arXiv subjects

Daisuke Yamada

Publications and source records attributed to Daisuke Yamada.

13 recordsLinked to original sources

arXiv2026

Tree-Structured Orthonormal Decomposition of the Aitchison Simplex

Compositional data -- vectors encoding relative proportions -- arise across scientific domains, including ecology, geochemistry, and genomics. The features in these data often come with known hierarchical structure (e.g., taxonomies, phylogenies, ontologies), yet existing methods either ignore this structure, discard the intrinsic Aitchison geometry, are designed for binary trees, or yield incomplete coordinate systems. We describe PolyILR, a canonical orthonormal decomposition of the Aitchison tangent space aligned with any tree topology. Our construction defines a weighted local geometry at each internal node capturing full branching structure, then lifts these to a global orthonormal basis where every coordinate corresponds to a specific tree location. On microbiome and single-cell benchmarks, PolyILR yields stable, interpretable features and enables inference at multiscale tree resolution. We also establish a novel theoretical connection to softmax classifiers, suggesting possible applications to probabilistic modeling.

cs.LG↗

arXiv2026

Recursive Binding on a Budget: Subspace Carving in Order-p Tensor Memories

Tensor Product Representations provide the structural fidelity required for symbolic reasoning in models but suffer from exponential dimensionality growth when encoding deep recursive structures. Conversely, Vector Symbolic Architectures maintain constant dimensionality but sacrifice capacity and fidelity due to noisy compression via superposition. In this work, we propose Orthogonal Subspace Carving (OSC), a memory architecture that binds fillers to roles by projecting onto the null space of the role basis before aggregating into a fixed order-p tensor. OSC uses projections to enforce geometric orthogonality between bound structures within a static memory trace. We show that this mechanism decouples the tensor order from the structural depth, enabling deep recursive binding within a constant memory footprint. By performing retrieval via recognition, this construction allows for component vectors that are orders of magnitude smaller than the memory tensor, giving superior memory efficiency in settings involving high superposition. We also show that TPR is a special case of binding in Clifford algebra, and give a Clifford formulation of OSC.

cs.LG↗

arXiv2026

Association of Progressive PPFE and Mortality in Lung Cancer Screening Cohorts

Background: Pleuroparenchymal fibroelastosis (PPFE) is an upper lobe predominant fibrotic lung abnormality associated with increased mortality in established interstitial lung disease. However, the clinical significance of radiologic PPFE progression in lung cancer screening (LCS) populations remains unclear. Methods: We analysed longitudinal low-dose CT scans and clinical data from two LCS studies: National Lung Screening Trial (NLST; n=7,980); SUMMIT study (n=8,561). An automated algorithm quantified PPFE volume on baseline and follow-up scans. Annualised change in PPFE was derived and dichotomised using a distribution-based threshold to define progressive PPFE. Associations between progressive PPFE and mortality were evaluated using Cox proportional hazards models adjusted for demographic and clinical variables. In SUMMIT cohort, associations between progressive PPFE and clinical outcomes were assessed using incidence rate ratios (IRR) and odds ratios (OR). Findings: Progressive PPFE independently associated with mortality in both LCS cohorts (NLST: Hazard Ratio (HR)=1.25, 95% Confidence Interval (CI): 1.01--1.56, p=0.042; SUMMIT: HR=3.14, 95% CI: 1.66--5.97, p<0.001). Within SUMMIT, progressive PPFE was strongly associated with higher respiratory admissions (IRR=2.79, p<0.001), increased antibiotic and steroid use (IRR=1.55, p=0.010), and showed a trend towards higher modified medical research council scores (OR=1.40, p=0.055). Interpretation: Radiologic PPFE progression independently associates with mortality across two large LCS cohorts, and associates with adverse clinical outcomes. Quantitative assessment of PPFE progression may provide a clinically relevant imaging biomarker to identify individuals at increased risk of respiratory morbidity within LCS programmes.

