🎯 Top Personalized Recommendations
Texas A&M University
Why we think this paper is great for you:
This paper directly addresses building robust and adaptive control strategies, which is highly relevant to ensuring system stability in challenging environments. You will find its focus on resilience in critical infrastructure particularly insightful.
This paper directly addresses building robust and adaptive control strategies, which is highly relevant to ensuring system stability in challenging environments. You will find its focus on resilience in critical infrastructure particularly insightful.
Abstract
Power systems remain highly vulnerable to disturbances and cyber-attacks, underscoring the need for resilient and adaptive control strategies. In this work, we investigate a data-driven Federated Learning Control (FLC) framework for transient stability resilience under cyber-physical disturbances. The FLC employs interpretable neural controllers based on the Chebyshev Kolmogorov-Arnold Network (ChebyKAN), trained on a shared centralized control policy and deployed for distributed execution. Simulation results on the IEEE 39-bus New England system show that the proposed FLC consistently achieves faster stabilization than distributed baselines at moderate control levels (10\%--60\%), highlighting its potential as a scalable, resilient, and interpretable learning-based control solution for modern power grids.
Power systems remain highly vulnerable to disturbances and cyber-attacks, underscoring the need for resilient and adaptive control strategies. In this work, we investigate a data-driven Federated Learning Control (FLC) framework for transient stability resilience under cyber-physical disturbances. The FLC employs interpretable neural controllers based on the Chebyshev Kolmogorov-Arnold Network (ChebyKAN), trained on a shared centralized control policy and deployed for distributed execution. Simulation results on the IEEE 39-bus New England system show that the proposed FLC consistently achieves faster stabilization than distributed baselines at moderate control levels (10\%--60\%), highlighting its potential as a scalable, resilient, and interpretable learning-based control solution for modern power grids.
AI Summary
- The proposed Federated Learning Control (FLC) framework, leveraging interpretable ChebyKANs, consistently achieves faster transient stability stabilization than distributed baselines at moderate control levels (10%–60%). [3]
- The ChebyKAN-based neural controllers provide enhanced interpretability of decision-making, which is crucial for trust and deployment in critical infrastructure like smart grids. [3]
- Chebyshev Kolmogorov-Arnold Network (ChebyKAN): A novel neural architecture that extends Kolmogorov-Arnold Networks by utilizing Chebyshev polynomials instead of B-splines for edge-wise learnable functions, offering improved numerical stability, stronger non-linear approximation, and enhanced interpretability. [3]
- FLC demonstrates robust generalization to unseen fault scenarios, indicating its adaptability and resilience in dynamic power system environments despite being trained on a single fault type. [2]
- FLC offers a smoother power control profile compared to Distributed Parametric Feedback Linearization (DPFL), potentially reducing high-frequency fluctuations that can accelerate battery degradation in energy storage systems. [2]
- The framework balances energy demand, generally offering stronger control responses than DPFL with lower energy consumption than fully centralized control (CPFL) in most evaluated scenarios. [2]
- Performance benefits of FLC diminish at higher distributed control penetration levels (beyond 60%), highlighting a critical need for improved training and federated aggregation strategies for large-scale deployments. [2]
- Transient Stability: The ability of a power system to maintain synchronism and return to a stable operating state following a large disturbance, such as a short-circuit fault. [2]
- The computational overhead of ChebyKAN inference presents a bottleneck for real-time deployment, necessitating optimization of its architecture and execution speed for practical applications. [1]
- Federated Learning Control (FLC): A data-driven framework for transient stability resilience in smart grids, employing interpretable neural controllers trained on a shared centralized control policy and deployed for distributed execution, while preserving data privacy. [1]
University of Connecticut
Why we think this paper is great for you:
This work tackles critical challenges in deploying large-scale models efficiently, especially concerning memory footprint and adaptability to changing conditions. It offers valuable insights into optimizing model inference for real-world applications.
This work tackles critical challenges in deploying large-scale models efficiently, especially concerning memory footprint and adaptability to changing conditions. It offers valuable insights into optimizing model inference for real-world applications.
Abstract
Mixture-of-Experts (MoE) models scale LLM capacity efficiently, but deployment on consumer GPUs is limited by the large memory footprint of inactive experts. Static post-training quantization reduces storage costs but cannot adapt to shifting activation patterns, causing accuracy loss under aggressive compression. So we present DynaExq, a runtime system that treats expert precision as a first-class, dynamically managed resource. DynaExq combines (1) a hotness-aware precision controller that continuously aligns expert bit-widths with long-term activation statistics, (2) a fully asynchronous precision-switching pipeline that overlaps promotion and demotion with MoE computation, and (3) a fragmentation-free memory pooling mechanism that supports hybrid-precision experts with deterministic allocation. Together, these components enable stable, non-blocking precision transitions under strict HBM budgets. Across Qwen3-30B and Qwen3-80B MoE models and six representative benchmarks, DynaExq deploys large LLMs on single RTX 5090 and A6000 GPUs and improves accuracy by up to 4.03 points over static low-precision baselines. The results show that adaptive, workload-aware quantization is an effective strategy for memory-constrained MoE serving.
