# catboost

CatBoost is an open-source machine-learning library and algorithm that uses gradient boosting on decision trees, including support for categorical features.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## GridForge: An Open Physics-Informed Digital Twin and Benchmark Suite for UK 11 kV Underground Distribution Cables

DevFeed: [GridForge: An Open Physics-Informed Digital Twin and Benchmark Suite for UK 11 kV Underground Distribution Cables](<https://devfeed.tech/articles/gridforge-an-open-physics-informed-digital-twin-and-benchmark-suite-for-uk-11-kv-underground-distribution-cables-143366.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23241>)

Author: Lilliane Linnet Musoke

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Dataset development](<https://devfeed.tech/topics/dataset-development.md>), [Open Source Models & Datasets](<https://devfeed.tech/topics/open-source-models-datasets.md>), [catboost](<https://devfeed.tech/topics/catboost.md>), [Open Source](<https://devfeed.tech/topics/open-source.md>), [Digital Public Good](<https://devfeed.tech/topics/digital-public-goods.md>)

Tags: [benchmark](<https://devfeed.tech/tags/benchmark.md>), [benchmark-dataset](<https://devfeed.tech/tags/benchmark-dataset.md>), [benchmarking](<https://devfeed.tech/tags/benchmarking.md>), [cables](<https://devfeed.tech/tags/cables.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [diagnosis](<https://devfeed.tech/tags/diagnosis.md>), [digital-twin](<https://devfeed.tech/tags/digital-twin.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [gradient-boosted-trees](<https://devfeed.tech/tags/gradient-boosted-trees.md>), [maintenance](<https://devfeed.tech/tags/maintenance.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [neural-network](<https://devfeed.tech/tags/neural-network.md>), [open](<https://devfeed.tech/tags/open.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [operational](<https://devfeed.tech/tags/operational.md>), [physics-informed](<https://devfeed.tech/tags/physics-informed.md>), [public](<https://devfeed.tech/tags/public.md>), [standards](<https://devfeed.tech/tags/standards.md>), [uk](<https://devfeed.tech/tags/uk.md>)

### AI overview

GridForge provides an open-source benchmark for fault diagnosis and prediction in UK 11 kV underground cables. It combines IEC 60287/60853 thermal models, a calibrated synthetic dataset covering 64 cables and 320 cable-years, and a PINN surrogate that enforces IEC 60287 algebraic balance. The PINN reports 0.148 °C RMSE against the analytical reference on a held-out set and trained in two minutes on the stated platform. Five benchmark tasks include pure-physics, gradient-boosted-tree, and PINN baselines, but only the virtual-sensor task distinguishes them on the reported test split; other tasks tie or are undefined. The dataset, test sets, models, and MIT-licensed code are made available for validation.

### Source excerpt

arXiv:2611.23241v1 Announce Type: new Abstract: The UK distribution-cable monitoring sector is data-rich but framework-poor: each Distribution Network Operator (DNO) collects megabytes of telemetry per asset per day, but there is no public benchmark dataset, no shared reference model, and no commonly agreed evaluation protocol for benchmarking sectoral, data-intensive models for operational decision support; notably for maintenance planning and prioritisation. We address this gap with GridForge, an open-source physics-informed digital twin and benchmark suite for fault diagnosis and prediction, integrating four building blocks: (i) a verified steady state and transient thermal model based on IEC 60287 and IEC 60853; (ii) a calibrated synthetic dataset of 64 cables (covering 320 cable-years) representing four canonical UK 11 kV thermal archetypes and four condition modes; (iii) a Physics-Informed Neural Network (PINN) surrogate that satisfies the algebraic balance defined by IEC 60287 during training and reaches an RMSE of 0.148 {\deg}C on an independent held out set against the reference analytical oracle; and (iv) a five-task benchmark suite that includes three reference baseline models (pure physics, gradient-boosted trees, and the PINN). All numerical parameters are sourced from public standards, the code is licensed under the MIT licence, and the trained PINN converges in only two minutes (as run on the computing platform in Section 10.1), enabling independent verification by DNO engineering and research practitioners. At present only the virtual-sensor task (T4) discriminates between the reference baselines; the other tasks tie or are undefined on this test split, so extending the baselines with task-specific logic is a priority future enhancement. For open validation, the dataset, test sets and models are made available at https://github.com/mlily2024/Gridforge

## Gradient Boosting vs. Random Forest vs. Neural Networks: A 2026 Performance Benchmark

DevFeed: [Gradient Boosting vs. Random Forest vs. Neural Networks: A 2026 Performance Benchmark](<https://devfeed.tech/articles/gradient-boosting-vs-random-forest-vs-neural-networks-a-2026-performance-benchmark-137767.md>)

Original publisher: [Read original article](<https://machinelearningmastery.com/gradient-boosting-vs-random-forest-vs-neural-networks-a-2026-performance-benchmark/>)

Author: Vinod Chugani

Published: 2026-11-17T11:00:00Z

Content type: comparison

Language: en

Sources: [MachineLearningMastery.com](<https://devfeed.tech/sources/machinelearningmastery-com.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Benchmark](<https://devfeed.tech/topics/benchmark.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [AI Foundation Models](<https://devfeed.tech/topics/ai-foundation-models.md>), [Deep learning](<https://devfeed.tech/topics/deep-learning.md>), [Hyperparameter optimization](<https://devfeed.tech/topics/hyperparameter-optimization.md>), [Transformer](<https://devfeed.tech/topics/transformer.md>), [catboost](<https://devfeed.tech/topics/catboost.md>), [mlp](<https://devfeed.tech/topics/mlp.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [audio](<https://devfeed.tech/tags/audio.md>), [backpropagation](<https://devfeed.tech/tags/backpropagation.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [catboost](<https://devfeed.tech/tags/catboost.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cnns](<https://devfeed.tech/tags/cnns.md>), [compare](<https://devfeed.tech/tags/compare.md>), [compute](<https://devfeed.tech/tags/compute.md>), [context-window](<https://devfeed.tech/tags/context-window.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [foundation-models](<https://devfeed.tech/tags/foundation-models.md>), [gradient-boosting](<https://devfeed.tech/tags/gradient-boosting.md>), [neural-networks](<https://devfeed.tech/tags/neural-networks.md>), [performance](<https://devfeed.tech/tags/performance.md>), [practical-machine-learning](<https://devfeed.tech/tags/practical-machine-learning.md>), [requirements](<https://devfeed.tech/tags/requirements.md>), [speed](<https://devfeed.tech/tags/speed.md>)

### AI overview

The excerpt describes two regimes for tabular data: tabular foundation models (TFMs) lead TabArena accuracy results on small-to-medium datasets, while gradient boosting remains stronger for large-scale data and regression. TabPFN-3.5 is reported to support up to 1 million rows and 20,000 features, but TFM limits vary and inference requires GPU memory to hold data in context. XGBoost, LightGBM, and CatBoost offer CPU-based training and low-latency inference, making them practical under compute or latency constraints. Random forest is a forgiving, quickly tuned baseline, especially with noisy data or limited tuning resources. Specialized networks have narrower advantages: TabR can outperform XGBoost on some classification tasks, while TabM and RealMLP reach parity with boosted trees on large datasets when carefully tuned and ensembled. For images, text, and audio, transformer architectures lead.

### Source excerpt

In this article, you will learn how gradient boosting, random forest, neural networks, and tabular foundation models compare on accuracy, speed, and data requirements in...

