Categories
Tech News

MIT programmable fiber can infer physical activity

Researchers at MIT have created the first fiber that has digital capabilities. The fiber can sense, store, analyze, and infer activity and be sewn into his shirt. MIT professor Yoel Fink says the digital fiber has expanded the possibilities for fabrics to uncover hidden patterns in the human body that could potentially be used for physical performance monitoring, medical inference, and early disease detection.

Until the MIT breakthrough, electronic fibers have been analog, carrying a continuous electric signal rather than digital where bits of information can be encoded and processed within the fiber. The new work is the first realization of a fabric with the ability to store and process data digitally. Fink says the new fiber allows fabrics to be programmed.

Researchers created the fiber by placing hundreds of square silicon microscale digital chips into a preform to create a polymer fiber. By controlling the polymer flow precisely, the researchers created a fiber providing a continuous electrical connection between the chips over tens of meters. The resulting fiber is thin and flexible, allowing it to be passed through a needle to be sewn into fabrics.

The digital fibers are also robust and can be washed at least ten times without breaking down. The fiber is also thin enough and comfortable enough that it can’t be felt by the wearer when woven into a shirt. Data storage is also possible within the fiber. The researchers can write, store, and read information on the fiber, including a 767-kilobit full-color movie file and a 0.48-megabyte music file. Files can be stored within the fiber for up to two months without power.

The new fiber also has potential for medical use and, in testing, was integrated into the armpit of a shirt and used to collect 270 minutes of surface body temperature data. The team confirmed with 96 percent accuracy what activity the person wearing the shirt was engaged in when it was recorded. A small external device currently controls the fiber, and researchers are designing a new chip as a microcontroller that can be connected within the fiber.

Repost: Original Source and Author Link

Categories
AI

AI Weekly: MIT aims to reconcile data sharing with EU AI regulations

Join Transform 2021 this July 12-16. Register for the AI event of the year.


This week, the European Union (EU) unveiled regulations to govern the use of AI across the bloc’s 27 member states. The first-of-its-kind proposal spans more than 100 pages and will take years to implement, but the ramifications are far-reaching. It imposes a ban — with some exceptions — on the use of biometric identification systems in public, including facial recognition. Other prohibited applications of AI include social credit scoring, the infliction of harm, and subliminal behavior manipulation.

The regulations are emblematic of an increased desire on the part of consumers for privacy-preserving, responsible implementations of AI and machine learning. A study by Capgemini found that customers and employees will reward organizations that practice ethical AI with greater loyalty, more business, and even a willingness to advocate for them — and in turn, punish those that don’t. And 87% of executives told Juniper in a recent survey that they believe organizations have a responsibility to adopt policies that minimize the negative impacts of AI

Dovetailing with the EU proposal is a data privacy-focused initiative to bring computer science research together with public policy engagement, also announced this week. MIT professor Srini Devadas says that the MIT Future of Data, Trust, and Privacy (FOD) initiative, which will involve collaboration between experts in specific technical areas, gets at the heart of what the EU AI regulations hope to accomplish.

“Enterprises would like to legally share data to do collaborative analytics and consumers want privacy of their data, location, and pictures, and we would like to address both scenarios,” Devadas, a co-director of the initiative, told VentureBeat via email. “The initiative is focused on legal sharing, use and processing of data … Some of the new [EU] regulations ban the use of certain technologies, except for security reasons. I can imagine, for example, that surveillance cameras produce encrypted video streams, and face recognition technology is only applied in a scenario where safety and security considerations are paramount. This might mean that the encrypted streams are processed without decryption — something that is actually possible with available technologies.”

The initiative is the brainchild of MIT Computer Science and Artificial Intelligence Laboratory managing director Lori Glover and Danny Weitzner, who runs the Internet Policy Research Initiative at MIT. As Devadas explains, the goal is integrating research on privacy policy with privacy-preserving technologies to create a virtuous cycle of R&D and regulation. “In many fields, such as medical diagnostics and finance, sharing of data can produce significantly better outcomes and predictions, but sharing is disallowed by [laws] such as HIPAA,” Devadas said. “There are private collaborative analytics techniques that can help with this problem, but it is not always clear if particular techniques or approaches satisfy regulations, because regulations are oftentimes vague. The initiative would like to address this issue.”

