fbpx
Connect with us

The Conversation

Here’s how machine learning can violate your privacy

Published

on

theconversation.com – Jordan Awan, Assistant Professor of Statistics, Purdue – 2024-05-23 07:29:26

If your data was used to train an AI, it might – or might not – be safe from prying eyes.

ValeryBrozhinsky/iStock via Getty Images

Jordan Awan, Purdue University

Machine learning has pushed the boundaries in several fields, personalized medicine, self-driving cars and customized advertisements. Research has shown, however, that these memorize aspects of the data they were trained with in order to learn patterns, which raises concerns for privacy.

Advertisement

In statistics and machine learning, the goal is to learn from past data to make new predictions or inferences about future data. In order to achieve this goal, the statistician or machine learning expert selects a model to capture the suspected patterns in the data. A model applies a simplifying structure to the data, which makes it possible to learn patterns and make predictions.

Complex machine learning models have some inherent pros and cons. On the positive side, they can learn much more complex patterns and work with richer datasets for tasks such as image recognition and predicting how a specific person will respond to a treatment.

However, they also have the risk of overfitting to the data. This means that they make accurate predictions about the data they were trained with but start to learn additional aspects of the data that are not directly related to the task at hand. This to models that aren't generalized, meaning they perform poorly on new data that is the same type but not exactly the same as the data.

While there are techniques to address the predictive error associated with overfitting, there are also privacy concerns from being able to learn so much from the data.

Advertisement

How machine learning algorithms make inferences

Each model has a certain number of parameters. A parameter is an element of a model that can be changed. Each parameter has a value, or setting, that the model derives from the training data. Parameters can be thought of as the different knobs that can be turned to affect the performance of the algorithm. While a straight-line pattern has only two knobs, the slope and intercept, machine learning models have a great many parameters. For example, the language model GPT-3, has 175 .

In order to choose the parameters, machine learning methods use training data with the goal of minimizing the predictive error on the training data. For example, if the goal is to predict whether a person would respond well to a certain medical treatment based on their medical history, the machine learning model would make predictions about the data where the model's developers know whether someone responded well or poorly. The model is rewarded for predictions that are correct and penalized for incorrect predictions, which leads the algorithm to adjust its parameters – that is, turn some of the “knobs” – and try again.

The basics of machine learning explained.

To avoid overfitting the training data, machine learning models are checked against a validation dataset as well. The validation dataset is a separate dataset that is not used in the training . By checking the machine learning model's performance on this validation dataset, developers can ensure that the model is able to generalize its learning beyond the training data, avoiding overfitting.

While this process succeeds at ensuring good performance of the machine learning model, it does not directly prevent the machine learning model from memorizing information in the training data.

Advertisement

Privacy concerns

Because of the large number of parameters in machine learning models, there is a potential that the machine learning method memorizes some data it was trained on. In fact, this is a widespread phenomenon, and users can extract the memorized data from the machine learning model by using queries tailored to get the data.

If the training data contains sensitive information, such as medical or genomic data, then the privacy of the people whose data was used to train the model could be compromised. Recent research showed that it is actually necessary for machine learning models to memorize aspects of the training data in order to get optimal performance solving certain problems. This indicates that there may be a fundamental trade-off between the performance of a machine learning method and privacy.

Machine learning models also make it possible to predict sensitive information using seemingly nonsensitive data. For example, Target was able to predict which customers were likely pregnant by analyzing habits of customers who registered with the Target baby registry. Once the model was trained on this dataset, it was able to send pregnancy-related advertisements to customers it suspected were pregnant because they purchased items such as supplements or unscented lotions.

Is privacy protection even possible?

While there have been many proposed methods to reduce memorization in machine learning methods, most have been largely ineffective. Currently, the most promising solution to this problem is to ensure a mathematical limit on the privacy risk.

Advertisement

The -of-the-art method for formal privacy protection is differential privacy. Differential privacy requires that a machine learning model does not change much if one individual's data is changed in the training dataset. Differential privacy methods achieve this guarantee by introducing additional randomness into the algorithm learning that “covers up” the contribution of any particular individual. Once a method is protected with differential privacy, no possible attack can violate that privacy guarantee.

Even if a machine learning model is trained using differential privacy, however, that does not prevent it from making sensitive inferences such as in the Target example. To prevent these privacy violations, all data transmitted to the organization needs to be protected. This approach is called local differential privacy, and Apple and Google have implemented it.

Differential privacy is a method for protecting people's privacy when their data is included in large datasets.