## A Fast Algorithm for the Familywise Loss

DevFeed: [A Fast Algorithm for the Familywise Loss](<https://devfeed.tech/articles/a-fast-algorithm-for-the-familywise-loss-138829.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19065>)

Author: Leelavathi Raja, Amirreza Eshraghi, Vinay Prabhakar, Mark Blacher, Han Bao, Yutong Wang

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [catboost](<https://devfeed.tech/topics/catboost.md>), [computational complexity](<https://devfeed.tech/topics/computational-complexity.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [classes](<https://devfeed.tech/tags/classes.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cross](<https://devfeed.tech/tags/cross.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [entropy](<https://devfeed.tech/tags/entropy.md>), [fast](<https://devfeed.tech/tags/fast.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [kernels](<https://devfeed.tech/tags/kernels.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [optimized](<https://devfeed.tech/tags/optimized.md>), [speed](<https://devfeed.tech/tags/speed.md>), [theoretical](<https://devfeed.tech/tags/theoretical.md>), [time](<https://devfeed.tech/tags/time.md>)

### AI overview

The paper addresses whether the familywise (FW) loss can match cross entropy (CE) computationally while retaining its stated theoretical advantage for multiclass classification. It proves that the FW gradient can be computed in O(K) time for K classes and reports optimized forward- and backward-pass kernels that run faster than CE on both CPU and GPU. The supplied excerpt does not include benchmark details or specify the conditions behind those speed comparisons.

### Source excerpt

arXiv:2611.19065v1 Announce Type: new Abstract: The familywise (FW) loss for multiclass classification, introduced by Duchi, Ruan, and Khosravi, achieves universal equivalence to the 0-1 loss, and thus has stronger theoretical properties than cross entropy (CE). Prior work, however, left open whether FW is as computationally efficient as CE. First, on the theoretical side, we prove that the FW gradient can be computed in $O(K)$ time, where $K$ is the number of classes. Second, on the practical side, we develop optimized kernels for computing the forward and backward passes of FW which outperform CE in terms of speed on both CPU and GPU.

## Reconstructing Boundary Layer Wind Profiles with Machine Learning: A Comparative Study on IGRA Radiosonde Data

DevFeed: [Reconstructing Boundary Layer Wind Profiles with Machine Learning: A Comparative Study on IGRA Radiosonde Data](<https://devfeed.tech/articles/reconstructing-boundary-layer-wind-profiles-with-machine-learning-a-comparative-study-on-igra-radiosonde-data-128697.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.13359>)

Author: Fedor Buzaev (HSE University), Dmitry Efremenko (HSE University), Daria Pugacheva (HSE University, Generalizable AI Systems), Fedor Ratnikov (HSE University), Egor Bugaev (HSE University), Vladimir Go

Published: 2026-11-11T05:00:00Z

Content type: comparison

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [catboost](<https://devfeed.tech/topics/catboost.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [analysis](<https://devfeed.tech/tags/analysis.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [architectures](<https://devfeed.tech/tags/architectures.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [catboost](<https://devfeed.tech/tags/catboost.md>), [cohort-analysis](<https://devfeed.tech/tags/cohort-analysis.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ne](<https://devfeed.tech/tags/cs-ne.md>), [dispersion](<https://devfeed.tech/tags/dispersion.md>), [energy](<https://devfeed.tech/tags/energy.md>), [error](<https://devfeed.tech/tags/error.md>), [features](<https://devfeed.tech/tags/features.md>), [fixed](<https://devfeed.tech/tags/fixed.md>), [forecasting](<https://devfeed.tech/tags/forecasting.md>), [law](<https://devfeed.tech/tags/law.md>), [layer](<https://devfeed.tech/tags/layer.md>), [leaving](<https://devfeed.tech/tags/leaving.md>), [less](<https://devfeed.tech/tags/less.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [models](<https://devfeed.tech/tags/models.md>), [physics-ao-ph](<https://devfeed.tech/tags/physics-ao-ph.md>), [study](<https://devfeed.tech/tags/study.md>)

### AI overview

On 158,097 IGRA soundings, CatBoost, TabM, and FT-Transformer achieved wind-profile RMSEs of 1.3-1.6 m/s, compared with about 3 m/s for power-law and Monin-Obukhov fits. The three machine-learning models differed by less than 0.05 m/s with predictors held fixed, suggesting predictor choice mattered more than architecture in this benchmark. Analysis of within-bin profile dispersion indicated an empirical error floor near 1.77 m/s for point-wise prediction from standard radiosonde features. A one-dimensional Fourier Neural Operator used to correct frozen tabular predictions lowered near-surface RMSE at U.S. stations from 1.5 to 1.2 m/s but did not change upper-layer error. The abstract concludes that further gains require additional predictors.

### Source excerpt

arXiv:2611.13359v1 Announce Type: new Abstract: Wind profiles in the atmospheric boundary layer (ABL) drive short-range forecasting and wind-energy siting, yet sparse radiosonde coverage forces practitioners to interpolate them from a handful of physical predictors. We benchmark modern tabular learners against analytical parameterizations on 158,097 IGRA soundings. CatBoost, TabM, and FT-Transformer reduce RMSE to 1.3-1.6 m/s, against approximately 3 m/s for power-law and Monin-Obukhov fits; once the predictor set is fixed the three architectures differ by less than 0.05 m/s, so accuracy is driven by predictors rather than architecture. A cohort analysis of within-bin profile dispersion places an empirical floor of approximately 1.77 m/s on any point-wise predictor that uses standard radiosonde features, consistent with prior eddy-viscosity inversions. A one-dimensional Fourier Neural Operator applied as a post-hoc corrector to frozen tabular predictions reduces near-surface RMSE on U.S. stations from 1.5 to 1.2 m/s, while leaving upper-layer error unchanged. Further gains should come from new predictors rather than larger models.

## Scalable Bayesian Optimization for Efficient Exploration of Moire Patterns

DevFeed: [Scalable Bayesian Optimization for Efficient Exploration of Moire Patterns](<https://devfeed.tech/articles/scalable-bayesian-optimization-for-efficient-exploration-of-moire-patterns-125496.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.12010>)

Author: Panagiotis Krokidas, Christos Lazaridis, Vassilis Sioros, Efi-Maria Papia, Vassilios Constantoudis, George Giannakopoulos

Published: 2026-11-10T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Operations research and optimization](<https://devfeed.tech/topics/operations-research-and-optimization.md>), [catboost](<https://devfeed.tech/topics/catboost.md>)

Tags: [bayesian](<https://devfeed.tech/tags/bayesian.md>), [catboost](<https://devfeed.tech/tags/catboost.md>), [compare](<https://devfeed.tech/tags/compare.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [design](<https://devfeed.tech/tags/design.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [evaluations](<https://devfeed.tech/tags/evaluations.md>), [exploration](<https://devfeed.tech/tags/exploration.md>), [fairness](<https://devfeed.tech/tags/fairness.md>), [gaussian](<https://devfeed.tech/tags/gaussian.md>), [global](<https://devfeed.tech/tags/global.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [gpu-memory](<https://devfeed.tech/tags/gpu-memory.md>), [learning](<https://devfeed.tech/tags/learning.md>), [lightweight](<https://devfeed.tech/tags/lightweight.md>), [machine](<https://devfeed.tech/tags/machine.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [materials](<https://devfeed.tech/tags/materials.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [patterns](<https://devfeed.tech/tags/patterns.md>), [physics-app-ph](<https://devfeed.tech/tags/physics-app-ph.md>)

### AI overview

FruBO with a lightweight CatBoost surrogate and committee-based uncertainty estimation found the best pore-uniformity configuration among 600,000 multilayer Moiré patterns after 1,470 acquired samples, taking about 1.15 hours. Random exploration required 533,012 evaluations and about 31 hours to find the same optimum, while Gaussian Process Bayesian Optimization became impractical after 290 samples because of GPU memory exhaustion. The benchmark used six rotated hexagonal lattices and measured pore-size uniformity with the Jain fairness index. These results suggest CatBoost-based surrogates can make Bayesian optimization more feasible for this large design space.