q-bio.QM↗

arXiv2025

Composing Linear Layers from Irreducibles

Contemporary large models often exhibit behaviors suggesting the presence of low-level primitives that compose into modules with richer functionality, but these fundamental building blocks remain poorly understood. We investigate this compositional structure in linear layers by asking: can we identify/synthesize linear transformations from a minimal set of geometric primitives? Using Clifford algebra, we show that linear layers can be expressed as compositions of bivectors -- geometric objects encoding oriented planes -- and introduce a differentiable algorithm that decomposes them into products of rotors. This construction uses only O(log^2 d) parameters, versus O(d^2) required by dense matrices. Applied to the key, query, and value projections in LLM attention layers, our rotor-based layers match the performance of strong baselines such as block-Hadamard and low-rank approximations. Our findings provide an algebraic perspective on how these geometric primitives can compose into higher-level functions within deep models.

cs.LG↗

arXiv2025

A computationally frugal open-source foundation model for thoracic disease detection in lung cancer screening programs

Low-dose computed tomography (LDCT) imaging employed in lung cancer screening (LCS) programs is increasing in uptake worldwide. LCS programs herald a generational opportunity to simultaneously detect cancer and non-cancer-related early-stage lung disease. Yet these efforts are hampered by a shortage of radiologists to interpret scans at scale. Here, we present TANGERINE, a computationally frugal, open-source vision foundation model for volumetric LDCT analysis. Designed for broad accessibility and rapid adaptation, TANGERINE can be fine-tuned off the shelf for a wide range of disease-specific tasks with limited computational resources and training data. Relative to models trained from scratch, TANGERINE demonstrates fast convergence during fine-tuning, thereby requiring significantly fewer GPU hours, and displays strong label efficiency, achieving comparable or superior performance with a fraction of fine-tuning data. Pretrained using self-supervised learning on over 98,000 thoracic LDCTs, including the UK's largest LCS initiative to date and 27 public datasets, TANGERINE achieves state-of-the-art performance across 14 disease classification tasks, including lung cancer and multiple respiratory diseases, while generalising robustly across diverse clinical centres. By extending a masked autoencoder framework to 3D imaging, TANGERINE offers a scalable solution for LDCT analysis, departing from recent closed, resource-intensive models by combining architectural simplicity, public availability, and modest computational requirements. Its accessible, open-source lightweight design lays the foundation for rapid integration into next-generation medical imaging tools that could transform LCS initiatives, allowing them to pivot from a singular focus on lung cancer detection to comprehensive respiratory disease management in high-risk populations.

eess.IV↗

arXiv2025

ASAT: Adaptive Scoring and Thresholding with Human Feedback for Robust Out-of-Distribution Detection

Machine Learning (ML) models are trained on in-distribution (ID) data but often encounter out-of-distribution (OOD) inputs during deployment---posing serious risks in safety-critical domains. Recent works have focused on designing scoring functions to quantify OOD uncertainty, with score thresholds typically set based solely on ID data to achieve a target true positive rate (TPR), since OOD data is limited before deployment. However, these TPR-based thresholds leave false positive rates (FPR) uncontrolled, often resulting in high FPRs where OOD points are misclassified as ID. Moreover, fixed scoring functions and thresholds lack the adaptivity needed to handle newly observed, evolving OOD inputs, leading to sub-optimal performance. To address these challenges, we propose \emph{ASAT}, a human-in-the-loop framework that \textit{safely updates both scoring functions and thresholds on the fly} based on real-world OOD inputs. ASAT maximizes TPR while controlling FPR at all times under stationary conditions, even as the system adapts over time. Under nonstationary conditions, the method adapts to distribution shifts with only transient FPR violations during the adaptation period. We provide theoretical guarantees for FPR control under stationary conditions and present extensive empirical evaluations on OpenOOD benchmarks to demonstrate that our approach outperforms existing methods by achieving higher TPRs while maintaining FPR control.

cs.LG↗

arXiv2007

Metastability of R-Charged Black Holes

The global stability of R-charged AdS black holes in a grand canonical ensemble is examined by eliminating the constraints from the action, but without solving the equations of motion, thereby constructing the reduced action of the system. The metastability of the system is found to set in at a critical value of the chemical potential which is conjugate to the R-charge. The relation among the small black hole, large black hole and the instability is discussed. The result is consistent with the metastability found in the AdS/CFT-conjectured dual field theory. The "renormalized" temperature of AdS black holes, which has been rather ad hoc, is suggested to be the boundary temperature in the sense of AdS/CFT correspondence. As a byproduct of the analysis, we find a more general solution of the theory and its properties are briefly discussed.

hep-th↗

arXiv2007

Scattering Rule in Soliton Cellular Automaton associated with Crystal Base of $U_q(D_4^{(3)})$

