On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

168,742 papers · 148 categories

Trend · papers per month

count % of corpus

13 results for “SDMs”

Heterogeneous graph neural networks for species distribution modelingcs.LG

Heterogeneous GNN improves species distribution modeling.

problem Predicting species occurrences and habitat suitability using environmental factors.

method Graph Neural Networks (GNN) for presence-only species distribution modeling.

result Heterogeneous GNN model outperforms single-species SDMs and baseline models.

Species distribution modeling Graph Neural Networks Presence-only modeling Environmental factors Fine-grained interactions

2025-03-14abs ↗pdf ↗

StatEcoNet: Statistical Ecology Neural Networks for Species Distribution Modelingcs.LG

StatEcoNet models species distribution using neural networks to correct observation errors.

problem Correcting observation errors in wildlife surveys for accurate species distribution modeling.

method StatEcoNet integrates a graphical generative model with neural networks to address SDM challenges.

result StatEcoNet outperforms traditional methods on simulated and real datasets.

species distribution modeling neural networks observation errors statistical ecology wildlife surveys

2021-02-17abs ↗pdf ↗

Conditional score-based diffusion models for Bayesian inference in infinite dimensionsstat.ML

New method for Bayesian inference in infinite dimensions using SDMs.

problem Bayesian inference in infinite-dimensional function spaces.

method Amortized conditional score-based diffusion models (SDMs).

result Validated approach for large-scale, discretization-invariant Bayesian inference.

Bayesian inference infinite dimensions score-based models SDMs conditional score

2023-05-28abs ↗pdf ↗

Bayesian Off-Policy Evaluation and Learning for Large Action Spacescs.LG

Unified Bayesian framework for efficient off-policy evaluation and learning in large action spaces.

problem Efficient off-policy evaluation and learning in systems with correlated actions.

method Unified Bayesian framework with structured priors and sDM approach.

result sDM leverages action correlations without compromising computational efficiency.

Bayesian Off-Policy Evaluation Large Action Spaces Correlated Actions Online Bayesian Bandits

2024-02-22abs ↗pdf ↗

DiffFlow: A Unified SDE Framework for Score-Based Diffusion Models and Generative Adversarial Networksstat.ML

Unified framework for SDMs and GANs with improved sampling and quality.

problem Limitations of SDMs and GANs in achieving fast sampling and high sample quality.

method Introducing a novel SDE named DiffFlow to describe the learning dynamics of SDMs and GANs, and proving the asymptotic optimality and maximal likelihood training scheme.

result Unified framework allows smooth transition between SDMs and GANs with flexible trade-off between sample quality and speed.

Score-based diffusion models Generative adversarial networks Unified framework Diffusion flow SDE

2023-07-05abs ↗pdf ↗

Score and Distribution Matching Policy: Advanced Accelerated Visuomotor Policies via Matched Distillationcs.RO

SDM Policy accelerates inference for robotic tasks while maintaining high action quality.

problem Prolonged inference times in diffusion-based policies for high-frequency control tasks.

method Two-stage optimization: score matching and distribution matching; dual-teacher mechanism.

result 6x inference speedup with state-of-the-art action quality.

robotics policy learning inference speedup diffusion models visuomotor control

2024-12-12abs ↗pdf ↗

Applying the maximum entropy principle to neural networks enhances multi-species distribution modelscs.LG

DeepMaxent uses neural networks to improve species distribution models.

problem Sampling biases and lack of absence data in presence-only observations.

method DeepMaxent employs neural networks to learn shared features among species using the maximum entropy principle.

result DeepMaxent outperforms traditional methods in predicting species distributions, especially in unevenly sampled regions.

species distribution models deep learning maximum entropy neural networks presence-only data

2024-12-26abs ↗pdf ↗

Diagrams-to-Dynamics (D2D): Exploring Causal Loop Diagram Leverage Points under Uncertaintycs.LG

D2D converts CLDs into SDMs to explore leverage points under uncertainty.

problem Limited dynamic analysis of CLDs for intervention strategies.

method Minimal user input protocol to convert CLDs into SDMs, simulating interventions.

result D2D helps distinguish leverage points and provides uncertainty estimates.

Causal Loop Diagrams System Dynamics Leverage Points Uncertainty Health Research

2025-07-30abs ↗pdf ↗

Prompt-Response Semantic Divergence Metrics for Faithfulness Hallucination and Misalignment Detection in Large Language Modelscs.CL

Paper introduces SDM for detecting LLM hallucinations, improving on entropy tests.

problem Challenges of Large Language Models (LLMs) with non-factual, nonsensical responses.

method Joint clustering on sentence embeddings to measure semantic divergence between prompts and responses.

result SDM framework detects deeper form of arbitrariness in LLM responses.

Large Language Models Semantic Divergence Metrics Hallucinations Confabulations Semantic Entropy

2025-08-13abs ↗pdf ↗

CB-SLICE: Concept-Based Interpretable Error Slice Discoverycs.LG

CB-SLICE identifies concept-based error slices in deep learning models.

problem Systematic errors in deep learning models on specific groups.

method Concept Bottleneck Models (CBMs) and concept representations.

result CB-SLICE outperforms state-of-the-art methods in error slice identification.

deep learning error slices bias mitigation concept-based interpretability

2026-05-28abs ↗pdf ↗

Classification of High-dimensional Time Series in Spectral Domain using Explainable Featuresstat.ML

Proposes a model for classifying high-dimensional time series with interpretable parameters.

problem Challenges in classifying high-dimensional time series, especially in neuroscience.

method Model-based approach using sparsity in inverse spectral density matrices, with interpretability of model parameters.

result Model demonstrates consistency and sure screening property, enabling nuanced inferences.

time series classification spectral domain sparsity interpretability neuroscience

2024-08-15abs ↗pdf ↗

Topic Identification in LLM Input-Output Pairs through the Lens of Information Bottleneckcs.CL

New method identifies shared topics in LLM inputs and outputs for better detection of hallucinations.

problem Detecting semantic drift in LLM responses from context.

method Transformed Deterministic Information Bottleneck (DIB) into UDIB for high-dimensional data.

result UDIB generates more informative topic representations for SDM, improving hallucination detection.

Large Language Models Semantic Divergence Metrics Information Bottleneck K-means Entropy-regularized

2025-08-26abs ↗pdf ↗

An Unconditional Representation of the Conditional Score in Infinite-Dimensional Linear Inverse Problemsstat.ML

UCoS avoids forward model evaluations in sampling for large-scale linear inverse problems.

problem Efficient sampling from posterior distributions in large-scale linear inverse problems.

method UCoS approach that learns a task-dependent score function offline and uses affine transformations to derive the conditional score.

result UCoS eliminates the need for forward model evaluations during sampling, making it more efficient.

Score-based models Bayesian inverse problems Linear inverse problems Unconditional score Infinite-dimensional

2024-05-24abs ↗pdf ↗