Mixture-of-Experts (MoE) models scale LLM capacity efficiently, but deployment on consumer GPUs is limited by the large memory footprint of inactive experts. Static post-training quantization reduces storage costs but cannot adapt to shifting activation patterns, causing accuracy loss under aggressive compression. So we present DynaExq, a runtime system that treats expert precision as a first-class, dynamically managed resource. DynaExq combines (1) a hotness-aware precision controller that continuously aligns expert bit-widths with long-term activation statistics, (2) a fully asynchronous precision-switching pipeline that overlaps promotion and demotion with MoE computation, and (3) a fragmentation-free memory pooling mechanism that supports hybrid-precision experts with deterministic allocation. Together, these components enable stable, non-blocking precision transitions under strict HBM budgets. Across Qwen3-30B and Qwen3-80B MoE models and six representative benchmarks, DynaExq deploys large LLMs on single RTX 5090 and A6000 GPUs and improves accuracy by up to 4.03 points over static low-precision baselines. The results show that adaptive, workload-aware quantization is an effective strategy for memory-constrained MoE serving.
Japan
Why we think this paper is great for you:
This report explores how AI foundation models are transforming development environments, including aspects like code generation and testing. You will appreciate its forward-looking perspective on enhancing productivity and tools in the development lifecycle.
This report explores how AI foundation models are transforming development environments, including aspects like code generation and testing. You will appreciate its forward-looking perspective on enhancing productivity and tools in the development lifecycle.
Abstract
Generative Artificial Intelligence (GenAI) models are achieving remarkable performance in various tasks, including code generation, testing, code review, and program repair. The ability to increase the level of abstraction away from writing code has the potential to change the Human-AI interaction within the integrated development environment (IDE). To explore the impact of GenAI on IDEs, 33 experts from the Software Engineering, Artificial Intelligence, and Human-Computer Interaction domains gathered to discuss challenges and opportunities at Shonan Meeting 222. This is the report
Generative Artificial Intelligence (GenAI) models are achieving remarkable performance in various tasks, including code generation, testing, code review, and program repair. The ability to increase the level of abstraction away from writing code has the potential to change the Human-AI interaction within the integrated development environment (IDE). To explore the impact of GenAI on IDEs, 33 experts from the Software Engineering, Artificial Intelligence, and Human-Computer Interaction domains gathered to discuss challenges and opportunities at Shonan Meeting 222. This is the report
University of Gothenburg
Why we think this paper is great for you:
This paper is essential for understanding how to integrate and verify ML components effectively by specifying clear requirements. It offers a practical approach to streamlining the development and validation processes for ML-enabled systems.
This paper is essential for understanding how to integrate and verify ML components effectively by specifying clear requirements. It offers a practical approach to streamlining the development and validation processes for ML-enabled systems.
Abstract
In software engineering processes for machine learning (ML)-enabled systems, integrating and verifying ML components is a major challenge. A prerequisite is the specification of ML component requirements, including models and data, an area where traditional requirements engineering (RE) processes face new obstacles. An underexplored source of RE-relevant information in this context is ML documentation such as ModelCards and DataSheets. However, it is uncertain to what extent RE-relevant information can be extracted from these documents. This study first investigates the amount and nature of RE-relevant information in 20 publicly available ModelCards and DataSheets. We show that these documents contain a significant amount of potentially RE-relevant information. Next, we evaluate how effectively three established RE representations (EARS, Rupp's template, and Volere) can structure this knowledge into requirements. Our results demonstrate that there is a pathway to transform ML-specific knowledge into structured requirements, incorporating ML documentation in software engineering processes for ML systems.
In software engineering processes for machine learning (ML)-enabled systems, integrating and verifying ML components is a major challenge. A prerequisite is the specification of ML component requirements, including models and data, an area where traditional requirements engineering (RE) processes face new obstacles. An underexplored source of RE-relevant information in this context is ML documentation such as ModelCards and DataSheets. However, it is uncertain to what extent RE-relevant information can be extracted from these documents. This study first investigates the amount and nature of RE-relevant information in 20 publicly available ModelCards and DataSheets. We show that these documents contain a significant amount of potentially RE-relevant information. Next, we evaluate how effectively three established RE representations (EARS, Rupp's template, and Volere) can structure this knowledge into requirements. Our results demonstrate that there is a pathway to transform ML-specific knowledge into structured requirements, incorporating ML documentation in software engineering processes for ML systems.
Beihang University
Why we think this paper is great for you:
This research provides methods for improving the reliability of software through iterative testing and verification. You will find its focus on accurate diagnostics and strong verification signals highly relevant to ensuring robust systems.
This research provides methods for improving the reliability of software through iterative testing and verification. You will find its focus on accurate diagnostics and strong verification signals highly relevant to ensuring robust systems.
Abstract
Large language models have advanced software engineering automation, yet resolving real-world software issues remains difficult because it requires repository-level reasoning, accurate diagnostics, and strong verification signals. Existing agent-based and pipeline-based methods often rely on insufficient tests, which can lead to patches that satisfy verification but fail to fix the underlying defect. We present InfCode, an adversarial multi-agent framework for automated repository-level issue resolution. InfCode iteratively refines both tests and patches through adversarial interaction between a Test Patch Generator and a Code Patch Generator, while a Selector agent identifies the most reliable fix. The framework runs inside a containerized environment that supports realistic repository inspection, modification, and validation. Experiments on SWE-bench Lite and SWE-bench Verified using models such as DeepSeek-V3 and Claude 4.5 Sonnet show that InfCode consistently outperforms strong baselines. It achieves 79.4% performance on SWE-bench Verified, establishing a new state-of-the-art. We have released InfCode as an open-source project at https://github.com/Tokfinity/InfCode.