Particularly in health care, consumers often aren’t fully aware their information is included in the datasets used to train AI systems. In 2019, The Wall Street Journal reported details on Project Nightingale, Google’s partnership with Ascension, which is America’s second-largest health system, that is collecting the personal health data of tens of millions of patients for the purposes of developing AI-based services for medical providers. A U.K. regulator concluded that The Royal Free London NHS Foundation Trust, a division of the U.K.’s National Health Service based in London, provided Google parent company Alphabet’s DeepMind with data on 1.6 million patients without their consent. And in 2019, Google and the University of Chicago Medical Center were sued for allegedly failing to scrub timestamps from anonymized medical records. (A judge tossed the suit in September.)

The EU regulations impose requirements on “high-risk” applications of AI, including medical devices and equipment. Companies developing them will have to use “high-quality” training data to avoid bias, agree to “human oversight,” and create detailed documentation that explains how the software works to both regulators and users. Moreover, in an effort to provide transparency about what technologies are in use, all high-risk AI systems will be indexed in an EU-wide database.

Devadas sees a number of technologies enabling ethical approaches to AI that align with the EU regulations. For example, secure multiparty computation can be used to arrive at an aggregated statistic in a way that doesn’t expose users’ data. On the horizon are approaches like secure hardware and homographic encryption, which allow AI to derive insights from encrypted data. While homomorphic encryption is possible today, it’s upwards of 100,000 times slower than unsecured techniques. As for secure hardware, it’s often not without security vulnerabilities and requires complete trust in hardware manufacturers.

“There are a lot of techniques that are perhaps close to being deployable but not quite there,” Devadas said. “[However], multiple projects in the initiative will directly address this performance gap.”

For AI coverage, send news tips to Kyle Wiggers — and be sure to subscribe to the AI Weekly newsletter and bookmark our AI channel, The Machine.

Thanks for reading,

Kyle Wiggers

AI Staff Writer

VentureBeat

VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact.

Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:

  • up-to-date information on the subjects of interest to you
  • our newsletters
  • gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
  • networking features, and more

Become a member

Repost: Original Source and Author Link

Categories
Tech News

MIT robot uses radio waves to find and retrieve hidden objects

MIT researchers have developed a robot that can detect and grab objects that are hidden behind walls or pieces of clutter.

The system, called RF-Grasp, uses radio waves to locate items beyond the line-of-sight of a robot’s cameras. It could help warehouse robots grab customer orders or tools that are occluded behind obstacles.

Existing mechanical search systems struggle with these tasks due to the constraints of their sensors. If an object is concealed, they typically need to explore the environment and search for the item.

Unlike visible light and infrared, RF (radio frequency) signals can traverse cardboard boxes, wooden walls, plastic covers, and colored glass to perceive objects fitted with RFID tags.

[Read: How to use AI to better serve your customers]

“Researchers have been giving robots human-like perception,” said study co-author Fadel Adib. “We’re trying to give robots superhuman perception.”

Robotic perception

RF-Grasp is comprised of a camera on the robot’s wrist and a separate RF reader. Together, they collect tracking data and create a visual map of the environment.

The system first pings the object’s RF tag to identify its location. It then determines the optimal path around the obstacles to reach the item.

As the robot gets closer to the object and starts manipulating it, computer vision provides more precise directions.

In tests, RF-Grasp successfully identified and moved objects that were concealed behind packaging and other obstacles. The researchers say the system completed the tasks with about half as much movement as similar robots equipped with only a camera.

The system does depend on target objects being tagged with RFIDs. But the widespread adoption of these chips as barcode replacements in retail, manufacturing, and warehousing means RF-Grasp could already have a practical impact.

You can read the study paper here.

Greetings Humanoids! Did you know we have a newsletter all about AI? You can subscribe to it right here.

