KAIST develops AI that automatically detects defects in smart factory manufacturing processes even when conditions change​

The Korea Advanced Institute of Science and Technology (KAIST)

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

(From left) Ph.D candidate Jihye Na, Professor Jae-Gil Lee 

view more 

Credit: KAIST

Recently, defect detection systems using artificial intelligence (AI) sensor data have been installed in smart factory manufacturing sites. However, when the manufacturing process changes due to machine replacement or variations in temperature, pressure, or speed, existing AI models fail to properly understand the new situation and their performance drops sharply. KAIST researchers have developed AI technology that can accurately detect defects even in such situations without retraining, achieving performance improvements up to 9.42%. This achievement is expected to contribute to reducing AI operating costs and expanding applicability in various fields such as smart factories, healthcare devices, and smart cities.

KAIST (President Kwang Hyung Lee) announced on the 26th of August that a research team led by Professor Jae-Gil Lee from the School of Computing has developed a new “time-series domain adaptation” technology that allows existing AI models to be utilized without additional defect labeling, even when manufacturing processes or equipment change.

Time-series domain adaptation technology enables AI models that handle time-varying data (e.g., temperature changes, machine vibrations, power usage, sensor signals) to maintain stable performance without additional training, even when the training environment (domain) and the actual application environment differ.

Professor Lee’s team paid attention to the fact that the core problem of AI models becoming confused by environmental (domain) changes lies not only in differences in data distribution but also in changes in defect occurrence patterns (label distribution) themselves. For example, in semiconductor wafer processes, the ratio of ring-shaped defects and scratch defects may change due to equipment modifications.

The research team developed a method for decomposing new process sensor data into three components—trends, non-trends, and frequencies—to analyze their characteristics individually. Just as humans detect anomalies by combining pitch, vibration patterns, and periodic changes in machine sounds, AI was enabled to analyze data from multiple perspectives.

In other words, the team developed TA4LS (Time-series domain Adaptation for mitigating Label Shifts) technology, which applies a method of automatically correcting predictions by comparing the results predicted by the existing model with the clustering information of the new process data. Through this, predictions biased toward the defect occurrence patterns of the existing process can be precisely adjusted to match the new process.

In particular, this technology is highly practical because it can be easily combined like an additional plug-in module inserted into existing AI systems without requiring separate complex development. That is, regardless of the AI technology currently being used, it can be applied immediately with only simple additional procedures.

In experiments using four benchmark datasets of time-series domain adaptation (i.e., four types of sensor data in which changes had occurred), the research team achieved up to 9.42% improvement in accuracy compared to existing methods.[TT1]

Especially when process changes caused large differences in label distribution (e.g., defect occurrence patterns), the AI demonstrated remarkable performance improvement by autonomously correcting and distinguishing such differences. These results proved that the technology can be used more effectively without defects in environments that produce small batches of various products, one of the main advantages of smart factories.

Professor Jae-Gil Lee, who supervised the research, said, “This technology solves the retraining problem, which has been the biggest obstacle to the introduction of artificial intelligence in manufacturing. Once commercialized, it will greatly contribute to the spread of smart factories by reducing maintenance costs and improving defect detection rates.”

This research was carried out with Jihye Na, a Ph.D. student at KAIST, as the first author, with Youngeun Nam, a Ph.D. student, and Junhyeok Kang, a researcher at LG AI Research, as co-authors. The research results were presented in August 2025 at KDD (the ACM SIGKDD Conference on Knowledge Discovery and Data Mining), the world’s top academic conference in artificial intelligence and data.
※Paper Title: “Mitigating Source Label Dependency in Time-Series Domain Adaptation under Label Shifts”
※DOI: https://doi.org/10.1145/3711896.3737050

This technology was developed as part of the research outcome of the SW Computing Industry Original Technology Development Program’s SW StarLab project (RS-2020-II200862, DB4DL: Development of Highly Available and High-Performance Distributed In-Memory DBMS for Deep Learning), supported by the Ministry of Science and ICT and the Institute for Information & Communications Technology Planning & Evaluation (IITP).