Large language models have advanced software engineering automation, yet resolving real-world software issues remains difficult because it requires repository-level reasoning, accurate diagnostics, and strong verification signals. Existing agent-based and pipeline-based methods often rely on insufficient tests, which can lead to patches that satisfy verification but fail to fix the underlying defect. We present InfCode, an adversarial multi-agent framework for automated repository-level issue resolution. InfCode iteratively refines both tests and patches through adversarial interaction between a Test Patch Generator and a Code Patch Generator, while a Selector agent identifies the most reliable fix. The framework runs inside a containerized environment that supports realistic repository inspection, modification, and validation. Experiments on SWE-bench Lite and SWE-bench Verified using models such as DeepSeek-V3 and Claude 4.5 Sonnet show that InfCode consistently outperforms strong baselines. It achieves 79.4% performance on SWE-bench Verified, establishing a new state-of-the-art. We have released InfCode as an open-source project at https://github.com/Tokfinity/InfCode.
Chalmers
Why we think this paper is great for you:
This paper presents an enhanced machine learning pipeline that considers downstream value and interpretability, crucial for effective model deployment. It offers a comprehensive framework for developing and managing predictive models in practical settings.
This paper presents an enhanced machine learning pipeline that considers downstream value and interpretability, crucial for effective model deployment. It offers a comprehensive framework for developing and managing predictive models in practical settings.
Abstract
Objective: Machine learning (ML) predictive models are often developed without considering downstream value trade-offs and clinical interpretability. This paper introduces a cost-aware prediction (CAP) framework that combines cost-benefit analysis assisted by large language model (LLM) agents to communicate the trade-offs involved in applying ML predictions. Materials and Methods: We developed an ML model predicting 1-year mortality in patients with heart failure (N = 30,021, 22% mortality) to identify those eligible for home care. We then introduced clinical impact projection (CIP) curves to visualize important cost dimensions - quality of life and healthcare provider expenses, further divided into treatment and error costs, to assess the clinical consequences of predictions. Finally, we used four LLM agents to generate patient-specific descriptions. The system was evaluated by clinicians for its decision support value. Results: The eXtreme gradient boosting (XGB) model achieved the best performance, with an area under the receiver operating characteristic curve (AUROC) of 0.804 (95% confidence interval (CI) 0.792-0.816), area under the precision-recall curve (AUPRC) of 0.529 (95% CI 0.502-0.558) and a Brier score of 0.135 (95% CI 0.130-0.140). Discussion: The CIP cost curves provided a population-level overview of cost composition across decision thresholds, whereas LLM-generated cost-benefit analysis at individual patient-levels. The system was well received according to the evaluation by clinicians. However, feedback emphasizes the need to strengthen the technical accuracy for speculative tasks. Conclusion: CAP utilizes LLM agents to integrate ML classifier outcomes and cost-benefit analysis for more transparent and interpretable decision support.
Objective: Machine learning (ML) predictive models are often developed without considering downstream value trade-offs and clinical interpretability. This paper introduces a cost-aware prediction (CAP) framework that combines cost-benefit analysis assisted by large language model (LLM) agents to communicate the trade-offs involved in applying ML predictions. Materials and Methods: We developed an ML model predicting 1-year mortality in patients with heart failure (N = 30,021, 22% mortality) to identify those eligible for home care. We then introduced clinical impact projection (CIP) curves to visualize important cost dimensions - quality of life and healthcare provider expenses, further divided into treatment and error costs, to assess the clinical consequences of predictions. Finally, we used four LLM agents to generate patient-specific descriptions. The system was evaluated by clinicians for its decision support value. Results: The eXtreme gradient boosting (XGB) model achieved the best performance, with an area under the receiver operating characteristic curve (AUROC) of 0.804 (95% confidence interval (CI) 0.792-0.816), area under the precision-recall curve (AUPRC) of 0.529 (95% CI 0.502-0.558) and a Brier score of 0.135 (95% CI 0.130-0.140). Discussion: The CIP cost curves provided a population-level overview of cost composition across decision thresholds, whereas LLM-generated cost-benefit analysis at individual patient-levels. The system was well received according to the evaluation by clinicians. However, feedback emphasizes the need to strengthen the technical accuracy for speculative tasks. Conclusion: CAP utilizes LLM agents to integrate ML classifier outcomes and cost-benefit analysis for more transparent and interpretable decision support.
Rome Tor Vergata
Why we think this paper is great for you:
This work addresses the orchestration of compute resources for AI training, particularly in distributed and energy-conscious environments. It provides valuable insights into building sustainable and efficient machine learning infrastructure.
This work addresses the orchestration of compute resources for AI training, particularly in distributed and energy-conscious environments. It provides valuable insights into building sustainable and efficient machine learning infrastructure.