Published April 1, 2021 — 18:36 UTC



Repost: Original Source and Author Link

Categories
AI

MIT study finds ‘systematic’ labeling errors in popular AI benchmark datasets

Join Transform 2021 for the most important themes in enterprise AI & Data. Learn more.


The field of AI and machine learning is arguably built on the shoulders of a few hundred papers, many of which draw conclusions using data from a subset of public datasets. Large, labeled corpora have been critical to the success of AI in domains ranging from image classification to audio classification. That’s because their annotations expose comprehensible patterns to machine learning algorithms, in effect telling machines what to look for in future datasets so they’re able to make predictions.

But while labeled data is usually equated with ground truth, datasets can — and do — contain errors. The processes used to construct corpora often involve some degree of automatic annotation or crowdsourcing techniques that are inherently error-prone. This becomes especially problematic when these errors reach test sets, the subsets of datasets researchers use to compare progress and validate their findings. Labeling errors here could lead scientists to draw incorrect conclusions about which models perform best in the real world, potentially undermining the framework by which the community benchmarks machine learning systems.

A new paper and website published by researchers at MIT instill little confidence that popular test sets in machine learning are immune to labeling errors. In an analysis of 10 test sets from datasets that include ImageNet, an image database used to train countless computer vision algorithms, the coauthors found an average of 3.4% errors across all of the datasets. The quantities ranged from just over 2,900 errors in the ImageNet validation set to over 5 million errors in QuickDraw, a Google-maintained collection of 50 million drawings contributed by players of the game Quick, Draw!

The researchers say the mislabelings make benchmark results from the test sets unstable. For example, when ImageNet and another image dataset, CIFAR-10, were corrected for labeling errors, larger models performed worse than their lower-capacity counterparts. That’s because the higher-capacity models reflected the distribution of labeling errors in their predictions to a greater degree than smaller models — an effect that increased with the prevalence of mislabeled test data.

MIT dataset audit

Above: A chart showing the percentage of labeling errors in popular AI benchmark datasets.

In choosing which datasets to audit, the researchers looked at the most-used open source datasets created in the last 20 years, with a preference for diversity across computer vision, natural language processing, sentiment analysis, and audio modalities. In total, they evaluated six image datasets (MNIST, CIFAR-10, CIFAR-100, Caltech-256, and ImageNet), three text datasets (20news, IMDB, and Amazon Reviews), and one audio dataset (AudioSet).

The researchers estimate that QuickDraw had the highest percentage of errors in its test set, at 10.12% of the total labels. CIFAR was second, with around 5.85% incorrect labels, while ImageNet was close behind, with 5.83%. And 390,000 label errors make up roughly 4% of the Amazon Reviews dataset.

Errors included:

  • Mislabeled images, like one breed of dog being confused for another or a baby being confused for a nipple.
  • Mislabeled text sentiment, like Amazon product reviews described as negative when they were actually positive.
  • Mislabeled audio of YouTube videos, like an Ariana Grande high note being classified as a whistle.

A previous study out of MIT found that ImageNet has “systematic annotation issues” and is misaligned with ground truth or direct observation when used as a benchmark dataset. The coauthors of that research concluded that about 20% of ImageNet photos contain multiple objects, leading to a drop in accuracy as high as 10% among models trained on the dataset.

In an experiment, the researchers filtered out the erroneous labels in ImageNet and benchmarked a number of models on the corrected set. The results were largely unchanged, but when the models were evaluated only on the erroneous data, those that performed best on the original, incorrect labels were found to perform the worst on the correct labels. The implication is that the models learned to capture systematic patterns of label error in order to improve their original test accuracy.

Chihuahua mislabeled as a feather boa

Above: A Chihuahua mislabeled as a feather boa in ImageNet.

In a follow-up experiment, the coauthors created an error-free CIFAR-10 test set to measure AI models for “corrected” accuracy. The results show that powerful models didn’t reliably perform better than their simpler counterparts because performance was correlated with the degree of labeling errors. For datasets where errors are common, data scientists might be misled to select a model that isn’t actually the best model in terms of corrected accuracy, the study’s coauthors say.