 

---

DOI

10.1145/3711896.3737050 

Method of Research

Meta-analysis

Subject of Research

Not applicable

Article Title

Mitigating Source Label Dependency in Time-Series Domain Adaptation under Label Shifts

Article Publication Date

23-Aug-2025

University of Tennessee collaborates on NSF grants to improve outcomes through AI

University of Tennessee at Knoxville

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

Tabitha Samuel

view more 

Credit: University of Tennessee

Faculty members from the Min H. Kao Department of Electrical Engineering and Computer Science at University of Tennessee are involved in two collaborative National Science Foundation grants that aim to address health disparities research and enhance the performance and productivity of AI science.

Tabitha Samuel, the interim director and operations group leader for UT’s National Institute for Computational Sciences (NICS), is the principal investigator for UT on both projects.

AI Advancement in Health Research

The first grant ($82,824) is a statewide collaboration among Tennessee Tech University, UT, Meharry Medical College, and Vanderbilt University that is part of the NSF’s National Artificial Intelligence Research Resource (NAIRR) Pilot.

The project, Mid-South Conferences on Cyberinfrastructure Advances to Enable Interdisciplinary AI Research in Health, will train participants how to use high-performance computing, cloud-based AI applications, and open data tools in medical research and healthcare delivery.

According to the 2023 America’s Health Rankings report, Tennessee ranks 44th among the 50 states in national health outcomes. This project will advance the use of modern, AI/ML-enabled computer technology in medical research and healthcare delivery while fostering sustained collaboration among medical professionals, engineers, scientists, and students who participate. Adhering to UT’s land-grant mission, the researchers will share content and outcomes with the NSF NAIRR program and broadly with the public at no charge.

The project consists of three workshops to be held in Knoxville, Nashville, and Memphis every six months over an 18-month period.

The co-PIs on this grant from UT are Vasileios Maroulas, director of the AI Tennessee Initiative; Courtney Cronley, a professor in the College of Social Work; Hector Santos-Villalobos, EECS assistant professor; and Fatima Zahra, an assistant professor of evaluation, statistics, and research methodology in the Department of Educational Leadership and Policy Studies.

“We focused on bringing AI training for health disparities research in Tennessee and the Mid-South area because we are aware of the magnitude of research being done around health disparities unique to the region,” Samuel said. “We hope that this AI training, coupled with exposure to the expanse of NAIRR resources, will empower Tennessee researchers with a distinct advantage in addressing and mitigating health disparities through innovative and impactful research.”

Boosting AI Speed and Efficiency

The second grant ($800,000) is a collaboration among Tennessee Tech, UT, Illinois Institute of Technology, and Stony Brook University. This NSF Cyber Infrastructure for Sustained Scientific Innovation grant aims to improve how massively parallel computers run large-scale artificial intelligence (AI) applications by enhancing the Message Passing Interface (MPI), a widely used standard for coordinating work across many high-performance-computing nodes in parallel programs.

Currently, the enabling data-transfer software used in AI for communication between computers enhanced by Graphical Processing Units (GPUs) are often proprietary and/or limited in scope; they cannot be expanded or enhanced by an open community. That situation restricts innovation, making it harder for scientists to collaborate and enhance their science output on limited computer resources, while also creating dependency on a few vendors.

By contrast, this project, Enhancing Performance and Productivity of AI Science through Next-generation High Performance Communication Abstractions, builds on and advances Open MPI, a major open-source implementation of MPI with a long history of broad impact, to make it more efficient, flexible, and better suited for modern AI tasks.

“In this age of AI, how do we instrument and improve MPI to perform better for AI codes?” Samuel said. “The hardware is at the stage where AI can do fairly well on HPC hardware. But the next question is, how do AI codes perform across multiple nodes and scaling? That’s where this project comes in.”

 

Study: Fossils reveal reliable record of marine ecosystem functioning

New research confirms that fossilized marine invertebrates serve as a powerful tool for understanding long-term ecological change and informing modern conservation efforts.

University of Nevada, Las Vegas

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

(Image courtesy of Carrie Tyler/UNLV)

view more 

Credit: Carrie Tyler/UNLV

A new study published in the Proceedings of the National Academy of Sciences confirms that fossilized remains of marine invertebrates can accurately reflect the functional diversity of past ecosystems—offering a powerful tool for understanding long-term ecological change and informing modern conservation.

UNLV geoscience professor Carrie Tyler, in collaboration with Michał Kowalewski from the University of Florida, compared living marine communities with their corresponding skeletal remains and the fossil record across 51 coastal sites in Onslow Bay, North Carolina. 