### Source excerpt

arXiv:2611.12010v1 Announce Type: new Abstract: Large materials design spaces often contain rare high-performing configurations that are difficult to identify through trial-and-error exploration or purely representative machine-learning (ML) sampling. In this work, we present FruBO, a flexible Bayesian Optimization framework for accelerating the discovery of optimal configurations in computationally demanding materials design problems. As a case study, we investigate multilayer Moir\'e patterns generated by the superposition of six rotated hexagonal lattices, where the objective is to identify the rotational conformation that maximizes pore-size uniformity, quantified using the Jain fairness index. The resulting benchmark contains 600,000 unique configurations and exhibits a highly skewed objective distribution, making the global optimum difficult to locate by random sampling. We compare random exploration, Gaussian Process-based Bayesian Optimization, and Bayesian Optimization using a lightweight CatBoost surrogate with committee-based uncertainty estimation. Random sampling identifies the global optimum only after 533,012 evaluations, requiring approximately 31 hours, while Gaussian Process-based BO becomes impractical due to GPU memory exhaustion after 290 acquired samples. In contrast, FruBO with CatBoost identifies the global optimum after 1,470 acquired samples and approximately 1.15 hours. These results demonstrate that scalable surrogate models can substantially improve the feasibility and efficiency of BO in large nanomaterials design spaces.

## Generalized Adaptive Boosting and the Geometry of Mistakes

DevFeed: [Generalized Adaptive Boosting and the Geometry of Mistakes](<https://devfeed.tech/articles/generalized-adaptive-boosting-and-the-geometry-of-mistakes-125583.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.12696>)

Author: Marco Bressan, Nataly Brukhim, Nicol\\\`o Cesa-Bianchi, Emmanuel Esposito, Yishay Mansour, Shay Moran, Maximilian Thiessen

Published: 2026-11-10T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [catboost](<https://devfeed.tech/topics/catboost.md>)

Tags: [2025](<https://devfeed.tech/tags/2025.md>), [abstract](<https://devfeed.tech/tags/abstract.md>), [account](<https://devfeed.tech/tags/account.md>), [adaptive](<https://devfeed.tech/tags/adaptive.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [all](<https://devfeed.tech/tags/all.md>), [allows](<https://devfeed.tech/tags/allows.md>), [also](<https://devfeed.tech/tags/also.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [boosting](<https://devfeed.tech/tags/boosting.md>), [can](<https://devfeed.tech/tags/can.md>), [cases](<https://devfeed.tech/tags/cases.md>), [characterization](<https://devfeed.tech/tags/characterization.md>), [complete](<https://devfeed.tech/tags/complete.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [diagnosis](<https://devfeed.tech/tags/diagnosis.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [example](<https://devfeed.tech/tags/example.md>), [mistakes](<https://devfeed.tech/tags/mistakes.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>)

### AI overview

The paper introduces Z-learners, which guarantee joint false-positive and false-negative rates within a set Z, allowing weak-learning guarantees to reflect how mistake types and data distribution affect performance. It gives a geometric characterization of which sets Z can be boosted, encompassing classical boosting and a recent multi-objective extension. The authors also present an adaptive algorithm that boosts Z-learners without prior knowledge of Z, extending AdaBoost's handling of unknown margin parameters, and show that Z-learners can amplify confidence. The analysis is more involved than classical boosting and relies on the centerpoint theorem. This overview is based on the supplied abstract.

### Source excerpt

arXiv:2611.12696v1 Announce Type: new Abstract: Not all mistakes are alike. For example, in medical diagnosis, false positive and false negative mistakes have a different impact and should not be aggregated. By revisiting the classical theory of boosting through this lens, we introduce the notion of $Z$-learner, which returns weak hypotheses achieving joint false-positive/false-negative guarantees $(\mathrm{FP},\mathrm{FN})$ from some set $Z \subseteq [0,1]^2$. This formulation generalizes the standard notion of $\gamma$-weak learning--recovered by taking $Z=\{(z_1,z_2): z_1+z_2<1/2-\gamma\}$--and allows for more flexible notions of weak learnability that account for the impact of the distribution on the achievable learning guarantees. This raises a natural question: which sets $Z$ are boostable? We answer this by giving a complete geometric characterization that unifies, as special cases, both the classical boosting dichotomy and the recent multi-objective extension of Bressan et al. (COLT 2025). We also present an adaptive algorithm for boosting $Z$-learners without relying on the prior knowledge of $Z$, generalizing AdaBoost's ability to handle unknown margin parameters. Finally, we demonstrate that $Z$-learners can achieve confidence amplification. However, the analysis is significantly more involved than in classical boosting and relies on the centerpoint theorem from discrete and computational geometry.

## Topology as a Design Variable: Joint Architecture-Sizing Optimization for SRAM

DevFeed: [Topology as a Design Variable: Joint Architecture-Sizing Optimization for SRAM](<https://devfeed.tech/articles/topology-as-a-design-variable-joint-architecture-sizing-optimization-for-sram-123382.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.09248>)

Author: Can Wang, Junhao Ma, Yikai Wang, Shan Shen, Wei W. Xing

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Operations research and optimization](<https://devfeed.tech/topics/operations-research-and-optimization.md>), [Combinatorial optimization](<https://devfeed.tech/topics/combinatorial-optimization.md>), [catboost](<https://devfeed.tech/topics/catboost.md>), [Open Source](<https://devfeed.tech/topics/open-source.md>)

Tags: [architecture](<https://devfeed.tech/tags/architecture.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [circuits](<https://devfeed.tech/tags/circuits.md>), [compiler](<https://devfeed.tech/tags/compiler.md>), [continuous](<https://devfeed.tech/tags/continuous.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [design](<https://devfeed.tech/tags/design.md>), [few-shot](<https://devfeed.tech/tags/few-shot.md>), [global](<https://devfeed.tech/tags/global.md>), [local](<https://devfeed.tech/tags/local.md>), [macros](<https://devfeed.tech/tags/macros.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [optimizing](<https://devfeed.tech/tags/optimizing.md>), [performance](<https://devfeed.tech/tags/performance.md>), [power](<https://devfeed.tech/tags/power.md>), [ppa](<https://devfeed.tech/tags/ppa.md>), [replica](<https://devfeed.tech/tags/replica.md>), [research](<https://devfeed.tech/tags/research.md>), [sram](<https://devfeed.tech/tags/sram.md>), [topology](<https://devfeed.tech/tags/topology.md>)

### AI overview

OpenYieldV2 jointly optimizes SRAM bitcell topology and transistor sizing, addressing a gap between OpenRAM's layout generation without optimization and OpenYield's sizing optimization for a fixed 6T cell. Its parameterized compiler generates complete 6T and 10T macros for comparable evaluation; a few-shot TabPFN surrogate models topology-dependent power, performance, and area responses from limited full-macro SPICE samples; and a coarse-to-refined multi-objective optimizer explores topology choices while refining transistor sizes. Experiments report that the preferred topology boundary shifts with operating constraints, and joint optimization improves hypervolume by up to 17.3% over the strongest single-topology baseline. The supplied text is an abstract, so it does not provide detailed experimental conditions or results.