Abstract
The accelerating expansion of AI workloads is colliding with an energy landscape increasingly dominated by intermittent renewable generation. While vast quantities of zero-carbon energy are routinely curtailed, today's centralized datacenter architectures remain poorly matched to this reality in both energy proportionality and geographic flexibility. This work envisions a shift toward a distributed fabric of renewable-powered micro-datacenters that dynamically follow the availability of surplus green energy through live workload migration. At the core of this vision lies a formal feasibility-domain model that delineates when migratory AI computation is practically achievable. By explicitly linking checkpoint size, wide-area bandwidth, and renewable-window duration, the model reveals that migration is almost always energetically justified, and that time-not energy-is the dominant constraint shaping feasibility. This insight enables the design of a feasibility-aware orchestration framework that transforms migration from a best-effort heuristic into a principled control mechanism. Trace-driven evaluation shows that such orchestration can simultaneously reduce non-renewable energy use and improve performance stability, overcoming the tradeoffs of purely energy-driven strategies. Beyond the immediate feasibility analysis, the extended version explores the architectural horizon of renewable-aware AI infrastructures. It examines the role of emerging ultra-efficient GPU-enabled edge platforms, anticipates integration with grid-level control and demand-response ecosystems, and outlines paths toward supporting partially migratable and distributed workloads. The work positions feasibility-aware migration as a foundational building block for a future computing paradigm in which AI execution becomes fluid, geographically adaptive, and aligned with renewable energy availability.
The accelerating expansion of AI workloads is colliding with an energy landscape increasingly dominated by intermittent renewable generation. While vast quantities of zero-carbon energy are routinely curtailed, today's centralized datacenter architectures remain poorly matched to this reality in both energy proportionality and geographic flexibility. This work envisions a shift toward a distributed fabric of renewable-powered micro-datacenters that dynamically follow the availability of surplus green energy through live workload migration. At the core of this vision lies a formal feasibility-domain model that delineates when migratory AI computation is practically achievable. By explicitly linking checkpoint size, wide-area bandwidth, and renewable-window duration, the model reveals that migration is almost always energetically justified, and that time-not energy-is the dominant constraint shaping feasibility. This insight enables the design of a feasibility-aware orchestration framework that transforms migration from a best-effort heuristic into a principled control mechanism. Trace-driven evaluation shows that such orchestration can simultaneously reduce non-renewable energy use and improve performance stability, overcoming the tradeoffs of purely energy-driven strategies. Beyond the immediate feasibility analysis, the extended version explores the architectural horizon of renewable-aware AI infrastructures. It examines the role of emerging ultra-efficient GPU-enabled edge platforms, anticipates integration with grid-level control and demand-response ecosystems, and outlines paths toward supporting partially migratable and distributed workloads. The work positions feasibility-aware migration as a foundational building block for a future computing paradigm in which AI execution becomes fluid, geographically adaptive, and aligned with renewable energy availability.
Machine Learning Resilience
Mississippi State
Abstract
A key principle in resilience thinking is Embracing Change because change is, indeed, inevitable. In the face of a growing number of disasters, natural and human-made disasters, our critical infrastructures (CIs) are being challenged like never before. This recent trend has sparked a wave of interest among both practitioners and researchers in understanding and delving deeper into the concept of resilience across multiple disciplines. This paper provides an accessible review of these new insights, exploring various frameworks, guidebooks, and methodologies that define resilience through the lens of ecology, engineering, psychology, social science, community, and disaster management during crisis.
A key principle in resilience thinking is Embracing Change because change is, indeed, inevitable. In the face of a growing number of disasters, natural and human-made disasters, our critical infrastructures (CIs) are being challenged like never before. This recent trend has sparked a wave of interest among both practitioners and researchers in understanding and delving deeper into the concept of resilience across multiple disciplines. This paper provides an accessible review of these new insights, exploring various frameworks, guidebooks, and methodologies that define resilience through the lens of ecology, engineering, psychology, social science, community, and disaster management during crisis.
Machine Learning Testing
Hefei University of Techn
Abstract
Computerized Adaptive Testing (CAT) is a widely used technology for evaluating learners' proficiency in online education platforms. By leveraging prior estimates of proficiency to select questions and updating the estimates iteratively based on responses, CAT enables personalized learner modeling and has attracted substantial attention. Despite this progress, most existing works focus primarily on improving diagnostic accuracy, while overlooking the selection bias inherent in the adaptive process. Selection Bias arises because the question selection is strongly influenced by the estimated proficiency, such as assigning easier questions to learners with lower proficiency and harder ones to learners with higher proficiency. Since the selection depends on prior estimation, this bias propagates into the diagnosis model, which is further amplified during iterative updates, leading to misalignment and biased predictions. Moreover, the imbalanced nature of learners' historical interactions often exacerbates the bias in diagnosis models. To address this issue, we propose a debiasing framework consisting of two key modules: Cross-Attribute Examinee Retrieval and Selective Mixup-based Regularization. First, we retrieve balanced examinees with relatively even distributions of correct and incorrect responses and use them as neutral references for biased examinees. Then, mixup is applied between each biased examinee and its matched balanced counterpart under label consistency. This augmentation enriches the diversity of bias-conflicting samples and smooths selection boundaries. Finally, extensive experiments on two benchmark datasets with multiple advanced diagnosis models demonstrate that our method substantially improves both the generalization ability and fairness of question selection in CAT.
Computerized Adaptive Testing (CAT) is a widely used technology for evaluating learners' proficiency in online education platforms. By leveraging prior estimates of proficiency to select questions and updating the estimates iteratively based on responses, CAT enables personalized learner modeling and has attracted substantial attention. Despite this progress, most existing works focus primarily on improving diagnostic accuracy, while overlooking the selection bias inherent in the adaptive process. Selection Bias arises because the question selection is strongly influenced by the estimated proficiency, such as assigning easier questions to learners with lower proficiency and harder ones to learners with higher proficiency. Since the selection depends on prior estimation, this bias propagates into the diagnosis model, which is further amplified during iterative updates, leading to misalignment and biased predictions. Moreover, the imbalanced nature of learners' historical interactions often exacerbates the bias in diagnosis models. To address this issue, we propose a debiasing framework consisting of two key modules: Cross-Attribute Examinee Retrieval and Selective Mixup-based Regularization. First, we retrieve balanced examinees with relatively even distributions of correct and incorrect responses and use them as neutral references for biased examinees. Then, mixup is applied between each biased examinee and its matched balanced counterpart under label consistency. This augmentation enriches the diversity of bias-conflicting samples and smooths selection boundaries. Finally, extensive experiments on two benchmark datasets with multiple advanced diagnosis models demonstrate that our method substantially improves both the generalization ability and fairness of question selection in CAT.