Because differential privacy limits how much the machine learning model can depend on one individual's data, this prevents memorization. Unfortunately, it also limits the performance of the machine learning methods. Because of this trade-off, there are critiques on the usefulness of differential privacy, since it often results in a significant drop in performance.

Going forward

Due to the tension between inferential learning and privacy concerns, there is ultimately a societal question of which is more important in which contexts. When data does not contain sensitive information, it is easy to recommend using the most powerful machine learning methods available.

Advertisement

When working with sensitive data, however, it is important to weigh the consequences of privacy leaks, and it may be necessary to sacrifice some machine learning performance in order to protect the privacy of the people whose data trained the model.The Conversation

Jordan Awan, Assistant Professor of Statistics, Purdue University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

The Conversation

Poop has been an easy target for microbiome research, but voyages into the small intestine shed new light on ways to improve gut health

Published

on

theconversation.com – Christopher Damman, Associate Professor of Gastroenterology, School of Medicine, of Washington – 2024-06-14 07:38:49

Much of the small intestine microbiome remains an undiscovered frontier.

Stefano Madrigali/Moment via Getty Images

Christopher Damman, University of Washington

Microbiome research to date has been much like the parable of the blind men and the elephant. How much can be said about an elephant by examining just its tail? Researchers have studied what is most readily available – stool rescued from a flush down the toilet – but have been missing the microbial masterminds upstream in the small intestine. Until recently.

Advertisement

Likened by some scientists to another human organ, your microbiome is collectively the tens of trillions of microorganisms that live in interconnected populations on and in your body. They serve as miniature sentinels that protect your body's surfaces from pathogenic invaders. In the upper intestine, distinct microbial populations also aid in digestion, metabolism and even immunity.

I am a gastroenterologist who has spent the past 20 years studying the microbiome's role in health and disease. Advances in technology are helping scientists investigate the small intestine microbiome and the promise it holds for better understanding and treating many diseases.

Big transformations come from small places

Certain members of the small intestine microbiome are linked to obesity and overweight, while other microbial members are linked to a healthy metabolic state. Indeed, small intestine microbes aid in digestion by turning certain simple carbohydrates into the molecular building blocks of a healthy gut and body.

While analogous in function to the colon, small intestine metabolites can be quite distinct from the fiber-derived metabolites of the large intestine microbiome. Some small intestine metabolites help regulate the upper gut's production of GIP, a sister molecule to the lower gut hormone GLP-1, which makes up the weight loss and type 2 diabetes drugs Wegovy and Ozempic. Together, with another lower gut hormone called PYY, this triumvirate is critical for coordinating your body's response to food by regulating your appetite and blood sugar.

Advertisement

Monjaro is an incrementally more powerful combination of GIP and GLP-1 with Wegovy and Ozempic. The full complement of these hormones is naturally stimulated by the breakdown of products from both the large and small intestine microbiome.

The upshot on gut breakdown

Research has linked a disrupted small intestine microbiome to diseases of the gut. These include irritable bowel syndrome (IBS), small intestinal bacterial overgrowth (SIBO), Crohn's disease and Celiac disease.

These diseases are thought to arise partly from disturbances in the way the microbiome breaks down food. Celiac disease, for example, is associated with the small intestine microbiome's decreased ability to digest gluten. IBS and SIBO are linked to the opposite: the small intestine microbiome's ability to too readily ferment fibers and sugars.

Small intestinal bacterial overgrowth, or SIBO, shares similar symptoms with irritable bowel syndrome.

Foods like wheat, garlic, onion, beans and certain processed products that are high in FODMAPs – a set of fermentable short-chain carbohydrates – have been shown to contribute to symptoms in individuals with SIBO and IBS. Lactose-rich dairy is a high FODMAP food group implicated in lactose intolerance and linked to an overzealous small intestine microbiome.

Advertisement

The body's not-so-diplomatic immunity

Diseases associated with the small intestine microbiome aren't limited to metabolism and the gut. In the gut's lining resides a virtual embassy of immune cells that remain in an ever-vigilant state surveying the motley stream of microbial and nutritional antigens passing through your gut.

Compromise in the security that separate the fecal stream from the rest of the body and the processes that keep immune responses in check are hypothesized to play a role in triggering various autoimmune conditions in which the body becomes confused as to who's friend and who's foe.

Studies have linked inflammatory changes in the small intestine microbiome to type 1 diabetes, where the body's circulating immune cells attack insulin-producing cells in the pancreas, and to the extra-intestinal symptoms of Celiac disease, where immune cells can to destructive processes in the body's eyes, skin and joints.