### Source excerpt

arXiv:2611.09248v1 Announce Type: new Abstract: SRAM's power-performance-area-stability trade-off depends on bitcell topology, a discrete architectural choice that interacts non-separably with continuous transistor sizing. However, no open-source SRAM tool treats topology as an optimizable variable. OpenRAM generates SRAM layouts but performs no optimization, while OpenYield supports high-fidelity evaluation but restricts optimization to scalar sizing of a fixed 6T cell. OpenYieldV2 closes this gap through three integrated contributions. First, a fully parameterized compiler generates complete 6T and 10T SRAM macros with peripheral circuits and replica-bitline self-timing, enabling topology comparison under identical simulation conditions. Second, a few-shot TabPFN surrogate learns topology-dependent PPA responses with limited full-macro SPICE samples. Third, a coarse-to-refined multi-objective optimizer combines global exploration of discrete architectural choices with local refinement of continuous transistor sizes. Experiments across varying design conditions reveal a topology decision boundary that shifts with operating constraints rather than following a fixed design rule. Jointly optimizing across this boundary improves hypervolume by up to 17.3\% over the strongest single-topology baseline. OpenYieldV2 is open-source to support reproducible SRAM design research.

## Gradient Boosting for Efficient Rank Discovery in Tensor Network Kernel Machines

DevFeed: [Gradient Boosting for Efficient Rank Discovery in Tensor Network Kernel Machines](<https://devfeed.tech/articles/gradient-boosting-for-efficient-rank-discovery-in-tensor-network-kernel-machines-120286.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.08093>)

Author: Albert Saiapin, Kim Batselier

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [catboost](<https://devfeed.tech/topics/catboost.md>), [scikit-learn](<https://devfeed.tech/topics/scikit-learn.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [approach](<https://devfeed.tech/tags/approach.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [based](<https://devfeed.tech/tags/based.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [boosting](<https://devfeed.tech/tags/boosting.md>), [builds](<https://devfeed.tech/tags/builds.md>), [candidate](<https://devfeed.tech/tags/candidate.md>), [canonical](<https://devfeed.tech/tags/canonical.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [complex](<https://devfeed.tech/tags/complex.md>), [costs](<https://devfeed.tech/tags/costs.md>), [cross](<https://devfeed.tech/tags/cross.md>), [cross-validation](<https://devfeed.tech/tags/cross-validation.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [kernel](<https://devfeed.tech/tags/kernel.md>), [machines](<https://devfeed.tech/tags/machines.md>), [network](<https://devfeed.tech/tags/network.md>), [rank](<https://devfeed.tech/tags/rank.md>)

### AI overview

GB-CPKM avoids explicit tensor-rank search by sequentially adding fixed-rank canonical polyadic tensor network kernel learners, with ensemble size chosen by validation performance. On real-world regression benchmarks, it achieved comparable predictive performance to a cross-validation-based rank-selection method while training up to 6x faster and using up to 2x less peak memory. These reported results apply to the benchmarks described in the abstract.

### Source excerpt

arXiv:2611.08093v1 Announce Type: cross Abstract: Tensor network kernel machines (TNKMs) combine nonlinear features with low-rank tensor representations to model complex relationships efficiently. However, selecting an appropriate tensor rank remains challenging, as conventional rank-search methods require training multiple candidate models, increasing computational costs. In this work, we focus on a canonical polyadic (CP) tensor network parameterization and propose GB-CPKM, a gradient boosting framework that builds a higher-capacity predictor by sequentially adding fixed-rank learners. This approach provides a new perspective on rank selection by incrementally increasing model capacity, with ensemble size selected based on validation performance rather than through explicit rank search. Experiments on real-world regression benchmarks show that GB-CPKM trains up to 6x faster and uses up to 2x less peak memory than a conventional rank-selection method based on cross-validation, while achieving comparable predictive performance. These results highlight the potential of gradient boosting to improve the computational efficiency and scalability of tensor-network learning.

## How Amazon Marketplace evaluates content quality with Amazon SageMaker AI

DevFeed: [How Amazon Marketplace evaluates content quality with Amazon SageMaker AI](<https://devfeed.tech/articles/how-amazon-marketplace-evaluates-content-quality-with-amazon-sagemaker-ai-119744.md>)

Original publisher: [Read original article](<https://aws.amazon.com/blogs/machine-learning/how-amazon-marketplace-evaluates-content-quality-with-amazon-sagemaker-ai/>)

Author: James Kim

Published: 2026-11-05T18:06:43Z

Content type: article

Language: en

Sources: [Artificial Intelligence](<https://devfeed.tech/sources/artificial-intelligence.md>)

Topics: [Amazon SageMaker](<https://devfeed.tech/topics/amazon-sagemaker.md>), [catboost](<https://devfeed.tech/topics/catboost.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [model architecture](<https://devfeed.tech/topics/model-architecture.md>), [Amazon SageMaker JumpStart](<https://devfeed.tech/topics/amazon-sagemaker-jumpstart.md>), [lora](<https://devfeed.tech/topics/lora.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [amazon](<https://devfeed.tech/tags/amazon.md>), [amazon-marketplace](<https://devfeed.tech/tags/amazon-marketplace.md>), [amazon-sagemaker](<https://devfeed.tech/tags/amazon-sagemaker.md>), [amazon-sagemaker-ai](<https://devfeed.tech/tags/amazon-sagemaker-ai.md>), [best-practices](<https://devfeed.tech/tags/best-practices.md>), [catboost](<https://devfeed.tech/tags/catboost.md>), [content](<https://devfeed.tech/tags/content.md>), [expert-400](<https://devfeed.tech/tags/expert-400.md>), [language](<https://devfeed.tech/tags/language.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [marketplace](<https://devfeed.tech/tags/marketplace.md>), [model](<https://devfeed.tech/tags/model.md>), [ocr](<https://devfeed.tech/tags/ocr.md>), [partners](<https://devfeed.tech/tags/partners.md>), [product](<https://devfeed.tech/tags/product.md>), [quality](<https://devfeed.tech/tags/quality.md>), [readability](<https://devfeed.tech/tags/readability.md>), [recommendations](<https://devfeed.tech/tags/recommendations.md>), [rendering](<https://devfeed.tech/tags/rendering.md>), [scale](<https://devfeed.tech/tags/scale.md>), [shap](<https://devfeed.tech/tags/shap.md>)

### AI overview

Amazon Marketplace evaluates A+ Content holistically by rendering a product detail page as one snapshot and scoring it with a rubric prompted to a fixed vision-language model (VLM). Rather than fine-tune that model, the system feeds its four rubric scores and four OCR readability signals into CatBoost trained on expert labels; OCR adds layout-based measures such as text and character density and spacing, which complement the VLM's broader semantic and visual assessment. On a 20% holdout of 5,478 pages across 10 categories, the fused model reached 0.94 accuracy and 0.95 F1, versus 0.89 and 0.90 for VLM-only CatBoost and 0.80 and 0.84 for the averaged-score baseline. SHAP ranks factors behind each decision, and an LLM uses negative contributions, category context, best practices, and VLM rationales to prioritize seller recommendations. Revisions were associated with a statistically significant average 1.4-point quality-score increase on a 10-point scale. The described SageMaker deployment supports real-time, asynchronous, or batch VLM scoring; the authors note expert labels may not represent every category, language, or marketplace.

### Source excerpt

Learn how Amazon Marketplace evaluates A+ Content quality at scale on Amazon SageMaker AI by rendering each product page as a single snapshot, scoring it with a vision-language model fused with OCR readability signals in a CatBoost classifier, and using SHAP to turn each decision into prioritized, actionable recommendations for selling partners.