“Traditionally, machine learning practitioners choose which model to deploy based on test accuracy — our findings advise caution here, proposing that judging models over correctly labeled test sets may be more useful, especially for noisy real-world datasets,” the researchers wrote. “It is imperative to be cognizant of the distinction between corrected versus original test accuracy and to follow dataset curation practices that maximize high-quality test labels.”

To promote more accurate benchmarks, the researchers have released a cleaned version of each test set in which a large portion of the label errors have been corrected. The team recommends that data scientists measure the real-world accuracy they care about in practice and consider using simpler models for datasets with error-prone labels, especially for algorithms trained or evaluated with noisy labeled data.

Creating datasets in a privacy-preserving, ethical way remains a major blocker for researchers in the AI community, particularly those who specialize in computer vision. In January 2019, IBM released a corpus designed to mitigate bias in facial recognition algorithms that contained nearly a million photos of people from Flickr. But IBM failed to notify either the photographers or the subjects of the photos that their work would be canvassed. Separately, an earlier version of ImageNet, a dataset used to train AI systems around the world, was found to contain photos of naked children, porn actresses, college parties, and more — all scraped from the web without those individuals’ consent.

In July 2020, the creators of the 80 Million Tiny Images dataset from MIT and NYU took the collection offline, apologized, and asked other researchers to refrain from using the dataset and to delete any existing copies. Introduced in 2006 and containing photos scraped from internet search engines, 80 Million Tiny Images was found to have a range of racist, sexist, and otherwise offensive annotations, such as nearly 2,000 images labeled with the N-word, and labels like “rape suspect” and “child molester.” The dataset also contained pornographic content like nonconsensual photos taken up women’s skirts.

Biases in these datasets not uncommonly find their way into trained, commercially available AI systems. Back in 2015, a software engineer pointed out that the image recognition algorithms in Google Photos were labeling his Black friends as “gorillas.” Nonprofit AlgorithmWatch showed Cloud Vision API automatically labeled a thermometer held by a dark-skinned person as a “gun” while labeling a thermometer held by a light-skinned person as an “electronic device.” And benchmarks of major vendors’ systems by the Gender Shades project and the National Institute of Standards and Technology (NIST) suggest facial recognition technology exhibits racial and gender bias and facial recognition programs can be wildly inaccurate, misclassifying people upwards of 96% of the time.

Some in the AI community are taking steps to build less problematic corpora. The ImageNet creators said they plan to remove virtually all of about 2,800 categories in the “person” subtree of the dataset, which were found to poorly represent people from the Global South. And this week, the group released a version of the dataset that blurs people’s faces in order to support privacy experimentation.

VentureBeat

VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact.

Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:

  • up-to-date information on the subjects of interest to you
  • our newsletters
  • gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
  • networking features, and more

Become a member

Repost: Original Source and Author Link

Categories
AI

MIT CSAIL taps AI to reduce sheet metal waste

Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) say they’ve created an AI-powered tool that provides feedback on how different parts of laser-cut designs should be placed onto metal sheets. By analyzing how much material is used in real time, they claim that their tool — called Fabricaide — allows users to better plan designs in the context of available materials.

Laser cutting is a core part of industries spanning from manufacturing to construction. However, the process isn’t always efficient. Cutting sheets of metal requires time and expertise, and even the most skillful users can produce leftovers that go to waste.

Fabricaide ostensibly solves this with a workflow that “significantly” shortens the feedback loop between design and fabrication. The tool keeps an archive of what a user has done, tracking how much of each material they have left and allowing the user to assign multiple materials to different parts of the design to be cut. This simplifies the process so that it’s less of a headache for multimaterial designs.

Fabricaide also features a custom 2D packing algorithm that can arrange parts onto sheets in an efficient way, in real time. As the user creates their design, Fabricaide optimizes the placement of parts onto existing sheets and provides warnings if there’s insufficient material, with suggestions for material substitutes.