Lights shed in and on the tunnel

Up until very recently, small intestinal research has moved slowly. Scientists relied on upper endoscopy procedures, which involve sedation and inserting a small camera at the end of pinky-thick tubes through the mouth into the very first part of the small intestine.

Advertisement

One of the few alternatives to endoscopies has been studying patients who have had intestinal surgeries that direct portals into their small intestine via a hole in their abdominal wall.

Newly developed technologies are removing the need for sedating medications and unique anatomical situations by allowing scientists to more easily sample the furthest reaches of the gut. Such technologies include camera capsules tethered to angel-hair-thin filaments and other even more streamlined devices that create minimally invasive direct lines of access to the small intestine. Researchers have also developed capsules with sample compartments that open when they reach certain acidity levels in the body.

Close-up of person dangling pill-like device over tongue

Improvements in endoscopy techniques are making it easier to study the small intestine.

Simon Belcher/imageBROKER via Getty Images

These new sampling techniques have unlocked unprecedented access to the upper gut, paving the way for new insights and therapies. In a real- parallel to a childhood favorite, “The Magic School Bus, Inside the Human Body,” researchers can now ride along through the gut like Ms. Frizzle and her class, shining light on the microbial secrets held within.

Advertisement

Accrued alliance in a still-crude science

Therapies based on early understandings of the gut microbiome have included approaches ranging from probiotics to fecal transplants and prebiotics to fermented foods.

But new treatments for gut health are still in their early days. Studying the small intestine could insights to improve therapeutic . A couple of promising future possibilities include partnering small intestine bacteria with their preferred prebiotics and personalized combinations of low FODMAP prebiotics designed to avoid small intestine fermentation.

Treatments that partner food and the microbiome are likely early harbingers of what's to come in the rapidly developing field of microbiome medicine. Researching the small intestine – and not only the gut's tail end – might just be microbiome medicine's most pioneering upstream start.The Conversation

Christopher Damman, Associate Professor of Gastroenterology, School of Medicine, University of Washington

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Advertisement

Read More

The post Poop has been an easy target for microbiome research, but voyages into the small intestine shed new light on ways to improve gut health appeared first on .com

Continue Reading

The Conversation

Quantum computers are like kaleidoscopes − why unusual metaphors help illustrate science and technology

Published

on

theconversation.com – Sorin Adam Matei, Associate Dean for Research, Purdue – 2024-06-14 07:38:17
This image could give you a better grasp of how quantum computers work.
Crystal A Murray/Flickr, CC BY-NC-SA

Sorin Adam Matei, Purdue University

Quantum computing is like Forrest Gump's box of chocolates: You never know what you're gonna get. Quantum phenomena – the behavior of matter and energy at the atomic and subatomic levels – are not definite, one thing or another. They are opaque clouds of possibility or, more precisely, probabilities. When someone observes a quantum system, it loses its quantum-ness and “collapses” into a definite .

Quantum phenomena are mysterious and often counterintuitive. This makes quantum computing difficult to understand. People naturally reach for the familiar to attempt to explain the unfamiliar, and for quantum computing this usually means using traditional binary computing as a metaphor. But explaining quantum computing this way to major conceptual confusion, because at a base level the two are entirely different animals.

This problem highlights the often mistaken belief that common metaphors are more useful than exotic ones when explaining new technologies. Sometimes the opposite approach is more useful. The freshness of the metaphor should match the novelty of the discovery.

Advertisement

The uniqueness of quantum computers calls for an unusual metaphor. As a communications researcher who studies technology, I believe that quantum computers can be better understood as kaleidoscopes.

Digital certainty vs. quantum probabilities

The gap between understanding classical and quantum computers is a wide chasm. Classical computers store and information via transistors, which are electronic devices that take binary, deterministic states: one or zero, yes or no. Quantum computers, in contrast, handle information probabilistically at the atomic and subatomic levels.

Classical computers use the flow of electricity to sequentially open and close gates to record or manipulate information. Information flows through circuits, triggering actions through a of switches that record information as ones and zeros. Using binary math, bits are the foundation of all things digital, from the apps on your phone to the account at your bank and the Wi-Fi signals bouncing around your home.

In contrast, quantum computers use changes in the quantum states of atoms, ions, electrons or photons. Quantum computers link, or entangle, multiple quantum particles so that changes to one affect all the others. They then introduce interference patterns, like multiple stones tossed into a pond at the same time. Some waves combine to create higher peaks, while some waves and troughs combine to cancel each other out. Carefully calibrated interference patterns guide the quantum computer toward the solution of a problem.

