arXiv subjects

Yaroslav Kholodov

Publications and source records attributed to Yaroslav Kholodov.

13 recordsLinked to original sources

arXiv2026

Task-Aware Hybrid QUBO Optimization for Structured Neural Network Pruning

Neural network pruning can be formulated as a combinatorial optimization problem, yet many existing approaches rely on independent filter-importance scores or simplified objective functions. In this work, we propose a Hybrid Quadratic Unconstrained Binary Optimization (QUBO) framework for structured filter pruning that combines task-aware sensitivity information with interactions between candidate filters. The formulation incorporates first-order Taylor sensitivity and Weight-Fisher sensitivity into the linear component of the objective and can additionally incorporate activation similarity into the quadratic interactions. To control the target pruning cardinality without introducing an explicit quadratic cardinality penalty, we use a binary search over the capacity incentive to identify a coefficient that empirically yields the target pruning cardinality. We further investigate a two-stage QUBO--Tensor-Train refinement strategy in which the QUBO solution initializes gradient-free probabilistic black-box optimization to search for improved pruning masks using the downstream metric. Experiments on the SIDD image denoising task and a Half-UNet model show that the Hybrid QUBO achieves higher PSNR and SSIM than the evaluated Taylor and L1-based QUBO baselines at the studied pruning target. Multi-seed experiments under a fixed dataset protocol are used to assess robustness, while controlled sub-problem experiments demonstrate that Tensor-Train refinement becomes increasingly valuable as the combinatorial problem size grows. The results support Hybrid QUBO as a task-aware structured pruning framework for the evaluated setting, while also highlighting the computational and deployment limitations of mask-based pruning.

cs.LG↗

arXiv2026

Noninvasive H3 K27M screening in pediatric diffuse midline glioma using radiomics on heterogeneous T2-weighted MRI

Histone H3K27M mutation status defines a clinically aggressive subgroup of pediatric diffuse midline glioma and informs prognosis and trial eligibility, but confirmation usually requires tissue sampling from eloquent midline structures. We evaluated whether radiomics from routinely available T2-weighted MRI can provide an adjunctive screening signal in a heterogeneous referral-style cohort, where scans are often acquired externally and T2-weighted imaging is the only consistently available sequence. Ninety-eight pediatric patients with tissue-confirmed status were analyzed (73 mutation-positive, 25 wild-type). Expert tumor segmentations defined the regions of interest for PyRadiomics feature extraction after isotropic resampling, dual skull stripping, and multi-scale filtering. We systematically ablated preprocessing, correlation pruning with repeated recursive feature elimination, tumor volume, and TabDDPM synthetic minority augmentation across 100 stratified train/test splits with real-only test sets. Pure radiomics achieved accuracy 0.664 and F1-score 0.784. The best pipeline used preprocessing, feature selection, and volume with CatBoost, achieving accuracy 0.730$\pm$0.068 and F1-score 0.826$\pm$0.044. TabDDPM improved TabPFN to F1-score 0.81$\pm$0.05 at 200 augmented rows. These results support T2-weighted radiomics as a moderate screening and triage aid, not a replacement for tissue-based diagnosis.

q-bio.QM↗

arXiv2026

Higher-Order vs. Quadratic Binary Optimization: Which Is Better for Probability Optimization with Tensor Sampling?

This article explores the comparative strengths of Higher-Order Unconstrained Binary Optimization (HUBO) and Quadratic Unconstrained Binary Optimization (QUBO) models in the context of probability optimization using tensor sampling techniques. HUBO can represent interactions beyond pairwise terms without introducing auxiliary variables, while QUBO benefits from a simpler algebraic structure and a wide range of established solvers. We combine a theoretical analysis of formulation expressiveness and reduction overhead with empirical experiments on synthetic polynomial objectives, RSA factorization, and Max-Cut instances. The results show that native HUBO formulations are often preferable when the original problem contains high-order interactions, because reducing such problems to QUBO increases the effective dimension seen by tensor samplers. At the same time, QUBO remains competitive for naturally quadratic problems. These findings provide practical guidance for selecting an optimization framework that balances model fidelity, dimension growth, and solver performance.

math.OC↗

arXiv2026

Neural Network Pruning via QUBO Optimization

Neural network pruning can be formulated as a combinatorial optimization problem, yet most existing approaches rely on greedy heuristics that ignore complex interactions between filters. Formal optimization methods such as Quadratic Unconstrained Binary Optimization (QUBO) provide a principled alternative but have so far underperformed due to oversimplified objective formulations based on metrics like the L1-norm. In this work, we propose a unified Hybrid QUBO framework that bridges heuristic importance estimation with global combinatorial optimization. Our formulation integrates gradient-aware sensitivity metrics - specifically first-order Taylor and second-order Fisher information - into the linear term, while utilizing data-driven activation similarity in the quadratic term. This allows the QUBO objective to jointly capture individual filter relevance and inter-filter functional redundancy. We further introduce a dynamic capacity-driven search to strictly enforce target sparsity without distorting the optimization landscape. Finally, we employ a two-stage pipeline featuring a Tensor-Train (TT) Refinement stage - a gradient-free optimizer that fine-tunes the QUBO-derived solution directly against the true evaluation metric. Experiments on the SIDD image denoising dataset demonstrate that the proposed Hybrid QUBO significantly outperforms both greedy Taylor pruning and traditional L1-based QUBO, with TT Refinement providing further consistent gains at appropriate combinatorial scales. This highlights the potential of hybrid combinatorial formulations for robust, scalable, and interpretable neural network compression.