## Deep Meets Classical: A CPU-Efficient Decision-Fusion Model for Automatic Modulation Classification

DevFeed: [Deep Meets Classical: A CPU-Efficient Decision-Fusion Model for Automatic Modulation Classification](<https://devfeed.tech/articles/deep-meets-classical-a-cpu-efficient-decision-fusion-model-for-automatic-modulation-classification-118405.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.05937>)

Author: Zhiyi Yin, Zhizhao Wen, Ruoxin Zhang, Chao Wang, Puyang Xu, Yifan Jiang, Yusi Gong

Published: 2026-11-05T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [model architecture](<https://devfeed.tech/topics/model-architecture.md>), [catboost](<https://devfeed.tech/topics/catboost.md>)

Tags: [classification](<https://devfeed.tech/tags/classification.md>), [cnn](<https://devfeed.tech/tags/cnn.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cpu-efficient](<https://devfeed.tech/tags/cpu-efficient.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [embedded](<https://devfeed.tech/tags/embedded.md>), [fusion](<https://devfeed.tech/tags/fusion.md>), [logistic-regression](<https://devfeed.tech/tags/logistic-regression.md>), [radio](<https://devfeed.tech/tags/radio.md>), [signal](<https://devfeed.tech/tags/signal.md>), [snr](<https://devfeed.tech/tags/snr.md>)

### AI overview

FuseAMC combines a residual CNN on raw I/Q samples with a higher-order-cumulant classifier because their strengths differ by SNR: the CNN is more robust at low SNR, while the classical classifier is more accurate at high SNR. A logistic-regression combiner learns from both models' class posteriors using a validation split separate from the test set. On RadioML 2016.10a, across five seeds and a 44,000-frame held-out test set, it achieved 55.8% accuracy, 5.1 percentage points above the best cumulant baseline and 3.2 points above the CNN alone; reported paired-bootstrap confidence bounds for both gains were positive. The roughly 198,000-parameter CNN and full pipeline require about 3 MFLOPs per frame and run on one CPU core without hardware acceleration.

### Source excerpt

arXiv:2611.05937v1 Announce Type: cross Abstract: Automatic modulation classification (AMC) underpins spectrum sensing and non-cooperative signal identification in cognitive radio, yet the two standard CPU-feasible approaches to it, a compact convolutional neural network (CNN) on raw in-phase and quadrature (I/Q) samples and a classical higher-order-cumulant feature classifier, are rarely compared against each other conditioned on signal-to-noise ratio (SNR). We show that they are, in fact, complementary: the CNN is markedly more robust at low SNR, where hand-crafted statistics collapse toward chance, while the cumulant classifier is more accurate at high SNR, where higher-order statistics cleanly separate dense constellations. We exploit this complementarity with FuseAMC, a lightweight decision-fusion model that trains a logistic-regression meta-combiner over the concatenated class-posteriors of an approximately 198k-parameter residual CNN and a gradient-boosted-tree cumulant classifier, on a validation split disjoint from test. On RadioML 2016.10a (5 seeds, 44k-frame held-out test set), FuseAMC reaches 55.8% overall accuracy, improving on the best cumulant-feature baseline by 5.1 points and on the CNN alone by 3.2 points (paired-bootstrap 95% confidence-interval lower bounds of 5.0 and 3.1 points, both positive), while matching the best classical baseline in its own strong high-SNR regime and the CNN in its strong low-SNR regime. With about 0.2 million parameters, a model footprint of a few megabytes, and roughly 3 MFLOPs per frame, the whole pipeline trains and runs on one CPU core, targeting general-purpose and embedded processors without hardware acceleration.

## When Is Tracking Affordable? A Movement-Budget Feasibility Threshold for Dynamic Regret

DevFeed: [When Is Tracking Affordable? A Movement-Budget Feasibility Threshold for Dynamic Regret](<https://devfeed.tech/articles/when-is-tracking-affordable-a-movement-budget-feasibility-threshold-for-dynamic-regret-114740.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.02812>)

Author: Ridham Patel

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [catboost](<https://devfeed.tech/topics/catboost.md>)

Tags: [adversary](<https://devfeed.tech/tags/adversary.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [budget](<https://devfeed.tech/tags/budget.md>), [construction](<https://devfeed.tech/tags/construction.md>), [convex](<https://devfeed.tech/tags/convex.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-sy](<https://devfeed.tech/tags/cs-sy.md>), [diameter](<https://devfeed.tech/tags/diameter.md>), [domain](<https://devfeed.tech/tags/domain.md>), [eess-sy](<https://devfeed.tech/tags/eess-sy.md>), [environment](<https://devfeed.tech/tags/environment.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [gradient-descent](<https://devfeed.tech/tags/gradient-descent.md>), [learner](<https://devfeed.tech/tags/learner.md>), [linear](<https://devfeed.tech/tags/linear.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [math-oc](<https://devfeed.tech/tags/math-oc.md>), [minimax](<https://devfeed.tech/tags/minimax.md>), [moving](<https://devfeed.tech/tags/moving.md>), [online](<https://devfeed.tech/tags/online.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>), [tracking](<https://devfeed.tech/tags/tracking.md>)

### AI overview

A hard cap on an online learner's total movement can make tracking a moving comparator incur linear dynamic regret, even when the environment changes slowly. For convex losses on a domain of diameter D, a deterministic adversary forces regret Ω(GDT) when the movement budget S is at most a sufficiently small constant multiple of comparator path length P_T. By contrast, projected online gradient descent with a budget-aware step size achieves sublinear regret when S grows faster than D + P_T, in the learnable regime. The results distinguish the learner's travel forced by the comparator from the comparator's path length alone. Experiments support the predicted threshold and 1/S decay; the tight minimax rate above the threshold remains open.

### Source excerpt

arXiv:2611.02812v1 Announce Type: new Abstract: Dynamic regret measures an online learner against a moving comparator, and standard analyses assume the learner may move freely at every round. We study what happens when it may not: the learner is given a hard budget $S$ on its total movement $\sum_t \|x_t - x_{t-1}\|$. Our main result is that a movement budget creates a feasibility barrier. On a domain of diameter $D$ with $G$-Lipschitz convex losses, for every comparator path length $P_T \in [D, DT]$ we exhibit a deterministic adversary that forces $R_d = \Omega(GDT)$ on every budget-feasible algorithm whenever $S \le c P_T$ for a universal constant $c>0$. This barrier is caused by the budget rather than by a volatile environment: for fixed $D$, it applies when $P_T = o(T)$, where the unconstrained minimax rate $\Theta(G\sqrt{D(D+P_T)T})$ is sublinear, so the same instances are learnable with an unbounded budget and unlearnable without one. Conversely, projected online gradient descent with a budget-aware step size is budget-feasible and attains $o(GDT)$ once $S = \omega(D + P_T)$. Thus the critical movement scale is $D+P_T$: linear regret is unavoidable below a sufficiently small constant multiple of $P_T$, while sublinear regret is achievable when $S/(D+P_T)\to\infty$ in the learnable regime. The construction rests on a distinction we make explicit: what makes a comparator expensive to track under a movement budget is the distance it forces the learner to travel, not its path length. Above the threshold we give a budget-dependent upper bound and identify the tight minimax rate in this region as an open problem rather than claiming a product law. Experiments support the predicted transition location and the $1/S$ decay.

