Published research analyses and original notes
Browse ISOM's current public work, from paper breakdowns to original notes on search quality, publishing, and editorial decisions.
Publisher Pages
Core public pages that explain who operates ISOM, how pages are prepared, and how to request a correction.
About ISOM
Learn what ISOM publishes, how articles are prepared, and why English is the primary public edition.
Editorial Standards
Review disclosure, correction, update, and publication principles for public ISOM pages.
Contact ISOM
Use the public contact page for factual corrections, rights concerns, broken links, and editorial questions.
Site Map
Browse a human-readable map of ISOM's core pages, original notes, and published research analyses.
Transfer Briefs
Short, falsifiable research opportunities that map a scientific structure onto a concrete AI target problem.
Phase-transition thresholds for curriculum scheduling
Use percolation-style threshold estimation to decide when a curriculum should switch regimes instead of relying on fixed epoch cutoffs.
Diffusive deformation priors for medical image synthesis
Treat clinically meaningful image synthesis as transport over deformations and uncertainty, not only as intensity translation.
Variational energy shaping for planning networks
View neural planning modules as energy-shaping systems whose updates should stay inside a feasible value landscape.
Topology-aware distance fields for vascular reconstruction
Use distance-field supervision as a bridge between local geometry and global network validity in structured reconstruction tasks.
Uncertainty-calibrated confidence maps for robust sensing
Make confidence a first-class field that controls inference, not just a diagnostic overlay after prediction.
Paper Analyses
Each card opens the current public analysis for one paper.
Localising entropy production along non-equilibrium trajectories
Entropy production is a universal measure of irreversibility and energy dissipation in physical, chemical, and biological systems operating far from equilibrium.
On the Power of Small-size Graph Neural Networks for Linear Programming
Graph neural networks (GNNs) have recently emerged as powerful tools for addressing complex optimization problems.
Electrically switchable continuous phase liquid crystal Fresnel zone plate
We present the design, fabrication, and characterization of continuous phase Fresnel zone plates (FZPs) using two-photon polymerization direct laser writing in a polymerizable nematic liquid crystal (LC) confined...
Adaptive Stain Normalization for Cross-Domain Medical Histology
Deep learning advances have revolutionized automated digital pathology analysis.
Large phonon-drag thermopower polarity reversal in Ba-doped KTaO3
This study reports the observation of phonon-drag thermopower polarity reversal in Ba-doped KTaO3 thin films, mediated by electron-phonon Umklapp scattering.
Finite disorder critical point in ductile-to-brittle transition in amorphous solids with aspherical impurities
Enhancing the mechanical strength and stability of amorphous solids is crucial for material design, with microalloying being a common yet poorly understood method.
Dynamic behavior analysis of vegetation water model with double saturation transformation term
In this paper, we propose a vegetation-water system incorporating double saturation transformation terms, which more vividly depicts the mutual influence and transformation relationship between vegetation and water.
Rotated Runtime Smooth: Training-Free Activation Smoother for accurate INT4 inference
Large language models have demonstrated promising capabilities upon scaling up parameters.
Federated Model Heterogeneous Matryoshka Representation Learning
Model heterogeneous federated learning (MHeteroFL) enables FL clients to collaboratively train models with heterogeneous structures in a distributed fashion.
Multi-dimensional frequency-bin entanglement-based quantum key distribution network
Quantum networks enhance quantum communication schemes and link multiple users over large areas.
Paralinguistics-Aware Speech-Empowered Large Language Models for Natural Conversation
Recent work shows promising results in expanding the capabilities of large language models (LLM) to directly understand and synthesize speech.
Collective purification of interacting quantum networks beyond symmetry constraints
Following any quantum information processing protocol, it is essential to reset a mixed state of a many-body interacting spin-network to the computational-zero pure state.
Emergence of low-energy spin waves in superconducting electron-doped cuprates
In order to fully utilize the technological potential of unconventional superconductors, an enhanced understanding of the superconducting mechanism is necessary.
RFWave: Multi-band Rectified Flow for Audio Waveform Reconstruction
Recent advancements in generative modeling have significantly enhanced the reconstruction of audio waveforms from various representations.