Advertisement
Physicist Katie Mack explains quantum probability.

Achieving a quantum leap, conceptually

The term “bit” is a metaphor. The word suggests that during calculations, a computer can break up large values into tiny ones – bits of information – which electronic devices such as transistors can more easily process.

Using metaphors like this has a cost, though. They are not perfect. Metaphors are incomplete comparisons that transfer knowledge from something people know well to something they are working to understand. The bit metaphor ignores that the binary method does not deal with many types of different bits at once, as common sense might suggest. Instead, all bits are the same.

The smallest unit of a quantum computer is called the quantum bit, or qubit. But transferring the bit metaphor to quantum computing is even less adequate than using it for classical computing. Transferring a metaphor from one use to another blunts its effect.

The prevalent explanation of quantum computing is that while classical computers can store or process only a zero or one in a transistor or other computational unit, quantum computers supposedly store and handle both zero and one and other values in between at the same time through the process of superposition.

Advertisement

Superposition, however, does not store one or zero or any other number simultaneously. There is only an expectation that the values might be zero or one at the end of the computation. This quantum probability is the polar opposite of the binary method of storing information.

Driven by quantum science's uncertainty principle, the probability that a qubit stores a one or zero is like Schroedinger's cat, which can be either dead or alive, depending on when you observe it. But the two different values do not exist simultaneously during superposition. They exist only as probabilities, and an observer cannot determine when or how frequently those values existed before the observation ended the superposition.

Leaving behind these challenges to using traditional binary computing metaphors means embracing new metaphors to explain quantum computing.

Peering into kaleidoscopes

The kaleidoscope metaphor is particularly apt to explain quantum processes. Kaleidoscopes can create infinitely diverse yet orderly patterns using a limited number of colored glass beads, mirror-dividing walls and light. Rotating the kaleidoscope enhances the effect, generating an infinitely variable spectacle of fleeting colors and shapes.

Advertisement

The shapes not only change but can't be reversed. If you turn the kaleidoscope in the opposite direction, the imagery will generally remain the same, but the exact composition of each shape or even their structures will vary as the beads randomly mingle with each other. In other words, while the beads, light and mirrors could replicate some patterns shown before, these are never absolutely the same.

If you don't have a kaleidoscope handy, this is a good substitute.

Using the kaleidoscope metaphor, the solution a quantum computer provides – the final pattern – depends on when you stop the computing process. Quantum computing isn't about guessing the state of any given particle but using mathematical models of how the interaction among many particles in various states creates patterns, called quantum correlations.

Each final pattern is the answer to a problem posed to the quantum computer, and what you get in a quantum computing operation is a probability that a certain configuration will result.

New metaphors for new worlds

Metaphors make the unknown manageable, approachable and discoverable. Approximating the meaning of a surprising object or phenomenon by extending an existing metaphor is a method that is as old as calling the edge of an ax its “bit” and its flat end its “butt.” The two metaphors take something we understand from everyday very well, applying it to a technology that needs a specialized explanation of what it does. Calling the cutting edge of an ax a “bit” suggestively indicates what it does, adding the nuance that it changes the object it is applied to. When an ax shapes or splits a piece of wood, it takes a “bite” from it.

Advertisement

Metaphors, however, do much more than convenient labels and explanations of new processes. The words people use to describe new concepts change over time, expanding and taking on a life of their own.

When encountering dramatically different ideas, technologies or scientific phenomena, it's important to use fresh and striking terms as windows to open the mind and increase understanding. Scientists and engineers seeking to explain new concepts would do well to seek out originality and master metaphors – in other words, to think about words the way poets do.The Conversation

Sorin Adam Matei, Associate Dean for Research, Purdue University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Read More

Advertisement

The post Quantum computers are like kaleidoscopes − why unusual metaphors help illustrate science and technology appeared first on .com

Continue Reading

The Conversation

Space weather forecasting needs an upgrade to protect future Artemis astronauts

Published

on

theconversation.com – Lulu Zhao, Assistant Research Scientist in Climate and Sciences and Engineering, University of Michigan – 2024-06-13 07:39:39

The Sun can send out eruptions of energetic particles.

NASA/SDO via AP

Lulu Zhao, University of Michigan

NASA has set its sights on the Moon, aiming to send astronauts back to the lunar surface by 2026 and establish a long-term presence there by the 2030s. But the Moon isn't exactly a habitable place for people.