Fault tolerance
Imperial College
Abstract
Ensuring that agents satisfy safety specifications can be crucial in safety-critical environments. While methods exist for controller synthesis with safe temporal specifications, most existing methods restrict safe temporal specifications to probabilistic-avoidance constraints. Formal methods typically offer more expressive ways to express safety in probabilistic systems, such as Probabilistic Computation Tree Logic (PCTL) formulas. Thus, in this paper, we develop a new approach that supports more general temporal properties expressed in PCTL. Our contribution is twofold. First, we develop a theoretical framework for the Synthesis of safe-PCTL specifications. We show how the reducing global specification satisfaction to local constraints, and define CPCTL, a fragment of safe-PCTL. We demonstrate how the expressiveness of CPCTL makes it a relevant fragment for the Synthesis Problem. Second, we leverage these results and propose a new Value Iteration-based algorithm to solve the synthesis problem for these more general temporal properties, and we prove the soundness and completeness of our method.
Ensuring that agents satisfy safety specifications can be crucial in safety-critical environments. While methods exist for controller synthesis with safe temporal specifications, most existing methods restrict safe temporal specifications to probabilistic-avoidance constraints. Formal methods typically offer more expressive ways to express safety in probabilistic systems, such as Probabilistic Computation Tree Logic (PCTL) formulas. Thus, in this paper, we develop a new approach that supports more general temporal properties expressed in PCTL. Our contribution is twofold. First, we develop a theoretical framework for the Synthesis of safe-PCTL specifications. We show how the reducing global specification satisfaction to local constraints, and define CPCTL, a fragment of safe-PCTL. We demonstrate how the expressiveness of CPCTL makes it a relevant fragment for the Synthesis Problem. Second, we leverage these results and propose a new Value Iteration-based algorithm to solve the synthesis problem for these more general temporal properties, and we prove the soundness and completeness of our method.
University of Cape Town
Abstract
Arnold and Arvanitis (2020) introduced a novel bivariate conditionally specified distribution, a distribution in which dependence between two random variables is established by defining the distribution of one variable conditional on the other. This novel conditioning regime was achieved through the use of survival functions, and the approach was termed the accelerated failure conditionals model. In their work, the conditioning framework was constructed using the exponential distribution. Although further generalization was proposed, challenges emerged in deriving the necessary and sufficient conditions for valid joint survival functions. The present study achieves such generalization, extending the conditioning framework to encompass distributional families whose marginal densities may exhibit unimodality and skewness, moving beyond distributional families whose marginal densities are non-increasing. The resulting models are fully specified through closed-form expressions for their moments, with simulations implemented using either a copula-based procedure or the Metropolis-Hastings algorithm. Empirical applications to two datasets, each featuring variables which are unimodal and skewed, demonstrate that the models with flexible, non-monotonic marginal densities yield a superior fit relative to those models with marginal densities restricted to monotonically decaying forms.
Arnold and Arvanitis (2020) introduced a novel bivariate conditionally specified distribution, a distribution in which dependence between two random variables is established by defining the distribution of one variable conditional on the other. This novel conditioning regime was achieved through the use of survival functions, and the approach was termed the accelerated failure conditionals model. In their work, the conditioning framework was constructed using the exponential distribution. Although further generalization was proposed, challenges emerged in deriving the necessary and sufficient conditions for valid joint survival functions. The present study achieves such generalization, extending the conditioning framework to encompass distributional families whose marginal densities may exhibit unimodality and skewness, moving beyond distributional families whose marginal densities are non-increasing. The resulting models are fully specified through closed-form expressions for their moments, with simulations implemented using either a copula-based procedure or the Metropolis-Hastings algorithm. Empirical applications to two datasets, each featuring variables which are unimodal and skewed, demonstrate that the models with flexible, non-monotonic marginal densities yield a superior fit relative to those models with marginal densities restricted to monotonically decaying forms.
Data Science Development Environment and Productivity
University of Minnesota
Abstract
The scientific community increasingly relies on open data sharing, yet existing metrics inadequately capture the true impact of datasets as research outputs. Traditional measures, such as the h-index, focus on publications and citations but fail to account for dataset accessibility, reuse, and cross-disciplinary influence. We propose the X-index, a novel author-level metric that quantifies the value of data contributions through a two-step process: (i) computing a dataset-level value score (V-score) that integrates breadth of reuse, FAIRness, citation impact, and transitive reuse depth, and (ii) aggregating V-scores into an author-level X-index. Using datasets from computational social science, medicine, and crisis communication, we validate our approach against expert ratings, achieving a strong correlation. Our results demonstrate that the X-index provides a transparent, scalable, and low-cost framework for assessing data-sharing practices and incentivizing open science. The X-index encourages sustainable data-sharing practices and gives institutions, funders, and platforms a tangible way to acknowledge the lasting influence of research datasets.