cs.CV↗

arXiv2026

SCOPE: Smooth Convex Optimization for Planned Evolution of Deformable Linear Objects

We present SCOPE, a fast and efficient framework for modeling and manipulating deformable linear objects (DLOs). Unlike conventional energy-based approaches, SCOPE leverages convex approximations to significantly reduce computational cost while maintaining smooth and physically plausible deformations. This trade-off between speed and accuracy makes the method particularly suitable for applications requiring real-time or near-real-time response. The effectiveness of the proposed framework is demonstrated through comprehensive simulation experiments, highlighting its ability to generate smooth shape trajectories under geometric and length constraints.

cs.RO↗

arXiv2026

Quantum-Inspired Episode Selection for Monte Carlo Reinforcement Learning via QUBO Optimization

Monte Carlo (MC) reinforcement learning suffers from high sample complexity, especially in environments with sparse rewards, large state spaces, and correlated trajectories. We address these limitations by reformulating episode selection as a Quadratic Unconstrained Binary Optimization (QUBO) problem and solving it with quantum-inspired samplers. Our method, MC+QUBO, integrates a combinatorial filtering step into standard MC policy evaluation: from each batch of trajectories, we select a subset that maximizes cumulative reward while promoting state-space coverage. This selection is encoded as a QUBO, where linear terms favor high-reward episodes and quadratic terms penalize redundancy. We explore both Simulated Quantum Annealing (SQA) and Simulated Bifurcation (SB) as black-box solvers within this framework. Experiments in a finite-horizon GridWorld demonstrate that MC+QUBO outperforms vanilla MC in convergence speed and final policy quality, highlighting the potential of quantum-inspired optimization as a decision-making subroutine in reinforcement learning.

cs.LG↗

arXiv2025

Training-Free Out-Of-Distribution Segmentation With Foundation Models

Detecting unknown objects in semantic segmentation is crucial for safety-critical applications such as autonomous driving. Large vision foundation models, including DINOv2, InternImage, and CLIP, have advanced visual representation learning by providing rich features that generalize well across diverse tasks. While their strength in closed-set semantic tasks is established, their capability to detect out-of-distribution (OoD) regions in semantic segmentation remains underexplored. In this work, we investigate whether foundation models fine-tuned on segmentation datasets can inherently distinguish in-distribution (ID) from OoD regions without any outlier supervision. We propose a simple, training-free approach that utilizes features from the InternImage backbone and applies K-Means clustering alongside confidence thresholding on raw decoder logits to identify OoD clusters. Our method achieves 50.02 Average Precision on the RoadAnomaly benchmark and 48.77 on the benchmark of ADE-OoD with InternImage-L, surpassing several supervised and unsupervised baselines. These results suggest a promising direction for generic OoD segmentation methods that require minimal assumptions or additional data.

cs.CV↗

arXiv2025

Adaptive Backdoor Attacks with Reasonable Constraints on Graph Neural Networks

Recent studies show that graph neural networks (GNNs) are vulnerable to backdoor attacks. Existing backdoor attacks against GNNs use fixed-pattern triggers and lack reasonable trigger constraints, overlooking individual graph characteristics and rendering insufficient evasiveness. To tackle the above issues, we propose ABARC, the first Adaptive Backdoor Attack with Reasonable Constraints, applying to both graph-level and node-level tasks in GNNs. For graph-level tasks, we propose a subgraph backdoor attack independent of the graph's topology. It dynamically selects trigger nodes for each target graph and modifies node features with constraints based on graph similarity, feature range, and feature type. For node-level tasks, our attack begins with an analysis of node features, followed by selecting and modifying trigger features, which are then constrained by node similarity, feature range, and feature type. Furthermore, an adaptive edge-pruning mechanism is designed to reduce the impact of neighbors on target nodes, ensuring a high attack success rate (ASR). Experimental results show that even with reasonable constraints for attack evasiveness, our attack achieves a high ASR while incurring a marginal clean accuracy drop (CAD). When combined with the state-of-the-art defense randomized smoothing (RS) method, our attack maintains an ASR over 94%, surpassing existing attacks by more than 7%.

cs.LG↗

arXiv2024

Local Methods with Adaptivity via Scaling

The rapid development of machine learning and deep learning has introduced increasingly complex optimization challenges that must be addressed. Indeed, training modern, advanced models has become difficult to implement without leveraging multiple computing nodes in a distributed environment. Distributed optimization is also fundamental to emerging fields such as federated learning. Specifically, there is a need to organize the training process to minimize the time lost due to communication. A widely used and extensively researched technique to mitigate the communication bottleneck involves performing local training before communication. This approach is the focus of our paper. Concurrently, adaptive methods that incorporate scaling, notably led by Adam, have gained significant popularity in recent years. Therefore, this paper aims to merge the local training technique with the adaptive approach to develop efficient distributed learning methods. We consider the classical Local SGD method and enhance it with a scaling feature. A crucial aspect is that the scaling is described generically, allowing us to analyze various approaches, including Adam, RMSProp, and OASIS, in a unified manner. In addition to theoretical analysis, we validate the performance of our methods in practice by training a neural network.