The study included over 200 species from six major invertebrate groups and was unique in that it included more types of organisms than are typically studied.  

The results show high functional fidelity—meaning that key ecological traits such as feeding strategies, mobility, and habitat use are well preserved in the fossil record. Despite natural biases in preservation, the fossil and skeletal remains captured nearly all the functional roles found in the current living communities.

“We found that fossils don’t just tell us what species lived in the past—they also preserve how ecosystems functioned,” said Tyler. “That’s critical for understanding both ancient ecosystems and the baseline conditions of today’s marine environments.”

What does this mean for ecosystem conservation efforts?

The findings support the growing field of conservation paleobiology, which uses the youngest fossil records to assess how ecosystems have changed over time—especially in response to human impacts. By validating the use of functional diversity metrics in fossil data, this research opens the door to more accurate reconstructions of past ecosystems and better-informed strategies for marine conservation. 

Conservation organizations can reliably use fossil records when assessing functional diversity in an ecosystem to compare the species present, the functions those species carry out, and the overall health of the ecosystem – all keys to ecosystem protection and restoration. 

Knowing that functional diversity is preserved in the fossil record can help conservationists determine what functions are lacking within that ecosystem so they can determine what needs to be replaced to restore the ecosystem to a healthy state. 

“There are no pristine ecosystems left on the planet, so when you are trying to restore an ecosystem that can be a difficult and challenging task without having any idea of what it looked like before,” said Tyler. “It hasn’t been free of human impacts or pristine for thousands of years, so we don’t have records of what that ecosystem is ‘supposed’ to look like. This study shows the fossil record can be used to give us an idea of what that ecosystem used to look like, and what functions are needed to keep it healthy.”

About the Study

The study, "Fossil samples archive functional diversity in marine ecosystems: An empirical test from present-day coastal environment," was published July 28, 2025 in Proceedings of the National Academy of Sciences.

---

Journal

Proceedings of the National Academy of Sciences

DOI

10.1073/pnas.2405727122 

Method of Research

Data/statistical analysis

Subject of Research

Not applicable

Article Title

Fossil samples archive functional diversity in marine ecosystems: An empirical test from a present-day coastal environment

CRIMINAL CAPITALI$M

Most US neurologists prescribing MS drugs have received pharma industry cash

Higher volume prescribers more likely to receive payments; and recipients more likely to prescribe that company’s drugs, especially if payments were larger, sustained, and recent

BMJ Group

Facebook

XLinkedInWeChatBlueskyMessageWhatsAppEmail

Nearly 80% of US neurologists prescribing drugs for multiple sclerosis (MS) received at least one pharma industry payment, with higher volume prescribers more likely to be beneficiaries, finds a 5 year analysis of Medicare database payments, published in the open access journal BMJ Open.

And those in receipt of these payments were more likely to prescribe that company’s drugs, especially if the sums involved were larger, sustained, and recent, the findings indicate.

Because of the lifelong nature of MS, effective therapies are usually continued indefinitely unless a patient’s clinical response changes, explain the researchers. And MS drug prescriptions are Medicare’s largest neurological drug expense despite making up a relatively small portion of total claims, they add.

While previously published research indicates that industry payments are associated with increased prescribing of marketed products, none of these studies focused on a market as competitive as the MS drugs market, say the researchers. 

They therefore set out to characterise industry payments to neurologists prescribing MS drugs and find out if the receipt of such payments might be associated with the likelihood of the preferential prescribing of that company’s drugs.

They used publicly available data on payments made by pharma companies to doctors from the Centers for Medicare & Medicaid (CMS) Open Payments platform from 2015 to 2019.

Payments are classified as: research payments; ownership and investment interests; and general payments. The researchers focused on general payments to neurologists, linking these to Medicare Part D data, which covers prescription drugs, using National Provider Identification numbers and drug names. 

Their analysis included 7401 neurologists who had prescribed disease modifying therapies (DMTs) for at least 1 year, issuing a minimum of 11 prescriptions, and 20 DMTs manufactured by 10 companies.

In all, 5809 (78.5%) neurologists received 626,290 distinct industry payments from at least one drug company, totalling US$163.6 million between 2015 and 2019; 4999 (67.5%) neurologists received payments from two or more companies.

