New Amazon tool simplifies delivery of containerized machine learning models

As part of the flurry of announcements coming this week out of AWS re:Invent, Amazon announced the release of Amazon SageMaker Operators for Kubernetes, a way for data scientists and developers to simplify training, tuning and deploying containerized machine learning models.

Packaging machine learning models in containers can help put them to work inside organizations faster, but getting there often requires a lot of extra management to make it all work. Amazon SageMaker Operators for Kubernetes is supposed to make it easier to run and manage those containers, the underlying infrastructure needed to run the models, and the workflows associated with all of it.

“While Kubernetes gives customers control and portability, running ML workloads on a Kubernetes cluster brings unique challenges. For example, the underlying infrastructure requires additional management such as optimizing for utilization, cost and performance; complying with appropriate security and regulatory requirements; and ensuring high availability and reliability,” AWS’ Aditya Bindal wrote in a blog post introducing the new feature.

When you combine that with the workflows associated with delivering a machine learning model inside an organization at scale, it becomes part of a much bigger delivery pipeline, one that is challenging to manage across departments and a variety of resource requirements.

This is precisely what Amazon SageMaker Operators for Kubernetes has been designed to help DevOps teams do. “Amazon SageMaker Operators for Kubernetes bridges this gap, and customers are now spared all the heavy lifting of integrating their Amazon SageMaker and Kubernetes workflows. Starting today, customers using Kubernetes can make a simple call to Amazon SageMaker, a modular and fully-managed service that makes it easier to build, train, and deploy machine learning (ML) models at scale,” Bindal wrote.

The promise of Kubernetes is that it can orchestrate the delivery of containers at the right moment, but if you haven’t automated delivery of the underlying infrastructure, you can over (or under) provision and not provide the correct amount of resources required to run the job. That’s where this new tool combined with SageMaker can help.

“With workflows in Amazon SageMaker, compute resources are pre-configured and optimized, only provisioned when requested, scaled as needed, and shut down automatically when jobs complete, offering near 100% utilization,” Bindal wrote.

Amazon SageMaker Operators for Kubernetes are available today in select AWS regions.

AWS releases SageMaker to make it easier to build and deploy machine learning models

from Amazon – TechCrunch https://techcrunch.com/2019/12/02/new-amazon-tool-simplifies-delivery-of-containerized-machine-learning-models/

AWS announces DeepComposer, a machine-learning keyboard for developers

Today, as AWS re:Invent begins, Amazon announced DeepComposer, a machine-learning driven keyboard aimed at developers.

“AWS DeepComposer is a 32-key, 2-octave keyboard designed for developers to get hands on with Generative AI, with either pretrained models or your own,” AWS’ Julien Simon wrote in a blog post introducing the company’s latest machine learning hardware.

The keyboard is supposed to help developers learn about machine learning in a fun way, and maybe create some music along the way. The area involved in generating creative works in artificial intelligence is called “generative AI.” In other words, it helps you teach machines to generate something creative using “generative adversarial networks.”

“Developers, regardless of their background in ML or music, can get started with Generative Adversarial Networks (GANs). This Generative AI technique pits two different neural networks against each other to produce new and original digital works based on sample inputs. With AWS DeepComposer, you can train and optimize GAN models to create original music,” according to Amazon.

AWS DeepComposer keyboard.

Developers can train their own machine learning models or use ones supplied by Amazon to get started. Either way, you create the music based on the model, tweak it in the DeepComposer console on the AWS cloud, then generate your music. If you wish, you can share your machine-generated composition on SoundCloud when you’re done.

This is the third machine-learning teaching device from Amazon, joining the DeepLens camera introduced in 2017 and the DeepRacer racing cars introduced last year. It’s worth noting that this is just an announcement. The device isn’t quite ready yet, but Amazon is allowing account holders to sign up for a preview when it is.

AWS expands DeepRacer league, announces car updates

from Amazon – TechCrunch https://techcrunch.com/2019/12/02/aws-announces-deepcomposer-a-machine-learning-keyboard-for-developers/

Amazon debuts automatic speech recognition service, Amazon Transcribe Medical

Amazon is expanding its automatic transcription service for AWS, Amazon Transcribe, to include support for medical speech, the company announced this morning at its AWS Re:Invent conference. The new machine-learning powered service, Amazon Transcribe Medical, will allow physicians to quickly dictate their clinical notes and speech into accurate text in real-time, without any human intervention, Amazon claims.