MIT CSAIL Fabricaide

Fabricaide acts as an interface that integrates with existing design tools and is compatible with computer-assisted design software like AutoCAD, SolidWorks, and Adobe Illustrator. In the future, the researchers hope to incorporate more sophisticated properties of materials, like how strong or flexible they need to be.

“By giving feedback on the feasibility of a design as it’s being created, Fabricaide allows users to better plan their designs in the context of available materials,” Ticha Sethapakdi, a Ph.D. student who led the development of Fabricaide alongside MIT professor Stefanie Mueller, said in a statement. “A lot of these materials are very scarce resources, and so a problem that often comes up is that a designer doesn’t realize that they’ve run out of a material until after they’ve already cut the design. With Fabricaide, they’d be able to know earlier so that they can proactively determine how to best allocate materials.”

The AI in manufacturing market is expected to be valued at $1.1 billion in 2020 and is likely to reach $16.7 billion by 2026, according to Markets and Markets. AI-based solutions like Fabricaide, if commercialized, could help manufacturers to transform their operations by playing a crucial role in automating stages of manufacturing and augmenting human work.

VentureBeat

VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact.

Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:

  • up-to-date information on the subjects of interest to you
  • our newsletters
  • gated thought-leader content and discounted access to our prized events, such as Transform
  • networking features, and more

Become a member

Repost: Original Source and Author Link

Categories
Tech News

New MIT brain research shows how AI could help us understand consciousness

A team of researchers from MIT and Massachusetts General Hospital recently published a study linking social awareness to individual neuronal activity. To the best of our knowledge, this is the first time evidence for the ‘theory of mind‘ has been identified at this scale.

Measuring large groups of neurons is the bread-and-butter of neurology. Even a simple MRI can highlight specific regions of the brain and give scientists an indication of what they’re used for and, in many cases, what kind of thoughts are happening. But figuring out what’s going on at the single-neuron level is an entirely different feat.

According to the paper:

Here, using recordings from single cells in the human dorsomedial prefrontal cortex, we identify neurons that reliably encode information about others’ beliefs across richly varying scenarios and that distinguish self- from other-belief-related representations … these findings reveal a detailed cellular process in the human dorsomedial prefrontal cortex for representing another’s beliefs and identify candidate neurons that could support theory of mind.

In other words: the researchers believe they’ve observed individual brain neurons forming the patterns that cause us to consider what other people might be feeling and thinking. They’re identifying empathy in action.

This could have a huge impact on brain research, especially in the area of mental illness and social anxiety disorders or in the development of individualized treatments for people with autism spectrum disorder.

Perhaps the most interesting thing about it, however, is what we could potentially learn about consciousness from the team’s work.

[Read: How this company leveraged AI to become the Netflix of Finland]

The researchers asked 15 patients who were slated to undergo a specific kind of brain surgery (not related to the study) to answer a few questions and undergo an simple behavioral test. Per a press release from Massachusetts General Hospital:

Micro-electrodes inserted in the dorsomedial prefrontal cortex recorded the behavior of individual neurons as patients listened to short narratives and answered questions about them. For example, participants were presented with this scenario to evaluate how they considered another’s beliefs of reality: “You and Tom see a jar on the table. After Tom leaves, you move the jar to a cabinet. Where does Tom believe the jar to be?”

The participants had to make inferences about another’s beliefs after hearing each story. The experiment did not change the planned surgical approach or alter clinical care.

The experiment basically took a grand concept (brain activity) and dialed it in as much as possible. By adding this layer of knowledge to our collective understanding of how individual neurons communicate and work together to emerge what’s ultimately a theory of other minds within our own consciousness, it may become possible to identify and quantify other neuronal systems in action using similar experimental techniques.

It would, of course, be impossible for human scientists to come up with ways to stimulate, observe, and label 100 billion neurons – if for no other reason than the fact it would take thousands of years just to count them much less watch them respond to provocation.

Luckily, we’ve entered the artificial intelligence age and if there’s one thing AI is good at it’s doing really monotonous things, such as labeling 80 billion individual neurons, really quickly.