The average individual amount received was US$779, but 10% of recipients amassed US$155.7 million between them—95% of the total sums received–which suggests that drug companies may selectively target high-volume prescribers, say the researchers.

Higher prescription volumes were associated with a greater likelihood of receiving any payment type, particularly for consulting services, non-consulting services, such as speaking at an event, and travel/accommodation; the highest number of discrete payments was made for food and drink. 

The amount received was positively associated with prescription volume. Compared with those who received no payments from a company, those who did, were 13% more likely to prescribe that company’s drugs.

The strongest association between industry payment and prescribing tendencies was observed for non-consulting services. These neurologists were 53% more likely to prescribe that company’s drugs. 

Larger payments were also associated with a greater likelihood of preferential prescribing, rising in tandem with the size of the payment received: US$50 was associated with a 10% greater likelihood of prescribing that company’s drugs; US$500 with a 26% greater likelihood; US$1000 with a 29% greater likelihood; and US$5000 with a 50% greater likelihood.

Longer duration of payments was another seemingly influential factor, ranging from a 12% greater likelihood of prescribing that company’s drugs for one year of payments to 78% greater likelihood for 5 consecutive years. 

The recency of payments also seemed to be influential. A payment made 4 years earlier was associated with a 3% greater likelihood of prescribing that company’s drugs, but a 34% greater likelihood when made in the same year.

This is an observational study, and as such, no firm conclusions can be drawn about cause and effect. And the researchers acknowledge that their study was limited to the prescribing of Part D drugs, and couldn’t establish the appropriateness of prescribing, nor for which patients more expensive brand-name drugs were most suitable. 

A doctor’s decision to prescribe is informed by many different factors, including national guidelines and/or institutional protocols, insurance cover, and patient preferences. These drivers are difficult to assess using publicly available data, but should be considered when interpreting the findings, emphasise the researchers.

Nevertheless their “findings raise concerns about excess pharmaceutical promotion efforts and their implications for physician prescribing for patients,” they suggest.

“Promotional efforts to influence prescribing are especially concerning given the drugs’ substantial costs, particularly if more expensive brand-name drugs are being prescribed instead of appropriate, effective, generically available alternatives,” they point out.

“The Physician Payments Sunshine Act, which led to the creation of the Open Payments Database, was an important step forward in making transparent the financial conflicts of interest among physicians receiving industry payments.

“However, it remains unclear whether increased transparency has mitigated these conflicts of interest and their impact on prescribing behaviour, or simply given the public greater insight into the large scale of industry payments made to prescribers,” they conclude.

---

Journal

BMJ Open

DOI

10.1136/bmjopen-2024-095952 

Method of Research

Observational study

Subject of Research

People

Article Title

Industry payments to US neurologists related to multiple sclerosis drugs and prescribing (2015–2019): a retrospective cohort study

Article Publication Date

26-Aug-2025

COI Statement

All authors have completed the ICMJE uniform disclosure form at http://www.icmje.org/disclosure-of-interest/ and declare no support from any organisation for the submitted work; JSR was an expert witness at the request of Relator's attorneys, the Greene Law Firm, in a qui tam suit alleging violations of the False Claims Act and Anti-Kickback Statute against Biogen that was settled in September 2022.

How Big Pharma Bought Government to Protect its Racket

Brandon Novick

 August 22, 2025

FacebookTwitterRedditBlueskyEmail

The US government is pay-to-play – and drug lobbyists are buying a lot of playing time.

Pharmaceutical companies claim that the government shouldn’t negotiate lower drug prices because losing those excess profits will hurt innovation, but they can pour record amounts of money into lobbying the government. The premier lobbying group for Big Pharma – the Pharmaceutical Research and Manufacturers of America (PhRMA) –  spent over $20.6 million on lobbying the federal government in the first half of 2025, including more than $7.6 million in the second quarter.

Pharmaceutical and health products companies overall spent $105.4 million in the second quarter of 2025 and $226.8 million for the first half the year. This lobbying boom is an extension of growing spending over the last few years, as the industry spent around $22 million more than it had in the first half of 2024.

Lobbying literally means the act of influencing government actions, and PhRMA’s spending successfully reaped rewards in the recently signed reconciliation packagethat President Trump coined the “One Big Beautiful Bill” (OBBB).