The scientific community increasingly relies on open data sharing, yet existing metrics inadequately capture the true impact of datasets as research outputs. Traditional measures, such as the h-index, focus on publications and citations but fail to account for dataset accessibility, reuse, and cross-disciplinary influence. We propose the X-index, a novel author-level metric that quantifies the value of data contributions through a two-step process: (i) computing a dataset-level value score (V-score) that integrates breadth of reuse, FAIRness, citation impact, and transitive reuse depth, and (ii) aggregating V-scores into an author-level X-index. Using datasets from computational social science, medicine, and crisis communication, we validate our approach against expert ratings, achieving a strong correlation. Our results demonstrate that the X-index provides a transparent, scalable, and low-cost framework for assessing data-sharing practices and incentivizing open science. The X-index encourages sustainable data-sharing practices and gives institutions, funders, and platforms a tangible way to acknowledge the lasting influence of research datasets.
Machine Learning Infrastructure
The Ohio State University
Abstract
ImageNet-1K linear-probe transfer accuracy remains the default proxy for visual representation quality, yet it no longer predicts performance on scientific imagery. Across 46 modern vision model checkpoints, ImageNet top-1 accuracy explains only 34% of variance on ecology tasks and mis-ranks 30% of models above 75% accuracy. We present BioBench, an open ecology vision benchmark that captures what ImageNet misses. BioBench unifies 9 publicly released, application-driven tasks, 4 taxonomic kingdoms, and 6 acquisition modalities (drone RGB, web video, micrographs, in-situ and specimen photos, camera-trap frames), totaling 3.1M images. A single Python API downloads data, fits lightweight classifiers to frozen backbones, and reports class-balanced macro-F1 (plus domain metrics for FishNet and FungiCLEF); ViT-L models evaluate in 6 hours on an A6000 GPU. BioBench provides new signal for computer vision in ecology and a template recipe for building reliable AI-for-science benchmarks in any domain. Code and predictions are available at https://github.com/samuelstevens/biobench and results at https://samuelstevens.me/biobench.
ImageNet-1K linear-probe transfer accuracy remains the default proxy for visual representation quality, yet it no longer predicts performance on scientific imagery. Across 46 modern vision model checkpoints, ImageNet top-1 accuracy explains only 34% of variance on ecology tasks and mis-ranks 30% of models above 75% accuracy. We present BioBench, an open ecology vision benchmark that captures what ImageNet misses. BioBench unifies 9 publicly released, application-driven tasks, 4 taxonomic kingdoms, and 6 acquisition modalities (drone RGB, web video, micrographs, in-situ and specimen photos, camera-trap frames), totaling 3.1M images. A single Python API downloads data, fits lightweight classifiers to frozen backbones, and reports class-balanced macro-F1 (plus domain metrics for FishNet and FungiCLEF); ViT-L models evaluate in 6 hours on an A6000 GPU. BioBench provides new signal for computer vision in ecology and a template recipe for building reliable AI-for-science benchmarks in any domain. Code and predictions are available at https://github.com/samuelstevens/biobench and results at https://samuelstevens.me/biobench.
Online inference
University of Michigan
Abstract
Decision trees are widely used for non-linear modeling, as they capture interactions between predictors while producing inherently interpretable models. Despite their popularity, performing inference on the non-linear fit remains largely unaddressed. This paper focuses on classification trees and makes two key contributions. First, we introduce a novel tree-fitting method that replaces the greedy splitting of the predictor space in standard tree algorithms with a probabilistic approach. Each split in our approach is selected according to sampling probabilities defined by an exponential mechanism, with a temperature parameter controlling its deviation from the deterministic choice given data. Second, while our approach can fit a tree that, with high probability, approximates the fit produced by standard tree algorithms at high temperatures, it is not merely predictive- unlike standard algorithms, it enables valid inference by taking into account the highly adaptive tree structure. Our method produces pivots directly from the sampling probabilities in the exponential mechanism. In theory, our pivots allow asymptotically valid inference on the parameters in the predictive fit, and in practice, our method delivers powerful inference without sacrificing predictive accuracy, in contrast to data splitting methods.
Decision trees are widely used for non-linear modeling, as they capture interactions between predictors while producing inherently interpretable models. Despite their popularity, performing inference on the non-linear fit remains largely unaddressed. This paper focuses on classification trees and makes two key contributions. First, we introduce a novel tree-fitting method that replaces the greedy splitting of the predictor space in standard tree algorithms with a probabilistic approach. Each split in our approach is selected according to sampling probabilities defined by an exponential mechanism, with a temperature parameter controlling its deviation from the deterministic choice given data. Second, while our approach can fit a tree that, with high probability, approximates the fit produced by standard tree algorithms at high temperatures, it is not merely predictive- unlike standard algorithms, it enables valid inference by taking into account the highly adaptive tree structure. Our method produces pivots directly from the sampling probabilities in the exponential mechanism. In theory, our pivots allow asymptotically valid inference on the parameters in the predictive fit, and in practice, our method delivers powerful inference without sacrificing predictive accuracy, in contrast to data splitting methods.