Anisotropic magnetoresistance and magnetic field-tunable Weyl nodes in Weyl metal SrRuO3 thin films
Weyl semimetals are a unique class of topological materials, possessing Fermi-arc surface states and exhibiting the chiral anomaly effect.
Embedding Trajectory for Out-of-Distribution Detection in Mathematical Reasoning
Real-world data deviating from the independent and identically distributed (\textit{i.i.d.}) assumption of in-distribution training data poses security threats to deep networks, thus advancing out-of-distribution...
Multifunctional fiber-optic theranostic probe for closed-loop tumor photothermal therapy
The combination of optical fiber and phototheranostic agents has emerged as a promising strategy to address the challenges of limited light penetration depth and systemic toxicity of nanomaterials.
Extended forms of Legendre-Laguerre-based hybrid polynomials and their characteristics via fractional operator approach
This study presents an extensive generalization of Legendre–Laguerre polynomials along with their Appell-type counterparts.
MTSAM: Multi-Task Fine-Tuning for Segment Anything Model
The Segment Anything Model (SAM), with its remarkable zero-shot capability, has the potential to be a foundation model for multi-task learning.
Magic Tricycles: Efficient Magic-State Generation with Finite Block-Length Quantum LDPC Codes
The preparation of high-fidelity non-Clifford (magic) states is an essential subroutine for universal quantum computation but imposes substantial space-time overhead.
Differentiated optimal control strategy of brucellosis in Ningxia, China: insights from a two-patch dynamical model
As a high-incidence region of brucellosis in China, the incidence pattern of brucellosis in Ningxia shows a significant spatial-temporal heterogeneity, thus, it is of significance to allocate the differentiated...
QKAN: quantum Kolmogorov-Arnold networks with applications in machine learning and multivariate state preparation
We introduce quantum Kolmogorov-Arnold networks (QKAN), a quantum algorithmic framework inspired by the recently proposed Kolmogorov-Arnold Networks (KAN).
Accelerating atomic fine structure determination with graph reinforcement learning
Atomic data determined by analysis of observed atomic spectra are essential for plasma diagnostics.
Rich dynamics and data analysis of immune decline against SARS-CoV-2
The global pandemic of SARS-CoV-2 has constituted a serious threat to public health.
Generalization Bound and New Algorithm for Clean-Label Backdoor Attack
The generalization bound is a crucial theoretical tool for assessing the generalizability of learning methods and there exist vast literatures on generalizability of normal learning, adversarial learning, and data...
Robust magnetic polaron percolation in the antiferromagnetic CMR system EuCd2P2
The interplay between magnetism and charge transport is central to understanding colossal magnetoresistance (CMR), a phenomenon well studied in ferromagnets.
Layered KIK quantum error mitigation for dynamic circuits
Layered KIK works with mid-circuit measurements & error correction for super reliable quantum computers.
Balloon regime: Drop elasticity leads to complete rebound
New research shows tuning liquid & surface properties prevents splashing at high speeds.
Energy-Guided Continuous Entropic Barycenter Estimation for General Costs
This paper introduces a new, simpler way to average probability distributions that keeps their shape, with guaranteed quality and real-world applications.
Nonlocality, integrability and quantum chaos in the spectrum of bell operators
New research reveals maximal entanglement in 3-state systems leads to predictable, non-chaotic behavior.
Global exponential stability analysis for a delayed diffusive Nicholson’s blowflies equation
A modified Nicholson’s blowflies equation accompanying distinct time-varying delays is established in this paper.
Quantifying aerosol transmission distance for foot-and-mouth disease virus
Foot-and-mouth disease (FMD) is an acute, febrile, and highly contagious animal infectious disease that can be transmitted through multiple routes.
Highway Value Iteration Networks
Value iteration networks (VINs) enable end-to-end learning for planning tasks by employing a differentiable "planning module" that approximates the value iteration algorithm.
Non-Clifford Gates between Stabilizer Codes via Non-Abelian Topological Order
We propose protocols to implement non-Clifford logical gates between stabilizer codes by entangling into a non-Abelian topological order as an intermediate step.