Unlike some services, the physicians won’t have to say things like “comma” or “full stop,” but can speak normally during the dictation process. The text can then be fed to downstream systems including ER systems or AWS language services, like Amazon Comprehend Medical for entity extraction.

The service is also HIPAA compliant and scales with the users’ needs, meaning you’ll only pay for what you actually use and without upfront fees, notes Amazon.

From a technical perspective, the service works as follows.

You first capture audio through a device’s microphone, then send PCM audio to a streaming API, based on the Websocket protocol. The API responds with a series of JSON blobs wit the transcribed text, plus word-level time stamps and punctuation. This can also be optionally saved to an Amazon Simple Storage Service (S3) bucket.

Amazon Transcribe Medical, which builds on 2017’s debut of Amazon Transcribe, arrives at a time when Amazon is increasing its investments in the medical space — particularly in terms of the intersection of voice technology with medicine.

Last week, for example, Amazon launched a medication management service for Alexa which allows consumers to make voice requests for refills and get medication reminders.

The company has also made it possible for Alexa voice apps to be HIPAA compliant, acquired health startups like PillPack and Health Navigator, launched its own healthcare service for employees, Amazon Health, and has been piloting the use of Alexa in a hospital environment.

Amazon is not alone in working with speech recognition in the healthcare space — this is an area Google is working in as well, with Google Brain, plus Microsoft, established players like Nuance and Philips, and a wide range of startups. 

Amazon Transcribe Medical is initially available in the U.S. East (North Virginia) and U.S. West (Oregon) regions.

from Amazon – TechCrunch https://techcrunch.com/2019/12/02/amazon-debuts-automatic-speech-recognition-service-amazon-transcribe-medical/

A bug in Microsoft’s login system put users at risk of account hijacks

Microsoft has fixed a vulnerability in its login system, which security researchers say could have been used to trick unsuspecting victims into giving over complete access to their online accounts.

The bug allowed attackers to quietly steal account tokens, which websites and apps use to grant users access to their accounts without having them to constantly re-enter their passwords. These tokens are created by an app or a website in place of a username and password after a user logs in. That keeps the user persistently logged into the site, but also allows users to access third-party apps and websites without having to directly hand over their passwords.

Researchers at Israeli cybersecurity company CyberArk found that Microsoft left open an accidental loophole which, if exploited, could’ve been used to siphon off these account tokens used to access that victim’s account — potentially without ever alerting the user.

CyberArk’s latest research, shared exclusively with TechCrunch, found dozens of unregistered subdomains connected to a handful of apps built by Microsoft. These in-house apps are highly trusted and as such, associated subdomains can be used to generate access tokens automatically without requiring any explicit consent from the user.

With the subdomains in hand, all an attacker would need is trick an unsuspecting victim into clicking on a specially crafted link in an email or on a website, and the token can be stolen.

In some cases, the researchers said, this could be done in a “zero-click” way, which as the name suggests requires almost no user interaction at all. A malicious website hiding an embedded webpage could silently trigger the same request as a link in a malicious email to steal a user’s account token.

Luckily, the researchers registered as many of the subdomains they could find from the vulnerable Microsoft apps to prevent any malicious misuse, but warned there could be more.

The security flaw was reported to Microsoft in late October and was fixed three weeks later.

“We resolved the issue with the applications mentioned in this report in November and customers remain protected,” said a Microsoft spokesperson.

It’s not the first time Microsoft has acted to fix a bug in its login system. Almost exactly a year ago, the software and services giant fixed a similar vulnerability in which researchers were allowed to alter the records of an improperly configured Microsoft subdomain and steal Office account tokens.

Read more:

• A bug in Microsoft’s login system made it easy to hijack anyone’s Office account
• StockX was hacked, exposing millions of customers’ data
• DoorDash confirms data breach affected 4.9 million customers, workers and merchants
• Equifax breach was ‘entirely preventable’ had it used basic security measures, says House report
• Stop saying, ‘We take your privacy and security seriously’
• Capital One breach also hit other major companies, say researchers
• Macy’s said hackers stole customer credit cards — again

from Microsoft – TechCrunch https://techcrunch.com/2019/12/02/microsoft-login-flaw-account-hijack/

Facebook launches a photo portability tool, starting in Ireland

It’s not friend portability, but Facebook has announced the launch today of a photo transfer tool to enable users of its social network to port their photos directly to Google’s photo storage service, via encrypted transfer.