## Omniprediction for All Mixable Proper Losses and Beyond

DevFeed: [Omniprediction for All Mixable Proper Losses and Beyond](<https://devfeed.tech/articles/omniprediction-for-all-mixable-proper-losses-and-beyond-111533.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.41023>)

Author: Junyan Liu, Haipeng Luo, Lillian J. Ratliff

Published: 2026-11-02T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [catboost](<https://devfeed.tech/topics/catboost.md>), [Algorithms](<https://devfeed.tech/topics/algorithms.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [cs-ds](<https://devfeed.tech/tags/cs-ds.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [glm](<https://devfeed.tech/tags/glm.md>), [linear](<https://devfeed.tech/tags/linear.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [online-learning](<https://devfeed.tech/tags/online-learning.md>), [predictions](<https://devfeed.tech/tags/predictions.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>)

### AI overview

The paper establishes that no algorithm can achieve sublinear regret simultaneously for all bounded proper losses in general, even with only two linear predictors. For the narrower class of all mixable proper losses, it gives an optimal up to logarithmic factors regret rate of Õ(T^{1/3}(log|F|)^{2/3}), compared with constant-in-T regret for a finite class of mixable proper losses. Its extended algorithm combines this guarantee with existing guarantees for finite loss classes: with suitable choices, it attains constant regret for squared and log loss, the T^{1/3} rate for all mixable proper losses, and Õ(√(T log|F|)) regret for 0-1, Lipschitz, GLM, and other structured loss families.

### Source excerpt

arXiv:2610.41023v1 Announce Type: new Abstract: Omniprediction concerns the task of making predictions that are simultaneously accurate with respect to different losses. Earlier results by [CV09] show that in an online learning setting, $O(\log(|\mathcal{F}||\mathcal{L}|))$ regret is achievable simultaneously against every predictor in a finite predictor class $\mathcal{F}$ and for every loss in a finite class $\mathcal{L}$ of mixable proper losses, while recent advances [OKK25, FP26] show that $\widetilde{O}(\sqrt{T\log(|\mathcal{F}||\mathcal{L}|)})$ regret is achievable simultaneously for every loss in a finite class $\mathcal{L}$ of bounded losses. In this work, we ask whether sublinear regret remains possible when $\mathcal{L}$ is an infinite class of proper losses. Our first result gives a negative answer in general: even when $\mathcal{F}$ contains only two linear predictors, it is impossible to achieve sublinear regret simultaneously for all bounded proper losses. We then show that $\widetilde{O}(T^{1/3}(\log|\mathcal{F}|)^{2/3})$ regret is achievable simultaneously for \textit{all} mixable proper losses and that this bound is optimal up to logarithmic factors. This optimal rate stands in stark contrast to the constant (in $T$) regret achievable for a finite class of mixable proper losses. Finally, we extend our algorithm to obtain a best-of-all-worlds result: in addition to the new regret guarantee for all mixable proper losses, the resulting algorithm simultaneously achieves the existing guarantees for finite loss classes. With appropriate choices of these finite classes, our algorithm simultaneously achieves constant regret for squared loss and log loss, $\widetilde{O}(T^{1/3}(\log|\mathcal{F}|)^{2/3})$ regret for all mixable proper losses, and $\widetilde{O}(\sqrt{T\log|\mathcal{F}|})$ regret for $0$-$1$ loss, all Lipschitz losses, all GLM losses, and other structured families.

## B2P-Corr: Batch-to-Population Gradient Estimators for Non-Decomposable Correlation Losses

DevFeed: [B2P-Corr: Batch-to-Population Gradient Estimators for Non-Decomposable Correlation Losses](<https://devfeed.tech/articles/b2p-corr-batch-to-population-gradient-estimators-for-non-decomposable-correlation-losses-108657.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.38171>)

Author: Jingquan Yan, Yuwei Miao, Peiran Yu, Junzhou Huang

Published: 2026-10-30T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [catboost](<https://devfeed.tech/topics/catboost.md>)

Tags: [batch](<https://devfeed.tech/tags/batch.md>), [bias](<https://devfeed.tech/tags/bias.md>), [correlation](<https://devfeed.tech/tags/correlation.md>), [covariance](<https://devfeed.tech/tags/covariance.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [expectation](<https://devfeed.tech/tags/expectation.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [framework](<https://devfeed.tech/tags/framework.md>), [gap](<https://devfeed.tech/tags/gap.md>), [global](<https://devfeed.tech/tags/global.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [lightweight](<https://devfeed.tech/tags/lightweight.md>), [local](<https://devfeed.tech/tags/local.md>), [metrics](<https://devfeed.tech/tags/metrics.md>), [mini](<https://devfeed.tech/tags/mini.md>), [moving](<https://devfeed.tech/tags/moving.md>), [nature](<https://devfeed.tech/tags/nature.md>), [novel](<https://devfeed.tech/tags/novel.md>), [objectives](<https://devfeed.tech/tags/objectives.md>), [operators](<https://devfeed.tech/tags/operators.md>)

### AI overview

B2P-Corr targets a bias and variance problem in mini-batch optimization of Pearson and concordance correlation objectives, which depend on population-level moments. It replaces local-statistic objectives with decomposable pseudo-losses driven by global moment sketches. B2P-EMA uses exponential moving averages and stop-gradient operators for asymptotic gradient tracking at O(B) cost; B2P-U uses sampled cross-batch U-statistics with reservoir stabilization, also with an O(B) implementation. The abstract reports an O(B⁻¹) bias for naive mini-batch optimization, a tracking result for B2P-EMA under a fast-sketch/slow-parameter regime, and an idealized control-variate analysis for B2P-U. Experiments on two synthetic and three real-world datasets reportedly show more stable optimization and higher correlation than the naive approach, with negligible overhead.

### Source excerpt

arXiv:2610.38171v1 Announce Type: new Abstract: Correlation metrics such as the Pearson Correlation Coefficient (PCC) and Concordance Correlation Coefficient (CCC) are standard regression metrics and learning objectives, but their non-decomposable nature poses a challenge for stochastic optimization. Since these metrics depend on dataset-level moments (mean, variance, covariance), naive mini-batch training with local statistics yields biased, high-variance gradients whose expectation does not align with the population gradient. To bridge this ``batch-to-population'' (B2P) gap, we propose \textbf{B2P-Corr}, a novel framework that transforms correlation objectives into decomposable pseudo-losses driven by lightweight global ``moment sketches''. We introduce two efficient variants: \textbf{B2P-EMA}, which utilizes exponential moving averages and stop-gradient operators to ensure asymptotic gradient tracking with $O(B)$ cost ($B$ is batch size); and \textbf{B2P-U}, which uses sampled cross-batch U-statistic estimators and a reservoir-based stabilization with an $O(B)$ implementation. Theoretically, we quantify the $O(B^{-1})$ bias of naive mini-batch optimization, prove a moving-target tracking result for B2P-EMA under a fast-sketch / slow-parameter regime, and provide an idealized control-variate analysis for B2P-U. Experiments on two synthetic and three real-world datasets demonstrate B2P-Corr achieves stable optimization and higher correlation than naive mini-batch optimization with negligible overhead.