Seeing Beyond the Surface: Retinal Thickness Prediction from Color Fundus Photography for DME Management
New model turns basic eye scans into detailed maps, boosting DME diagnosis in low-resource areas.
Amortizing intractable inference in diffusion models for vision, language, and control
Diffusion models have emerged as effective distribution estimators in vision, language, and reinforcement learning, but their use as priors in downstream tasks poses an intractable posterior inference problem.
D3M: Deformation-Driven Diffusion Model for Synthesis of Contrast-Enhanced MRI with Brain Tumors
Contrast-enhanced magnetic resonance images (CEMRIs) provide valuable information for brain tumor diagnosis and treatment planning.
Magnetic excitations of the Kitaev model candidate RuBr3
New study reveals RuBr3's magnetic interactions push it from ideal spin liquid state, offering clues for quantum computing materials.
Power-efficient ultra-broadband soliton microcombs in resonantly-coupled microresonators
The drive to miniaturize optical frequency combs for practical deployment has spotlighted microresonator solitons as a promising chip-scale candidate.
Negativity percolation in continuous-variable quantum networks
New theory reveals unique "mixed-order" entanglement transitions, paving way for chip-scale quantum tech.
Enhancing the reachability of variational quantum algorithms via input-state design
Design smarter inputs to unlock deeper insights & boost accuracy in quantum algorithms.
Fluxonium as a Control Qubit for Bosonic Quantum Information
Bosonic codes in superconducting resonators are a hardware-efficient avenue for quantum error correction and benefit from the inherent bias toward relaxation errors provided by long-lived cavities compared to typical...
Experimental secure multiparty computation from quantum oblivious transfer with bit commitment
Secure multiparty computation enables collaborative computations across multiple users while preserving individual privacy, which has a wide range of applications in finance, machine learning and healthcare.
Contractive unitary and classical shadow tomography
Here's a breakdown of the abstract, designed for a zero-base reader:
On Hausdorff content maximal operator and Riesz potential for non-measurable functions
We introduce Riesz potentials for Lebesgue non-measurable functions by taking the integrals in the sense of Choquet with respect to Hausdorff content and prove boundedness results for these operators.
Approximation of Mellin convolution-type nonlinear integral operators in variable bounded variation spaces
In this paper, we investigate approximation properties using a family of Mellin convolution type integral operators within the framework of variable bounded variation spaces with the help of summability methods.
DentEval: Fine-tuning-Free Expert-Aligned Assessment in Dental Education via LLM Agents
Large language models (LLMs) have demonstrated considerable potential in automating assignment scoring within higher education, providing efficient and consistent evaluations.
Oscillatory behavior for higher-order nonlinear differential equations in the canonical case
In this paper, we study the oscillation of a class of higher-order neutral nonlinear differential equations.
Improving Quantum Machine Learning via Heat-Bath Algorithmic Cooling
This work introduces an approach rooted in quantum thermodynamics to enhance sampling efficiency in quantum machine learning (QML).
VesselSDF: Distance Field Priors for Vascular Network Reconstruction
VesselSDF uses a new "distance field" approach to perfectly map blood vessels from sparse CT scans, overcoming past limitations.
Efficient Adaptation in Mixed-Motive Environments via Hierarchical Opponent Modeling and Planning
Despite the recent successes of multi-agent reinforcement learning (MARL) algorithms, efficiently adapting to co-players in mixed-motive environments remains a significant challenge.
Scalp Diagnostic System With Label-Free Segmentation and Training-Free Image Translation
ScalpVision tackles data challenges for better, cheaper, and more accessible skin care.
FluoroSAM: A Language-promptable Foundation Model for Flexible X-ray Image Segmentation
Language promptable X-ray image segmentation would enable greater flexibility for human-in-the-loop workflows in diagnostic and interventional precision medicine.
PhoCoLens: Photorealistic and Consistent Reconstruction in Lensless Imaging
New AI reconstructs stunning images from simple sensors, overcoming past limitations.
Dynamical arrest in active nematic turbulence
The study of active fluids, materials driven by internal components such as molecular motors, cells, or synthetic particles, has been a vibrant area of reasearch for decades.