cs.LG↗

arXiv2024

Activations and Gradients Compression for Model-Parallel Training

Large neural networks require enormous computational clusters of machines. Model-parallel training, when the model architecture is partitioned sequentially between workers, is a popular approach for training modern models. Information compression can be applied to decrease workers communication time, as it is often a bottleneck in such systems. This work explores how simultaneous compression of activations and gradients in model-parallel distributed training setup affects convergence. We analyze compression methods such as quantization and TopK compression, and also experiment with error compensation techniques. Moreover, we employ TopK with AQ-SGD per-batch error feedback approach. We conduct experiments on image classification and language model fine-tuning tasks. Our findings demonstrate that gradients require milder compression rates than activations. We observe that $K=10\%$ is the lowest TopK compression level, which does not harm model convergence severely. Experiments also show that models trained with TopK perform well only when compression is also applied during inference. We find that error feedback techniques do not improve model-parallel training compared to plain compression, but allow model inference without compression with almost no quality drop. Finally, when applied with the AQ-SGD approach, TopK stronger than with $ K=30\%$ worsens model performance significantly.

cs.LG↗

arXiv2023

An evolutionary view on equilibrium models of transport flows

In this short paper we describe natural logit population games dynamics that explain equilibrium models of origin-destination matrix estimation and (stochastic) traffic assignment models (Beckmann, Nesterov--de Palma). Composition of the proposed dynamics allows to explain two-stages traffic assignment models.

math.OC↗

arXiv2016

Protein-protein docking by generalized Fourier transforms on 5D rotational manifolds

Energy evaluation using fast Fourier transforms enables sampling billions of putative complex structures and hence revolutionized rigid protein-protein docking. However, in current methods efficient acceleration is achieved only in either the translational or the rotational subspace. Developing an efficient and accurate docking method that expands FFT based sampling to 5 rotational coordinates is an extensively studied but still unsolved problem. The algorithm presented here retains the accuracy of earlier methods but yields at least tenfold speedup. The improvement is due to two innovations. First, the search space is treated as the product manifold $\mathbf{SO(3)x(SO(3)\setminus S^1)}$, where $\mathbf{SO(3)}$ is the rotation group representing the space of the rotating ligand, and $\mathbf{(SO(3)\setminus S^1)}$ is the space spanned by the two Euler angles that define the orientation of the vector from the center of the fixed receptor toward the center of the ligand. This representation enables the use of efficient FFT methods developed for $\mathbf{SO(3)}$. Second, we select the centers of highly populated clusters of docked structures, rather than the lowest energy conformations, as predictions of the complex, and hence there is no need for very high accuracy in energy evaluation. Therefore it is sufficient to use a limited number of spherical basis functions in the Fourier space, which increases the efficiency of sampling while retaining the accuracy of docking results. A major advantage of the method is that, in contrast to classical approaches, increasing the number of correlation function terms is computationally inexpensive, which enables using complex energy functions for scoring.

q-bio.BM↗

arXiv2015

Tubulin bond energies and microtubule biomechanics determined from nanoindentation in silico

Microtubules, the primary components of the chromosome segregation machinery, are stabilized by longitudinal and lateral non-covalent bonds between the tubulin subunits. However, the thermodynamics of these bonds and the microtubule physico-chemical properties are poorly understood. Here, we explore the biomechanics of microtubule polymers using multiscale computational modeling and nanoindentations in silico of a contiguous microtubule fragment. A close match between the simulated and experimental force-deformation spectra enabled us to correlate the microtubule biomechanics with dynamic structural transitions at the nanoscale. Our mechanical testing revealed that the compressed MT behaves as a system of rigid elements interconnected through a network of lateral and longitudinal elastic bonds. The initial regime of continuous elastic deformation of the microtubule is followed by the transition regime, during which the microtubule lattice undergoes discrete structural changes, which include first the reversible dissociation of lateral bonds followed by irreversible dissociation of the longitudinal bonds. We have determined the free energies of dissociation of the lateral (6.9+/-0.4 kcal/mol) and longitudinal (14.9+/-1.5 kcal/mol) tubulin-tubulin bonds. These values in conjunction with the large flexural rigidity of tubulin protofilaments obtained (18,000-26,000 pN*nm^2), support the idea that the disassembling microtubule is capable of generating a large mechanical force to move chromosomes during cell division. Our computational modeling offers a comprehensive quantitative platform to link molecular tubulin characteristics with the physiological behavior of microtubules. The developed in silico nanoindentation method provides a powerful tool for the exploration of biomechanical properties of other cytoskeletal and multiprotein assemblies

physics.bio-ph↗