It’s not much of a stretch to imagine the Massachusetts team’s methodology being automated. While it appears the current iteration requires the use of invasive sensors – hence the use of volunteers who were already slated to undergo brain surgery – it’s certainly within the realm of possibility that such fine readings could be achieved with an external device one day. 

The ultimate goal of such a system would be to identify and map every neuron in the human brain as it operates in real time. It’d be like seeing a hedge maze from a hot air balloon after an eternity lost in its twists.

This would give us a god’s eye view of consciousness in action and, potentially, allow us to replicate it more accurately in machines. 

Published January 27, 2021 — 20:34 UTC



Repost: Original Source and Author Link

Categories
AI

MIT CSAIL claims its breast cancer prediction AI works equally well across demographic groups

Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) say they’ve created a breast cancer risk-assessment algorithm that shows consistent performance across patients from the U.S., Europe, and Asia. This new algorithm captures the requirements of risk modeling, jointly learning a patient’s risk across multiple future time points and optionally drawing on clinical risk factors like age and family history.

Breast cancer affects more than 2 million women each year, with around one in eight women in the U.S. developing the disease over the course of their lifetime. In 2018 in the U.S. alone, there were also 2,550 new cases of breast cancer in men. And rates of breast cancer are increasing in nearly every region around the world.

The CSAIL team’s algorithm, Mirai, which is also designed to produce predictions consistent across clinical environments, like the choice of mammography machine, was trained on a dataset of over 200,000 exams from Massachusetts General Hospital (MGH). A mammogram image is put through an encoder, which creates an image representation labeled with which view the image came from. Mirai aggregates representations of other images from other views to obtain a representation from the entire mammogram. With the mammogram, a patient’s traditional risk factors are predicted using what’s called a Tyrer-Cuzick model, which accounts for age, weight, and hormonal factors if available. With this information, Mirai predicts a patient’s risk for each year over the next five years.

MIT CSAIL breast cancer

Mirai was designed to predict risk at all time points simultaneously via a tool called an “additive-hazard layer.” This layer anticipates a patient’s risk at a time point, like five years, as an extension of their risk at the previous time point (e.g., four years). Doing so allows the model to learn from data with variable amounts of follow-up.

To benefit from risk factors without requiring them, Mirai predicts that information at training time, and if it’s not there, it uses its own predictive version. Mammograms are rich sources of health information, and many factors including age and menopausal status can be predicted from imaging. As a result of this design, the same model can be used by any clinic globally, the researchers say.

Partly due to a reticence to release code, datasets, and techniques, much of the data used today to train AI algorithms for diagnosing diseases might perpetuate inequalities. To debias Mirai, the CSAIL team used an adversarial scheme where the model learned mammogram representations invariant to the source clinical environment, which helped it to produce accurate predictions.

MIT CSAIL breast cancer

According to the researchers, Mirai was “significantly” more accurate than prior methods — including a two-year-old method by a team of scientists from CSAIL and MIT’s Jameel Clinic — in predicting risk and identifying high-risk groups across test datasets from MGH, the Karolinska Institute in Sweden, and Chang Gung Memorial Hospital in Taiwan. When comparing high-risk cohorts on the MGH test set, the team found that their model identified nearly two times more future cancer diagnoses compared to the current clinical standard. Mirai was similarly accurate across patients of different races, age groups, and breast density categories in the MGH test set and with different cancer subtypes in the Karolinska test set.

Mirai is now installed at MGH, and the team says their collaborators are actively working on integrating the model into point-of-care.

Although Mirai doesn’t currently look at any of the patient’s previous imaging results, the team notes that changes in imaging over time contain a wealth of information. In the future, they aim to create methods that can effectively leverage a patient’s full imaging history.

VentureBeat

VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact.

Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:

  • up-to-date information on the subjects of interest to you
  • our newsletters
  • gated thought-leader content and discounted access to our prized events, such as Transform
  • networking features, and more

Become a member

Repost: Original Source and Author Link