Adversarial Attacks on Combinatorial Multi-Armed Bandits
New research reveals how to identify and exploit vulnerabilities in "reward poisoning" attacks on a type of AI decision making system, showing attacks are harder than previously thought.
A Walsh Hadamard Derived Linear Vector Symbolic Architecture
Vector Symbolic Architectures (VSAs) are one approach to developing Neuro-symbolic AI, where two vectors in are 'bound' together to produce a new vector in the same space.
How Spurious Features are Memorized: Precise Analysis for Random and NTK Features
This paper offers a theoretical explanation for why AI models memorize irrelevant data, revealing how model stability and feature alignment play key roles.
Beyond Shadows: Learning Physics-inspired Ultrasound Confidence Maps from Sparse Annotations
This paper introduces a novel user-centered approach for generating confidence maps in ultrasound imaging.
Vector-Quantization-Driven Active Learning for Efficient Multi-Modal Medical Segmentation with Cross-Modal Assistance
Multi-modal medical image segmentation leverages complementary information across different modalities to enhance diagnostic accuracy, but faces two critical challenges: the requirement for extensive paired...
TAPNext: Tracking Any Point (TAP) as Next Token Prediction
The problem of "correspondence" has been a foundational challenge in computer vision for decades.
milliMamba: Specular-Aware Human Pose Estimation via Dual mmWave Radar with Multi-Frame Mamba Fusion
The problem of Human Pose Estimation (HPE) using millimeter wave (mmWave) radar signals emerged primarily as a response to the limitations of traditional camera based (RGB) systems.
Generative Video Propagation
The problem of generative video propagation, as addressed in this paper, is rooted in the broader field of computer vision, specifically within the domain of video generation and editing.
INST-IT: Boosting Instance Understanding via Explicit Visual Prompt Instruction Tuning
The problem addressed in this paper precisely originates from the recent advancements and, paradoxically, the limitations of Large Multimodal Models (LMMs) in the field of artificial intelligence, specifically within...
All-in-one medical image-to-image translation
Unifies translation, offers semantic control, and works without fine-tuning.
Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow
Rectified Flow learns straight paths to efficiently generate and transfer data between distributions.
RedDino: A foundation model for red blood cell analysis
RedDino analyzes red blood cell images with unprecedented accuracy, paving the way for faster disease diagnosis.
Patient-specific radiomic feature selection with reconstructed healthy persona of knee MR images
New method combines interpretable "radiomic features" with AI-generated "healthy scans" for better, explainable medical image analysis.
MeDi: Metadata-Guided Diffusion Models for Mitigating Biases in Tumor Classification
Deep learning models have made significant advances in histological prediction tasks in recent years.
Multi-Level Gated U-Net for Denoising TMR Sensor-Based MCG Signals
New AI model dramatically cleans up heart signals from cheap sensors for better medical use.
Hierarchical Part-based Generative Model for Realistic 3D Blood Vessel
Advancements in 3D vision have increased the impact of blood vessel modeling on medical applications.
Prompt-DAS: Annotation-Efficient Prompt Learning for Domain Adaptive Semantic Segmentation of Electron Microscopy Images
Prompt DAS adapts AI to segment tiny cell parts in electron microscope images, offering flexible, efficient, and interactive annotation.
Explainable ADHD Diagnostic Framework Using Weakly-Supervised Action Recognition
The clinical diagnosis of Attention Deficit Hyperactivity Disorder (ADHD) primarily relies on scale questionnaires, clinical interviews, and executive function tests, which face challenges including limited medical...
LiteTracker: Leveraging Temporal Causality for Accurate Low-latency Tissue Tracking
LiteTracker achieves lightning fast, accurate endoscopic tissue tracking for real time surgery.
Regularized Low-Rank Adaptation for Few-Shot Organ Segmentation
New method auto-adjusts rank for better segmentation, outperforming others in few-shot learning.
Hybrid Graph Mamba: Unlocking Non-Euclidean Potential for Accurate Polyp Segmentation
Colorectal polyp segmentation can assist doctors in screening colonoscopy images, which is crucial for the prevention of colorectal cancer.