The photo portability feature is initially being offered to Facebook users in Ireland, where the company’s international HQ is based. Facebook says it is still testing and tweaking the feature based on feedback but slates “worldwide availability” as coming in the first half of 2020.

It also suggests porting to other photo storage services will be supported in the future, in addition to Google Photos — which specifying which services it may seek to add.

Facebook says the tool is based on code developed via its participation in the Data Transfer Project — a collaborative effort started last year that’s currently backed by five tech giants (Apple, Facebook, Google, Microsoft and Twitter) who have committed to build “a common framework with open-source code that can connect any two online service providers, enabling a seamless, direct, user initiated portability of data between the two platforms”.

Facebook also points to a white paper it published in September — where it advocates for “clear rules” to govern the types of data that should be portable and “who is responsible for protecting that data as it moves to different providers”.

Behind all these moves is of course the looming threat of antitrust regulation, with legislators and agencies on both sides of the Atlantic now closely eyeing platforms’ grip on markets, eyeballs and data.

Hence Facebook’s white paper couching portability tools as “helping keep competition vibrant among online services”. (Albeit, if the ‘choice’ being offered is to pick another tech giant to get your data that’s not exactly going to reboot the competitive landscape.)

It’s certainly true that portability of user uploaded data can be helpful in encouraging people to feel they can move from a dominant service.

However it is also something of a smokescreen — especially when A) the platform in question is a social network like Facebook (because it’s people who keep other people stuck to these types of services); and B) the value derived from the data is retained by the platform regardless of whether the photos themselves travel elsewhere.

Facebook processes user uploaded data such as photos to gain personal insights to profile users for ad targeting purposes. So even if you send your photos elsewhere that doesn’t diminish what Facebook has already learned about you, having processed your selfies, groupies, baby photos, pet shots and so on. (It has also designed the portability tool to send a copy of the data; ergo, Facebook still retains your photos unless you take additional action — such as deleting your account.)

The company does not offer users any controls (portability tools or access rights) over the inferences it makes based on personal data such as photos.

Or indeed control over insights it gleans from its analysis of people’s usage of its platform or wider browsing of the Internet (Facebook tracks both users and non users across the web via tools like social plug-ins and tracking pixels).

Given its targeted ads business is powered by a vast outgrowth of tracking (aka personal data processing), there’s little risk to Facebook to offer a portability feature buried in a sub-menu somewhere that lets a few in-the-know users click to send a copy of their photos to another tech giant.

Indeed, it may hope to benefit from similar incoming ports from other platforms in future.

“We hope this product can help advance conversations on the privacy questions we identified in our white paper,” Facebook writes. “We know we can’t do this alone, so we encourage other companies to join the Data Transfer Project to expand options for people and continue to push data portability innovation forward.”

Competition regulators looking to reboot digital markets will need to dig beneath the surface of such self-serving initiatives if they are to alight on a meaningful method of reining in platform power.

from Microsoft – TechCrunch https://techcrunch.com/2019/12/02/facebook-launches-a-photo-portability-tool-starting-in-ireland/

Deep connection

When someone tells you what you need to hear, instead of what you’re hoping to hear, you’ve found something priceless.

This takes care, generosity and guts to achieve.

When you offer this gift to someone else, it might seem like it’s unappreciated. But you didn’t do it to be appreciated, you did it because you care enough to work for a deep connection, one that makes things better.

Best to devote that energy to people and causes that can run with it.

       

from Seth Godin's Blog on marketing, tribes and respect https://feeds.feedblitz.com/~/610689894/0/sethsblog~Deep-connection/

The attention crisis is real

It’s all happening at once, and we have to choose.

What to read, listen to, answer?

A spam from a Nigerian prince

An @ mention on Slack

A voice mail from the boss

An email from a customer

A DM on Slack

A last-minute sale email from a store you’ve never visited

A year-old blog post

A new blog post

A new podcast

A …

 

[sorry, got distracted.]

The idea that we can strip mine attention, wasting what we don’t need, is long gone. Like oysters and oil before it, attention is a scarce resource, and we need to use it wisely. Too often, it feels cheaper to simply take what we can get, but when we overreach, the cost in trust is real.

And each of us gets the same amount of attention to spend each day. It’s a competitive advantage to figure out how to focus it to get something done.

       

from Seth Godin's Blog on marketing, tribes and respect https://feeds.feedblitz.com/~/610606712/0/sethsblog~The-attention-crisis-is-real/