Data Science Development Tools
MTC Web Services
Abstract
Exploring relationships across data sources is a crucial optimization for entities recognition. Since databases can store big amount of information with synthetic and organic data, serving all quantity of objects correctly is an important task to deal with. However, the decision of how to construct entity relationship model is associated with human factor. In this paper, we present flow-oriented entity-relationship tool. This is first and unique end-to-end solution that eliminates routine and resource-intensive problems of processing, creating and visualizing both of explicit and implicit dependencies for prominent SQL dialects on-the-fly. Once launched, FLOWER automatically detects built-in constraints and starting to create own correct and necessary one using dynamic sampling and robust data analysis techniques. This approach applies to improve entity-relationship model and data storytelling to better understand the foundation of data and get unseen insights from DB sources using SQL or natural language. Evaluated on state-of-the-art STATS benchmark, experiments show that FLOWER is superior to reservoir sampling by 2.4x for distribution representation and 2.6x for constraint learning with 2.15x acceleration. For data storytelling, our tool archives 1.19x for accuracy enhance with 1.86x context decrease compare to LLM. Presented tool is also support 23 languages and compatible with both of CPU and GPU. Those results show that FLOWER can manage with real-world data a way better to ensure with quality, scalability and applicability for different use-cases.
Exploring relationships across data sources is a crucial optimization for entities recognition. Since databases can store big amount of information with synthetic and organic data, serving all quantity of objects correctly is an important task to deal with. However, the decision of how to construct entity relationship model is associated with human factor. In this paper, we present flow-oriented entity-relationship tool. This is first and unique end-to-end solution that eliminates routine and resource-intensive problems of processing, creating and visualizing both of explicit and implicit dependencies for prominent SQL dialects on-the-fly. Once launched, FLOWER automatically detects built-in constraints and starting to create own correct and necessary one using dynamic sampling and robust data analysis techniques. This approach applies to improve entity-relationship model and data storytelling to better understand the foundation of data and get unseen insights from DB sources using SQL or natural language. Evaluated on state-of-the-art STATS benchmark, experiments show that FLOWER is superior to reservoir sampling by 2.4x for distribution representation and 2.6x for constraint learning with 2.15x acceleration. For data storytelling, our tool archives 1.19x for accuracy enhance with 1.86x context decrease compare to LLM. Presented tool is also support 23 languages and compatible with both of CPU and GPU. Those results show that FLOWER can manage with real-world data a way better to ensure with quality, scalability and applicability for different use-cases.
Columbia University
Abstract
Human-data interaction (HDI) presents fundamentally different challenges from traditional data management. HDI systems must meet latency, correctness, and consistency needs that stem from usability rather than query semantics; failing to meet these expectations breaks the user experience. Moreover, interfaces and systems are tightly coupled; neither can easily be optimized in isolation, and effective solutions demand their co-design. This dependence also presents a research opportunity: rather than adapt systems to interface demands, systems innovations and database theory can also inspire new interaction and visualization designs. We survey a decade of our lab's work that embraces this coupling and argue that HDI systems are the foundation for reliable, interactive, AI-driven applications.
Human-data interaction (HDI) presents fundamentally different challenges from traditional data management. HDI systems must meet latency, correctness, and consistency needs that stem from usability rather than query semantics; failing to meet these expectations breaks the user experience. Moreover, interfaces and systems are tightly coupled; neither can easily be optimized in isolation, and effective solutions demand their co-design. This dependence also presents a research opportunity: rather than adapt systems to interface demands, systems innovations and database theory can also inspire new interaction and visualization designs. We survey a decade of our lab's work that embraces this coupling and argue that HDI systems are the foundation for reliable, interactive, AI-driven applications.
Machine Learning Operations
University of Stuttgart
Abstract
As machine learning applications grow increasingly ubiquitous and complex, they face an increasing set of requirements beyond accuracy. The prevalent approach to handle this challenge is to aggregate a weighted combination of requirement violation penalties into the training objective. To be effective, this approach requires careful tuning of these hyperparameters (weights), involving trial-and-error and cross-validation, which becomes ineffective even for a moderate number of requirements. These issues are exacerbated when the requirements involve parities or equalities, as is the case in fairness and boundary value problems. An alternative technique uses constrained optimization to formulate these learning problems. Yet, existing approximation and generalization guarantees do not apply to problems involving equality constraints. In this work, we derive a generalization theory for equality-constrained statistical learning problems, showing that their solutions can be approximated using samples and rich parametrizations. Using these results, we propose a practical algorithm based on solving a sequence of unconstrained, empirical learning problems. We showcase its effectiveness and the new formulations enabled by equality constraints in fair learning, interpolating classifiers, and boundary value problems.
As machine learning applications grow increasingly ubiquitous and complex, they face an increasing set of requirements beyond accuracy. The prevalent approach to handle this challenge is to aggregate a weighted combination of requirement violation penalties into the training objective. To be effective, this approach requires careful tuning of these hyperparameters (weights), involving trial-and-error and cross-validation, which becomes ineffective even for a moderate number of requirements. These issues are exacerbated when the requirements involve parities or equalities, as is the case in fairness and boundary value problems. An alternative technique uses constrained optimization to formulate these learning problems. Yet, existing approximation and generalization guarantees do not apply to problems involving equality constraints. In this work, we derive a generalization theory for equality-constrained statistical learning problems, showing that their solutions can be approximated using samples and rich parametrizations. Using these results, we propose a practical algorithm based on solving a sequence of unconstrained, empirical learning problems. We showcase its effectiveness and the new formulations enabled by equality constraints in fair learning, interpolating classifiers, and boundary value problems.
