PRESS-NEWS.org - Press Release Distribution
PRESS RELEASES DISTRIBUTION

Common feature between forest fires and neural networks reveals the universal framework underneath

Scientists find that scaling laws apply to deep neural networks exhibiting absorbing phase transitions

2025-07-18
(Press-News.org)

Researchers from the University of Tokyo in collaboration with Aisin Corporation have demonstrated that universal scaling laws, which describe how the properties of a system change with size and scale, apply to deep neural networks that exhibit absorbing phase transition behavior, a phenomenon typically observed in physical systems. The discovery not only provides a framework describing deep neural networks but also helps predict their trainability or generalizability. The findings were published in the journal Physical Review Research.

In recent years, it seems no matter where we look, we come across artificial intelligence in one form or another. The current version of the technology is powered by deep neural networks: numerous layers of digital “neurons” with weighted connections between them. The network learns by modifying the weights between the “neurons” until it produces the correct output. However, a unified theory describing how the signal propagates between the layers of neurons in the system has eluded scientists so far.

“Our research was motivated by two drivers,” says Keiichi Tamai, the first author, “partially by industrial needs as brute-force tuning of these massive models takes a toll on the environment. But there was a second, deeper pursuit: the scientific understanding of the physics of intelligence itself.”

This is where Tamai’s background in statistical physics of phase transitions gave him the first hint. Absorbing phase transitions refer to a sharp shift at a tipping point from an active to an absorbing phase, from which the system cannot escape without outside help. An example of such a physical system would be a fire burning out. Crucially, these systems exhibit universal behaviors near the tipping point and can be described using universal scaling laws if certain properties are preserved. If deep neural networks exhibit absorbing phase transitions, then then universal scaling laws may apply, providing unified framework for describing how they function. Consequently, researchers would be able to predict whether a signal would “burn out” in a certain deep learning set up.

To investigate, the researchers combined theory with simulations. They derived the exponents, which are universal across systems, and scaling factors, which differ across systems, from theory when possible and used simulations to confirm the scaling laws in more complex cases.

“What a coincidence, I thought,” Tamai says, remembering when he first noticed the link between deep neural networks and absorbing phase transitions. “I never imagined I would end up doing research on deep learning, let alone finding an effective use of a concept I worked on as a doctoral student in physics.”

The finding also brings us closer to understanding the physics of intelligence itself, as it reinvigorates the brain criticality hypothesis, which states that some biological networks operate near phase transitions. Tamai is excited about the prospects of this line of research.

“Alan Turing hinted at this connection as early as 1950, but the tools weren’t ready back then. With the rapid accumulation of evidence in neuroscience and the rise of near-human-level AI, I believe we’re at a perfect moment to revisit and deepen our understanding of this fundamental relationship.”

END



ELSE PRESS RELEASES FROM THIS DATE:

New R package revolutionizes gene set enrichment analysis visualization for biomedical research

2025-07-18
Scientists at China Pharmaceutical University have launched GseaVis, a groundbreaking R package designed to transform how researchers visualize and interpret Gene Set Enrichment Analysis (GSEA) results in biomedical research. Gene Set Enrichment Analysis is a powerful computational method widely used to determine whether predefined sets of genes show statistically significant differences between biological states, such as healthy versus diseased conditions. However, effective visualization of GSEA results ...

Benign uterine tumor resembling cancer highlights diagnostic challenges

2025-07-18
“This case highlights that lipoleiomyomas can present many years after attaining menopause.” BUFFALO, NY — July 18, 2025 — A new case report was published in Volume 12 of Oncoscience on July 2, 2025, titled “Navigating diagnostic complexity: A case report on uterine lipoleiomyoma, unveiling its benign nature amidst characteristics resembling liposarcoma.” In this report, Seetu Palo, first corresponding author Mishu Mangla and colleagues from the All India Institute of Medical Sciences describe a rare case where a benign uterine tumor closely resembles a malignant tumor in an elderly patient. ...

Diagnostic performance of ATR-FTIR spectroscopy in discriminating normal breast tissue and breast tumors

2025-07-18
Background and objectives Histopathology is the gold standard in cancer diagnosis. However, attenuated total reflectance (ATR)-Fourier transform infrared (FTIR) spectroscopy has shown diagnostic potential in other settings. Therefore, this study aimed to investigate the sensitivity and specificity of the ATR-FTIR spectroscopy in evaluating breast lesions. Methods This study was conducted on formalin-fixed, paraffin-embedded biopsy blocks received at Ladoke Akintola University of Technology Teaching Hospital between ...

Recent advancements in the treatment of age-related macular degeneration

2025-07-18
Age-related macular degeneration (AMD) remains a leading cause of vision loss globally, with significant advancements in treatment options for both dry and wet AMD. As highlighted by Huang and colleagues, "Current therapies for dry AMD have limited effectiveness in halting the progression of geographic atrophy (GA), underscoring the need for innovative approaches". One of the most notable breakthroughs is the FDA approval of pegcetacoplan and avacincaptad pegol, targeting the complement system to slow GA progression. Pegcetacoplan, a C3 inhibitor, reduced GA lesion growth by 19–22% ...