## Transient Amplification in Cascaded Gradient Methods: Exact Minimax Laws and Tracking without Resonance

DevFeed: [Transient Amplification in Cascaded Gradient Methods: Exact Minimax Laws and Tracking without Resonance](<https://devfeed.tech/articles/transient-amplification-in-cascaded-gradient-methods-exact-minimax-laws-and-tracking-without-resonance-99420.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2606.04031>)

Author: Ahanaf Hasan Ariq, Davide Wiest

Published: 2026-10-26T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [catboost](<https://devfeed.tech/topics/catboost.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [3](<https://devfeed.tech/tags/3.md>), [abstract](<https://devfeed.tech/tags/abstract.md>), [accelerated](<https://devfeed.tech/tags/accelerated.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [block](<https://devfeed.tech/tags/block.md>), [blocks](<https://devfeed.tech/tags/blocks.md>), [c](<https://devfeed.tech/tags/c.md>), [can](<https://devfeed.tech/tags/can.md>), [code](<https://devfeed.tech/tags/code.md>), [coupling](<https://devfeed.tech/tags/coupling.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cubic](<https://devfeed.tech/tags/cubic.md>), [d](<https://devfeed.tech/tags/d.md>), [dimension](<https://devfeed.tech/tags/dimension.md>), [effective](<https://devfeed.tech/tags/effective.md>), [eigenvalues](<https://devfeed.tech/tags/eigenvalues.md>), [equals](<https://devfeed.tech/tags/equals.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [math-oc](<https://devfeed.tech/tags/math-oc.md>), [minimax](<https://devfeed.tech/tags/minimax.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>), [tracking](<https://devfeed.tech/tags/tracking.md>)

### AI overview

Cascaded leader-follower gradient methods can have large transient amplification even when asymptotically stable, so eigenvalues alone do not predict finite-time gain. For the stated block-matrix assumptions, the excerpt gives an exact dimension-free worst-case amplification at each horizon, attained by an aligned two-dimensional quadratic; the largest transient scales on the order of 1 + c/(1−γ). With martingale noise, the aligned system's stationary follower variance has cubic inverse-gap scaling. For quadratic followers, coupling is related to sensitivity R; with an isotropic leader and positive semidefinite follower map, amplification is bounded by 1 + ||R|| independently of spectral gaps. A look-ahead Newton update removes follower memory and is minimax-optimal among linear updates that track the leader exactly. Experiments report that accelerated or inexact follower solvers can exceed the sensitivity bound when their effective maps have negative eigenvalues.

### Source excerpt

arXiv:2606.04031v3 Announce Type: replace Abstract: Leader-follower gradient iterations, including bilevel and two-time-scale methods, can exhibit large finite-time excursions despite asymptotic stability. For cascaded iterations with block matrix J = [[A,0],[C,D]], symmetric diagonal blocks of norm at most gamma < 1, and coupling norm at most c, we prove an exact, dimension-free minimax formula: at every horizon t, the worst-case amplification equals the spectral norm of [[gamma^t,0],[c t gamma^(t-1),gamma^t]], attained by an aligned two-dimensional quadratic. Consequently, the worst transient is on the order of 1 + c/(1-gamma), and eigenvalues alone cannot predict finite-time gain. Under martingale noise, the aligned system has stationary follower variance sigma^2 c^2 (1+gamma^2)/(1-gamma^2)^3, establishing cubic inverse-gap scaling. For quadratic followers, the coupling is linked to follower sensitivity R by C = (I-D)R. When the leader is isotropic and the follower map is positive semidefinite, amplification is at most 1 plus the operator norm of R, independently of the spectral gaps. A look-ahead Newton update eliminates follower memory and is minimax-optimal among linear updates that track the leader exactly. Experiments validate the bounds and show that accelerated or inexact follower solvers can exceed the sensitivity bound when their effective maps have negative eigenvalues. Code and experiment scripts: https://github.com/Axym-Labs/transient-amplification

## Sampling and Curvature in First-Exit Comparisons for Momentum SGD

DevFeed: [Sampling and Curvature in First-Exit Comparisons for Momentum SGD](<https://devfeed.tech/articles/sampling-and-curvature-in-first-exit-comparisons-for-momentum-sgd-93438.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.25191>)

Author: Haiyi Li

Published: 2026-10-21T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [catboost](<https://devfeed.tech/topics/catboost.md>)

Tags: [comparisons](<https://devfeed.tech/tags/comparisons.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [exit](<https://devfeed.tech/tags/exit.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [momentum](<https://devfeed.tech/tags/momentum.md>), [sampling](<https://devfeed.tech/tags/sampling.md>), [switching](<https://devfeed.tech/tags/switching.md>)

### AI overview

Pointwise gradient laws alone do not determine momentum SGD's first-exit behavior: finite pools with identical pointwise laws and matched iid dynamics can produce different exits under reshuffling. For one-dimensional persistent sampling, the paper proves an O(h) mean-exit comparison with a finite-rate switching process under stated velocity and step-size conditions; iid sampling instead averages to an ODE. Beyond those conditions, using realized minibatches and sampled curvature gives finite-window exit brackets. Experiments report certified brackets meeting targets in 29 of 32 ridge-logistic runs and finite-mean certificates for three ResNet-18 checkpoints, while neural-network upper bounds remain loose and original-radius ViT guarantees unresolved. Persistence comparisons replicate across two networks, but refresh-versus-random exit ordering changes with update length. The interventions do not causally decompose raw momentum SGD, and neural step-size scaling does not validate the switching limit.

### Source excerpt

arXiv:2610.25191v1 Announce Type: new Abstract: First-exit prediction for momentum SGD depends on sampling information that pointwise gradient laws omit. Strongly convex finite pools with identical pointwise laws and matched iid momentum dynamics can have different reshuffled exits; even analytic pools with uniformly close pointwise laws can have separated exit probabilities. For persistent sampling in one dimension, we establish an explicit $O(h)$ mean-exit comparison with a finite-rate switching process under nonvanishing component velocities and step-size conditions; iid sampling instead averages to an ODE. Retaining realized minibatches and sampled curvature yields finite-window exit brackets beyond this regime. In 32 trained ridge-logistic runs, 29 meet prespecified targets with rounding-certified brackets; an enclosure control explains most of the width advantage. Exact-real certificates establish finite means at the original tested radii for three ResNet-18 checkpoints and give local finite-mean conditions for tanh and ViT networks; their upper bounds remain loose, and original-radius ViT mean guarantees remain unresolved despite 128 fresh continuation paths. A two-network replication gives 32/32 successful seeds in each of four finite-path persistence comparisons. Independent equal-length controls retain slower exits after refresh resets than without resets, but the refresh-versus-random ordering reverses with update length at both networks. These interventions do not causally decompose raw momentum SGD. Neural step-size scaling does not validate the switching limit.

## ITU-DESC: a missingness-aware machine-learning dementia risk score in the UK Biobank cohort

DevFeed: [ITU-DESC: a missingness-aware machine-learning dementia risk score in the UK Biobank cohort](<https://devfeed.tech/articles/itu-desc-a-missingness-aware-machine-learning-dementia-risk-score-in-the-uk-biobank-cohort-93646.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.26604>)

Author: Abdullah Kavakl{\\i}, Faik Boray Tek

Published: 2026-10-21T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [catboost](<https://devfeed.tech/topics/catboost.md>), [Feature Engineering](<https://devfeed.tech/topics/feature-engineering.md>), [Predictive AI](<https://devfeed.tech/topics/predictive-ai.md>)

Tags: [catboost](<https://devfeed.tech/tags/catboost.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [feature-selection](<https://devfeed.tech/tags/feature-selection.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [risk](<https://devfeed.tech/tags/risk.md>), [score](<https://devfeed.tech/tags/score.md>), [test](<https://devfeed.tech/tags/test.md>), [uk](<https://devfeed.tech/tags/uk.md>)

### AI overview

ITU-DESC combines broad feature selection with explicit control of missingness bias to predict incident all-cause dementia. In roughly 500,000 UK Biobank participants followed for a median of 16.6 years, unconditional random hot-deck imputation withheld missingness patterns from the model. BorutaSHAP selection was repeated across ten random-imputation seeds; 38 predictors selected in at least eight runs were used to train a CatBoost model. On a 50,041-person holdout set, it achieved average precision of 0.138 and ROC AUC of 0.853, compared with 0.118 and 0.841 for the strongest comparator, the UK Biobank Dementia Risk Prediction model. Newly identified predictors included leukocyte telomere length and liver enzymes.