Advertisement

Cosmic rays from distant and galaxies and solar energetic particles from the Sun bombard the surface, and exposure to these particles can pose a risk to human .

Both galactic cosmic rays and solar energetic particles, are high-energy particles that travel close to the speed of light.

While galactic cosmic radiation trickles toward the Moon in a relatively steady stream, energetic particles can come from the Sun in big bursts. These particles can penetrate human flesh and increase the risk of cancer.

Earth has a magnetic field that provides a shield against high-energy particles from space. But the Moon doesn't have a magnetic field, leaving its surface vulnerable to bombardment by these particles.

Advertisement

During a large solar energetic particle , the radiation dosage an astronaut receives inside a space suit could exceed 1,000 times the dosage someone on Earth receives. That would exceed an astronaut's recommended lifetime limit by 10 times.

NASA's Artemis program, which began in 2017, intends to reestablish a human presence on the Moon for the first time since 1972. My colleagues and I at the University of Michigan's CLEAR center, the Center for All-Clear SEP Forecast, are working on predicting these particle ejections from the Sun. Forecasting these may protect future Artemis crew members.

A group of astronauts in blue jumpsuits stand or kneel on a stage in front of a screen displaying the Artemis logo.

With Artemis, NASA plans to return humans to the lunar surface.

AP Photo/Michael Wyke

An 11-year solar cycle

The Moon is facing dangerous levels of radiation in 2024, since the Sun is approaching the maximum point in its 11-year solar cycle. This cycle is driven by the Sun's magnetic field, whose total strength changes dramatically every 11 years. When the Sun approaches its maximum activity, as many as 20 large solar energetic particle events can happen each year.

Advertisement

Both solar flares, which are sudden eruptions of electromagnetic radiation from the Sun, and coronal mass ejections, which are expulsions of a large amount of matter and magnetic fields from the Sun, can produce energetic particles.

A coronal mass ejection erupting from the Sun.

The Sun is expected to reach its solar maximum in 2026, the target launch time for the Artemis III mission, which will an astronaut crew on the Moon's surface.

While researchers can follow the Sun's cycle and predict trends, it's difficult to guess when exactly each solar energetic particle event will occur, and how intense each event will be. Future astronauts on the Moon will need a warning system that predicts these events more precisely before they happen.

Forecasting solar events

In 2023, NASA funded a five-year space weather center of excellence called CLEAR, which aims to the probability and intensity of solar energetic particle events.

Advertisement

Right now, forecasters at the National Oceanic and Atmospheric Administration Space Weather Prediction Center, the center that tracks solar events, can't issue a warning for an incoming solar energetic particle event until they actually detect a solar flare or a coronal mass ejection. They detect these by looking at the Sun's atmosphere and measuring X-rays that flow from the Sun.

Once a forecaster detects a solar flare or a coronal mass ejection, the high-energy particles usually arrive to Earth in less than an hour. But astronauts on the Moon's surface would need more time than that to seek shelter. My team at CLEAR wants to predict solar flares and coronal mass ejections before they happen.

Two illustrations of a sphere with purple and green lines coming off it. On the left, the purple lines are coming off the top and the green lines off the bottom. On the right, the lines are scattered around and overlapping.

The solar magnetic field is incredibly complex and can change throughout the solar cycle. On the left, the magnetic field has two poles and looks relatively simple, though on the right, later in the solar cycle, the magnetic field has changed. When the solar magnetic field looks like the illustration on the right, solar flares and coronal mass ejections are more common.

NASA's Goddard Space Flight Center/Bridgman, CC BY

While scientists don't totally understand what causes these solar events, they know that the Sun's magnetic field is one of the key drivers. Specifically, they're studying the strength and complexity of the magnetic field in certain regions on the Sun's surface.

Advertisement

At the CLEAR center, we will monitor the Sun's magnetic field using measurements from both ground-based and space-based telescopes and build machine learning models that predict solar events – hopefully more than 24 hours before they happen.

With the forecast framework developed at CLEAR, we also hope to predict when the particle flux falls back to a safe level. That way, we'll be able to tell the astronauts when it's safe to their shelter and continue their work on the lunar surface.The Conversation

Lulu Zhao, Assistant Research Scientist in Climate and Space Sciences and Engineering, University of Michigan

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Read More

Advertisement

The post Space weather forecasting needs an upgrade to protect future Artemis astronauts appeared first on .com

Continue Reading

News from the South

Trending