Comprehensive multi-omics integration reveals B cell-derived ELL2 as a novel diagnostic and prognostic biomarker in sepsis

2025-07-18
Sepsis is a life-threatening condition caused by organ dysfunction resulting from the body's dysregulated response to infection. Annually, sepsis affects more than 31 million people worldwide, with a mortality rate of approximately 30%. The disease is characterized by rapid progression, poor prognosis, and high mortality, posing significant challenges in critical care medicine. Early diagnosis and intervention are crucial for improving patient outcomes. In a study published in the journal "Med Research," a team of researchers ...

How tumor microbes shape cancer: New insights into microbial diversity in the tumor microenvironment

2025-07-18
Intratumoral microbiota—the microbial populations residing within solid tumors—have emerged as pivotal components of the tumor microenvironment (TME), influencing tumor initiation, progression, and therapeutic outcomes. In a comprehensive review published in Med Research, researchers from Southern Medical University and collaborating institutions systematically examine the heterogeneity of these microorganisms and their multifaceted roles in cancer biology. The authors categorize microbial heterogeneity at three distinct levels: among different ...

AI-powered biomarker system redefines thyroid cancer progression and subtypes

2025-07-18
Differentiated thyroid carcinoma (DTC) is typically indolent, and some patients may be managed conservatively without immediate surgery. However, it remains a clinical challenge to determine who is suitable for active surveillance and to identify when disease progression may occur. A new study led by researchers at The First Affiliated Hospital of Zhengzhou University addresses this issue by developing an innovative dynamic biomarker system. Using an optimized dynamic network biomarker (DNB) algorithm, the team identified ...

Digestive cancers in China: A two-decade burden with uneven gains in life expectancy

2025-07-18
Digestive system malignancies (DSMTs)—including oesophageal, gastric, colorectal, liver, and pancreatic cancers—represent a major public health challenge in China. A recent national epidemiological study by Hu et al., published in eGastroenterology, analysed trends from 2004 to 2021, revealing complex patterns of burden and survival. While progress has been made in reducing mortality from some cancers, the overall impact remains significant, with notable disparities across demographic and geographic lines. Below ...

ANXA2+ migratory hepatocytes identified as key players in human liver regeneration

2025-07-18
Acute liver failure (ALF) represents a life-threatening condition marked by rapid deterioration of liver function, often necessitating urgent liver transplantation. In the United States, acetaminophen (APAP) overdose remains the most common cause of drug-induced ALF, accounting for nearly half of all cases. Given the scarcity of viable donor livers—worsened by the prevalence of steatotic liver disease—understanding and promoting the liver's intrinsic regenerative capacity has become an urgent priority ...

Researchers demonstrate a new material to reduce power consumption of electronics

2025-07-18
MINNEAPOLIS / ST. PAUL (07/18/2025) — Research from the University of Minnesota Twin Cities gives new insight into a material that could make computer memory faster and more energy-efficient. The study was recently published in Advanced Materials, a peer-reviewed scientific journal. The researchers also have a patent on the technology. As technology continues to grow, so does the demand for emerging memory technology. Researchers are looking for alternatives and complements to existing memory solutions that can perform at high levels with ...

LAST 30 PRESS RELEASES:

Current and recommended diets in the USA have embedded forced labor risk

AI breakthrough helps astronomers spot cosmic events with just a handful of examples

New vaccine shows promise against typhoid and invasive salmonella in first human trial

Engineered “natural killer” cells could help fight cancer

New 3D printing method ‘grows’ ultra-strong materials

Lizard genetics provide new perspective on evolution

Can a Stevia-derived sweetener improve hair loss treatment?

Method to assess the status of wild reindeer may help with conservation efforts

Do imported cut flowers spread livestock viruses?

Does prior incarceration contribute to poor health later in life?

Could slime mold microbes be a source of potent antimicrobials?

Record-breaking 2024 Amazon fires drive unprecedented carbon emissions and ecosystem degradation

Birds thrive despite pollution from ‘forever’ chemicals

Deadwood brings wild orchids to life

Changes in gut microbiota influence which patients get AIG-related neuroendocrine tumors

Medicaid expansion linked to improved long-term survival in cancer patients

Women with surgical menopause may exit workforce earlier, but hormone therapy could help

Trailblazing Young Scientists honored with $250,000 prizes at Blavatnik National Awards Gala

Revolutionary blood test for ME / Chronic Fatigue unveiled

Calorie labelling linked to 2% average reduction in energy content of menu items

Widely prescribed opioid painkiller tramadol not that effective for easing chronic pain

Exercise snacks may boost cardiorespiratory fitness of physically inactive adults

15,000 women a year with breast cancer could benefit from whole genome sequencing, say researchers

Study highlights risks of Caesarean births to future pregnancies

GLP-1 agonists pose emerging challenge for PET-CT imaging, study finds

Scripps Research scientists unlock new patterns of protein behavior in cell membranes

Panama Canal may face frequent extreme water lows in coming decades

Flash Joule heating lights up lithium extraction from ores

COMBINEDBrain and MUSC announce partnership to establish biorepository for pediatric cerebrospinal fluid and CNS tissue bank

Questionable lead reporting for drinking water virtually vanished after Flint water crisis, study reveals

[Press-News.org] Common feature between forest fires and neural networks reveals the universal framework underneath
Scientists find that scaling laws apply to deep neural networks exhibiting absorbing phase transitions