SOO-Bench: Benchmarks for Evaluating the Stability of Offline Black-Box Optimization
The problem of Offline Black-Box Optimization (BBO) emerged from the practical necessity of optimizing complex systems where direct, real-time evaluation of the objective function is either too dangerous,...
Towards Generalizable 3D Human Pose Estimation via Ensembles on Flat Loss Landscapes
The quest to understand human movement in three dimensions from simple two-dimensional images—like those from a standard smartphone camera—is a cornerstone of modern computer vision.
Compute-Constrained Data Selection
The field of large language models (LLMs) has seen explosive growth, leading to models with billions of parameters capable of remarkable feats in natural language understanding and generation.
Single Image Test-Time Adaptation via Multi-View Co-Training
Test-time adaptation enables a trained model to adjust to a new domain during inference, making it particularly valuable in clinical settings where such on-the-fly adaptation is required.
CENet: Context Enhancement Network for Medical Image Segmentation
CENet boosts medical image segmentation by enhancing boundaries and preserving details across diverse image types.
Temporal Atlas-Guided Generation of Longitudinal Data via Geometric Latent Embeddings
This paper introduces a new AI model that creates realistic "time lapse" medical images from static scans, helping us understand how body parts grow and change.
Multi-Tube-Voltage vBMD Measurement via Dual-Branch Frequency Balancing and Asymmetric Channel Attention
Phantom-less volumetric bone mineral density (vBMD) measurement using computed tomography (CT) presents a cost-effective alternative to conventional phantom-based approaches, yet faces accuracy challenges across...
The Refining of Brain Connectivity Features on Residual Posterior Patterns
New AI model RP-LGN captures subtle connectivity changes for better disease diagnosis, outperforming others with improved accuracy & noise handling.
One-shot active learning for vessel segmentation
New AI learns from tiny, smart samples, saving time & resources for better disease insights.
Anatomical Structure Few-Shot Detection Utilizing Enhanced Human Anatomy Knowledge in Ultrasound Images
Deep learning-based models have significantly advanced clinical ultrasound tasks by detecting anatomical structures within vast ultrasound image datasets.
Cross-Modal Brain Graph Transformer via Function-Structure Connectivity Network for Brain Disease Diagnosis
Multi-modal brain networks represent the complex connectivity between different brain regions from both functional and structural perspectives, which is of great significance for brain disease diagnosis.
Counterfactual Explanations for Conformal Prediction Sets
New counterfactual explanations make complex "prediction sets" from AI understandable by showing minimal changes that alter the AI's output.
Pre-to-Post Operative MRI Generation with Retrieval-based Visual In-Context Learning
New AI generates realistic post op MRIs from pre op scans, aiding brain tumor surgery.
Radar-Based Imaging for Sign Language Recognition in Medical Communication
This paper introduces a privacy preserving radar system for recognizing Italian Sign Language in medical settings, achieving high accuracy.
Queue Test From Local PDF
Physics-informed neural networks (PINN) have achieved notable success in solving partial differential equations (PDE), yet solving the Navier-Stokes equations (NSE) with complex boundary conditions remains a...
Learning to increase matching efficiency in identifying additional b-jets in the process
To truly understand the significance of this paper, we have to travel back to the monumental discovery of the Higgs boson at the Large Hadron Collider (LHC) in 2012.
Hybrid Boundary Physics-Informed Neural Networks for Solving Navier-Stokes Equations with Complex Boundary
Physics-informed neural networks (PINN) have achieved notable success in solving partial differential equations (PDE), yet solving the Navier-Stokes equations (NSE) with complex boundary conditions remains a...
Wavelet-driven Decoupling and Physics-informed Mapping Network for Accelerated Multi-parametric MR Imaging
To understand the origin of this problem, we have to look at how doctors look inside the human body.
ISOM Notes
Original notes on workflow, ranking quality, and editorial decisions at ISOM.
Translation Flow for Global Research Posts
How ISOM keeps English source analyses, draft translations, and public publication states aligned without exposing every generated page by default.
Ranking Features for Isomorphic Search
How ISOM combines relevance, freshness, source quality, and editorial confidence so search results surface genuinely useful research first.