In terms of the crystal base of a quantum affine algebra $U_q(\mathfrak{g})$, we study a soliton cellular automaton (SCA) associated with the exceptional affine Lie algebra $\mathfrak{g}=D_4^{(3)}$. The solitons therein are labeled by the crystals of quantum affine algebra $U_q(A_1^{(1)})$. The scatteing rule is identified with the combinatorial $R$ matrix for $U_q(A_1^{(1)})$-crystals. Remarkably, the phase shifts in our SCA are given by {\em 3-times} of those in the well-known box-ball system.

math.QA↗

arXiv2007

Box ball system associated with antisymmetric tensor crystals

A new box ball system associated with an antisymmetric tensor crystal of the quantum affine algebra of type A is considered. This includes the so-called colored box ball system with capacity 1 as the simplest case. Infinite number of conserved quantities are constructed and the scattering rule of two olitons are given explicitly.

math.QA↗

arXiv2006

Perfect Crystals for U_q(D_4^{(3)})

A perfect crystal of any level is constructed for the Kirillov-Reshetikhin module of $U_q(D_4^{(3)})$ corresponding to the middle vertex of the Dynkin diagram. The actions of Kashiwara operators are given explicitly. It is also shown that this family of perfect crystals is coherent. A uniqueness problem solved in this paper can be applied to other quantum affine algebras.

math.QA↗

arXiv2006

Phase Diagram of N=4 Super-Yang-Mills Theory with R-Symmetry Chemical Potentials

The phase diagram of large Nc, weakly-coupled N=4 supersymmetric Yang-Mills theory on a three-sphere with non-zero chemical potentials is examined. In the zero coupling limit, a transition line in the mu-T plane is found, separating a "confined" phase in which the Polyakov loop has vanishing expectation value from a "deconfined" phase in which this order parameter is non-zero. For non-zero but weak coupling, perturbative methods may be used to construct a dimensionally reduced effective theory valid for sufficiently high temperature. If the maximal chemical potential exceeds a critical value, then the free energy becomes unbounded below and no genuine equilibrium state exists. However, the deconfined plasma phase remains metastable, with a lifetime which grows exponentially with Nc (not Nc^2). This metastable phase persists with increasing chemical potential until a phase boundary, analogous to a spinodal decomposition line, is reached. Beyond this point, no long-lived locally stable quasi-equilibrium state exists. The resulting picture for the phase diagram of the weakly coupled theory is compared with results believed to hold in the strongly coupled limit of the theory, based on the AdS/CFT correspondence and the study of charged black hole thermodynamics. The confinement/deconfinement phase transition at weak coupling is in qualitative agreement with the Hawking-Page phase transition in the gravity dual of the strongly coupled theory. The black hole thermodynamic instability line may be the counterpart of the spinodal decomposition phase boundary found at weak coupling, but no black hole tunneling instability, analogous to the instability of the weakly coupled plasma phase is currently known.

hep-th↗

arXiv2004

CFL Phase of High Density QCD at Non Zero Strange Quark Mass

We compute free energy of quark matter at asymptotically high baryon number density in the presence of non zero strange quark mass including dynamics of pseudo Nambu-Goldstone bosons due to chiral symmetry breaking, extending previously existing analysis based on perturbative expansion in $m_s^2/4μΔ.$ We demonstrate that the CFL$K^0$ state has lower free energy than the symmetric CFL state for $0<m_s^2/4μΔ<2/3$. We also calculate the spectrum of the fermionic quasiparticle excitations about the kaon condensed ground state in the regime $m_s^2/4μΔ\sim 1$ and find that $(m_s^2/4μΔ)_{crit}=2/3$ for the CFL-gCFL phase transition, the leading order result reported in [1], is not modified.

hep-ph↗

arXiv2003

Construction of bosonic string theory on infinitely curved Anti-de Sitter space

Free scalar field theory in the sector with a large number of particles can be interpreted as bosonic string theory on anti-de Sitter space of vanishing radius. Different ways of writing the field theory Hamiltonian translate to different ways of reparametrizing the world-sheet sigma coordinate. Adding a mass term in the field theory corresponds to cutting off the warped AdS direction, with cut-off inversely proportional to the mass. The string theory has neither tachyon, nor critical dimension.

hep-th↗