The United States spends far more than other countries for the same prescription drugs. Compared to the 38 countries in the Organisation for Economic Co-operation and Development (OECD) – which are mostly other democratic, developed nations – US prices were roughly three times as high for the same products. Estimates haveshown two-thirds to three-fourths of global pharmaceutical profits come from the US alone. The pharmaceutical industry achieved two massive wins in the OBBB to help ensure that this price gouging of Americans continues.

What PhRMA Won in the OBBB

Unlike other countries, the US government doesn’t use its significant purchasing power to negotiate and lower prices. However, the 2022 Inflation Reduction Act (IRA) introduced very limited drug negotiation in the Medicare program for the 50 drugs with the highest amount of spending in Medicare Parts B and D each. More specifically, the IRA allows Medicare to negotiate prices for a whopping 10 drugs starting in 2026, adding another 15 in 2027, another 15 in 2028, and 20 in 2029 and beyond.

The OBBB increased the number of drugs exempted from the limited price negotiation program, which the Congressional Budget Office estimates will save the industry $5 billion over ten years. Before the OBBB, drugs that the Food and Drug Administration (FDA) approved to treat patient populations of under 200,000 for a single rare disease – known as orphan drugs – were exempt from price negotiation.

The OBBB tweaked the law so that drugs approved to treat multiple rare diseases are also exempt. This change is significant, as pharmaceutical companies often chase orphan drug designations for their products because they provide significant financial incentives like tax credits, fee exemptions, grants, and market exclusivities. Orphan drugs are also significantly more expensive than nonorphan drugs. One analysis of 242 FDA-approved orphan and nonorphan drugs from 2017-2021 found that the median cost of orphan drugs was roughly $219,000 compared to $13,000 for nonorphan drugs.

The other significant win for the pharmaceutical industry was the exclusion of a supposed Trump policy to lower drug prices. In May 2025, President Trump issuedan executive order claiming to create a most-favored-nation policy to lower drug prices. Essentially, such a policy would set the prices of drugs in the US to the lowest level paid by comparable countries. However, President Trump issued a similar executive order in his first administration, which the courts struck down for violatinglegal procedures.

To put President Trump’s executive order into law, Representatives Ro Khanna (D-CA), Anna Paulina Luna (R-FL), Marcy Kaptur (D-OH), and Andy Biggs (R-AZ) introduced the Global Fairness in Drug Pricing Act in May to codify key provisions of the executive order. Yet, the OBBB did not include this bipartisan policy. The pharmaceutical industry won out again.

The industry’s lobbying successes do not end there. Among its other wins, the OBBB nor any other piece of legislation has included Department of Health & Human Services (HHS) Secretary Robert Kennedy’s desire to rein in direct-to-consumer advertising. Currently, the United States and New Zealand are the only two countries that allow pharmaceutical companies to spend money directly advertising their drugs to consumers. Additionally, companies are financially incentivized to spend big on such advertising as these expenditures are exempt from federal taxes. Secretary Kennedy has openly endorsed ending this tax exemption and Senators Josh Hawley (R-MO) and Jeanne Shaheen (D-NH) have introduced the No Handouts for Drug Advertisements Act to do so.

Senators Bernie Sanders (I-VT) and Angus King (I-ME) also introduced legislation to ban such direct-to-consumer advertising outright, a policy that Secretary Kennedy has repeatedly advocated for. Unsurprisingly, Congress has not passed such a ban.

Why do drug companies spend millions on lobbying the federal government when they also price-gouge Americans so that they reap every last penny while willingly and knowingly denying life-saving treatments to patients? The same question applies to the tens of millions of dollars the industry spends through campaign contributions to members of Congress.

The answer is clear: Influence. Companies know that by spending more money to get favorable candidates elected, and by bombarding elected officials, their staff, and regulators with their priorities, they can reap massive profits.

Spending millions on lobbying and campaign contributions can influence policy that creates billions in returns – either from changes in public policy or by maintaining the status quo. Demonstrating the magnitude of this dynamic, one study found that every $1 corporations spent on lobbying for a tax holiday provision in the American Jobs Creation Act of 2004 yielded $220 – a 22,000 percent return on investment. For many pharmaceutical companies and other corporate interests, buying political power is arguably the most lucrative investment they can make.

This first appeared on CEPR.

Brandon Novick is a Program Outreach Assistant for the Domestic Team at the Center for Economic and Policy Research in Washington, D.C.