devices, becoming a bigger player in AI, enabling applications with improved software, and making more custom silicon.  

    "These are where we think we can win in terms of differentiation," AMD CEO Lisa Su said in opening remarks at the event. "It's about compute technology leadership. It's about expanding datacenter leadership. It's about expanding our AI footprint. It's expanding our software capability. And then it's really bringing together a broader custom solutions effort because we think this is a growth area going forward."

    Continue reading

Biting the hand that feeds IT © 1998–2022