### Source excerpt

arXiv:2610.26604v1 Announce Type: new Abstract: Background: Existing dementia risk scores rely on fixed, pre-selected features and linear models, or ignore bias from informative missingness in cohort data. Here we introduce the Istanbul Technical University Dementia Risk Score (ITU-DESC), an explainable machine-learning score for incident all-cause dementia that combines broad feature selection across roughly 8,000 candidate features with explicit control of missingness bias. Results: We studied approximately 500,000 UK Biobank participants with a median follow-up of 16.6 years. We withheld missingness patterns from the model using unconditional random hot-deck imputation. We stabilized feature selection by repeating BorutaSHAP-based feature selection with ten random-imputation seeds and selecting predictors chosen in at least eight of the ten seeds, which yielded 38 predictors. We trained a CatBoost gradient-boosted decision-tree model and evaluated it on a holdout test set of 50,041 participants. For all-cause incident dementia, ITU-DESC achieved an average precision of 0.138 (95% confidence interval 0.123-0.156) and an area under the receiver operating characteristic curve (ROC AUC) of 0.853 (0.844-0.863), significantly outperforming other risk scores, including the strongest comparator, the UK Biobank Dementia Risk Prediction model (0.118 and 0.841). We identified dementia risk predictors absent from established scores, including leukocyte telomere length and liver enzymes. Conclusions: ITU-DESC is the first dementia risk score to combine broad feature selection with explicit control of missingness bias.

## Improving Fake Profile Detection Through Multimodal Learning and Research-Guided Feature Integration

DevFeed: [Improving Fake Profile Detection Through Multimodal Learning and Research-Guided Feature Integration](<https://devfeed.tech/articles/improving-fake-profile-detection-through-multimodal-learning-and-research-guided-feature-integration-93196.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.23926>)

Author: Jean Claude Hirwa, Jamiu Dahunsi, Candide Giramata, Lowami Uwimana, Thierry Henry Ishimwe

Published: 2026-10-21T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [fake account fraud](<https://devfeed.tech/topics/fake-account-fraud.md>), [multimodal-ai](<https://devfeed.tech/topics/multimodal-ai.md>), [Feature Engineering](<https://devfeed.tech/topics/feature-engineering.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [catboost](<https://devfeed.tech/topics/catboost.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [bot](<https://devfeed.tech/tags/bot.md>), [bot-detection](<https://devfeed.tech/tags/bot-detection.md>), [catboost](<https://devfeed.tech/tags/catboost.md>), [clip](<https://devfeed.tech/tags/clip.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [detection](<https://devfeed.tech/tags/detection.md>), [fake](<https://devfeed.tech/tags/fake.md>), [fusion](<https://devfeed.tech/tags/fusion.md>), [metadata](<https://devfeed.tech/tags/metadata.md>), [model](<https://devfeed.tech/tags/model.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [multimodal-learning](<https://devfeed.tech/tags/multimodal-learning.md>), [network](<https://devfeed.tech/tags/network.md>), [profile](<https://devfeed.tech/tags/profile.md>)

### AI overview

On a 30,000-user subsample of TwiBot-22, account metadata alone was highly predictive of fake profiles, while adding text and image signals produced only marginal gains that were largely statistically insignificant. The study combines XLM-RoBERTa text embeddings and CLIP image embeddings with account features such as follower counts, posting frequency, and account age, evaluating modality combinations with CatBoost and an R-GCN. The graph model underperformed classical variants under the study's subsampled graph conditions. These findings suggest that, in this setup, metadata richness and feature quality mattered more than multimodal fusion complexity.

### Source excerpt

arXiv:2610.23926v1 Announce Type: new Abstract: Fake accounts continue to undermine online platforms, yet it remains unclear whether multimodal learning offers meaningful improvements over strong metadata baselines or graph-based approaches. In this work, we study bot detection by integrating textual, visual, and account metadata signals on the TwiBot-22 benchmark using a subsample of 30,000 users. We propose a multimodal framework combining pretrained textual (XLM-RoBERTa) and visual (CLIP) embeddings with structured account-level features such as follower counts, posting frequency, and account age, and evaluate multiple modality combinations using CatBoost alongside a Relational Graph Convolutional Network (R-GCN). Results show that metadata alone is highly predictive, while multimodal fusion yields only marginal and largely statistically insignificant gains. The graph-based model underperforms classical variants under the subsampled graph conditions of this study. Overall, our findings suggest that under practical constraints, feature quality and metadata richness matter more than fusion complexity.

## Graph-to-Table Interfaces Control Relational Generalisation

DevFeed: [Graph-to-Table Interfaces Control Relational Generalisation](<https://devfeed.tech/articles/graph-to-table-interfaces-control-relational-generalisation-93581.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.25993>)

Author: Alessandro Cagiano, Lorenzo Molfetta, Gianluca Moro, Matteo Zignani, Pasquale Minervini

Published: 2026-10-21T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [catboost](<https://devfeed.tech/topics/catboost.md>), [Graphs](<https://devfeed.tech/topics/graphs.md>), [Graph Mining](<https://devfeed.tech/topics/graph-mining.md>)

Tags: [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoder](<https://devfeed.tech/tags/decoder.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [features](<https://devfeed.tech/tags/features.md>), [graph](<https://devfeed.tech/tags/graph.md>), [knowledge-graphs](<https://devfeed.tech/tags/knowledge-graphs.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [table](<https://devfeed.tech/tags/table.md>)

### AI overview

Graph-to-table encoder performance depends on the context constructor and decoder, so component comparisons need to hold the rest of the pipeline and evaluation constant. Across logistic regression, CatBoost, and TABPFN-3 experiments, absolute positional encodings fit short CLUTRR paths nearly perfectly but generalise worse than relation-count features when longer paths break the supported-prefix shortcut. On GraphLog, encoder preference varies by decoder, and relation-count features degrade with deeper CatBoost trees. Across 57 knowledge graphs, adding a frozen ULTRA score to local features is consistently useful, while adding local features to the score has a smaller, less stable effect. A pointwise TABPFN-3 readout using the full encoder underperforms native ULTRA on 48 of 57 datasets. The findings also caution that panel-level conclusions depend on evaluation and aggregation protocols.

### Source excerpt

arXiv:2610.25993v1 Announce Type: new Abstract: Graph-to-table systems combine three design choices: an encoder that maps relational evidence to feature rows, a context constructor that selects labelled examples, and a tabular decoder. We investigate whether these components can be evaluated independently through matched comparisons that change one component while holding the rest of the pipeline and the evaluation constant. Across logistic regression, CATBOOST, and TABPFN-3 on CLUTRR, 57 GraphLog worlds, and 57 knowledge graphs, we show that conclusions about one component can change when the others change. On CLUTRR, absolute positional encodings maintain all distinctions among the evaluated queries and fit the short-path setting nearly perfectly; yet, they generalise worse than relation-count features because the supported prefix acts as a shortcut that breaks down on longer paths. On GraphLog, encoder preference depends on the decoder, while the relation-count representation degrades as CatBoost tree depth increases. Across different knowledge graphs, adding the frozen ULTRA score to local features is consistently useful, whereas adding local features to the score has a smaller and less stable effect. A pointwise TABPFN-3 readout using the full encoder underperforms native ULTRA on 48 of 57 datasets. These results show that encoder assessment depends jointly on context and decoder, while panel-level conclusions can also depend on evaluation and aggregation protocols.