AixMarseille Universit
Abstract
The Kitagawa-Oaxaca-Blinder decomposition splits the difference in means between two groups into an explained part, due to observable factors, and an unexplained part. In this paper, we reformulate this framework using potential outcomes, highlighting the critical role of the reference outcome. To address limitations like common support and model misspecification, we extend Neumark's (1988) weighted reference approach with a doubly robust estimator. Using Neyman orthogonality and double machine learning, our method avoids trimming and extrapolation. This improves flexibility and robustness, as illustrated by two empirical applications. Nevertheless, we also highlight that the decomposition based on the Neumark reference outcome is particularly sensitive to the inclusion of irrelevant explanatory variables.
The Kitagawa-Oaxaca-Blinder decomposition splits the difference in means between two groups into an explained part, due to observable factors, and an unexplained part. In this paper, we reformulate this framework using potential outcomes, highlighting the critical role of the reference outcome. To address limitations like common support and model misspecification, we extend Neumark's (1988) weighted reference approach with a doubly robust estimator. Using Neyman orthogonality and double machine learning, our method avoids trimming and extrapolation. This improves flexibility and robustness, as illustrated by two empirical applications. Nevertheless, we also highlight that the decomposition based on the Neumark reference outcome is particularly sensitive to the inclusion of irrelevant explanatory variables.
Machine Learning Validation
IBM Research
Abstract
Recently, using Large Language Models (LLMs) to generate optimization models from natural language descriptions has became increasingly popular. However, a major open question is how to validate that the generated models are correct and satisfy the requirements defined in the natural language description. In this work, we propose a novel agent-based method for automatic validation of optimization models that builds upon and extends methods from software testing to address optimization modeling . This method consists of several agents that initially generate a problem-level testing API, then generate tests utilizing this API, and, lastly, generate mutations specific to the optimization model (a well-known software testing technique assessing the fault detection power of the test suite). In this work, we detail this validation framework and show, through experiments, the high quality of validation provided by this agent ensemble in terms of the well-known software testing measure called mutation coverage.
Recently, using Large Language Models (LLMs) to generate optimization models from natural language descriptions has became increasingly popular. However, a major open question is how to validate that the generated models are correct and satisfy the requirements defined in the natural language description. In this work, we propose a novel agent-based method for automatic validation of optimization models that builds upon and extends methods from software testing to address optimization modeling . This method consists of several agents that initially generate a problem-level testing API, then generate tests utilizing this API, and, lastly, generate mutations specific to the optimization model (a well-known software testing technique assessing the fault detection power of the test suite). In this work, we detail this validation framework and show, through experiments, the high quality of validation provided by this agent ensemble in terms of the well-known software testing measure called mutation coverage.
Model Monitoring
Mlardalen University
Abstract
Complex software-driven systems often interleave distributed, concurrent computation processes with physical interactions with the environment. Developing these systems more efficiently and safely can be achieved by employing actionable, software-based models. From a high-level system model, engineers often need to derive multiple specialized models for different purposes, including simulation, deployment, and formal verification. Each of these target models usually rely on its own formalism, specification language, and execution platform. Traditionally, a compiler analyzes a program written in a programming language and generates executable code. In contrast, a model compiler processes a source model written in a modeling language and should ideally support the generation of multiple heterogeneous targets. However, most existing modeling languages are designed with a narrow focus, typically targeting only simulation or implementation. Multi-target compilation, when not considered during the language's early design, becomes significantly harder to achieve. In this paper, we introduce our initiative: a toolchain and modeling language called M, designed to support system modeling and multi-target compilation for model-driven engineering of complex, concurrent, and time-aware systems. M is a textual, grammar-driven language based on the actor model and extended with discrete-event scheduling semantics. It provides constructs for modeling system entities, message-based interactions, and time- or state-triggered reactions. From such models, M enables the systematic generation of diverse target artifacts while preserving semantic conformance to the original model. Moreover, M can serve as a middle language to which other modeling languages may anchor, thereby allowing them to benefit from its compilation framework.
Complex software-driven systems often interleave distributed, concurrent computation processes with physical interactions with the environment. Developing these systems more efficiently and safely can be achieved by employing actionable, software-based models. From a high-level system model, engineers often need to derive multiple specialized models for different purposes, including simulation, deployment, and formal verification. Each of these target models usually rely on its own formalism, specification language, and execution platform. Traditionally, a compiler analyzes a program written in a programming language and generates executable code. In contrast, a model compiler processes a source model written in a modeling language and should ideally support the generation of multiple heterogeneous targets. However, most existing modeling languages are designed with a narrow focus, typically targeting only simulation or implementation. Multi-target compilation, when not considered during the language's early design, becomes significantly harder to achieve. In this paper, we introduce our initiative: a toolchain and modeling language called M, designed to support system modeling and multi-target compilation for model-driven engineering of complex, concurrent, and time-aware systems. M is a textual, grammar-driven language based on the actor model and extended with discrete-event scheduling semantics. It provides constructs for modeling system entities, message-based interactions, and time- or state-triggered reactions. From such models, M enables the systematic generation of diverse target artifacts while preserving semantic conformance to the original model. Moreover, M can serve as a middle language to which other modeling languages may anchor, thereby allowing them to benefit from its compilation framework.
Universitt Oldenburg
Abstract
The steadily increasing level of automation in human-centred systems demands rigorous design methods for analysing and controlling interactions between humans and automated components, especially in safety-critical applications. The variability of human behaviour poses particular challenges for formal verification and synthesis. We present a model-based framework that enables design-time exploration of safe shared-control strategies in human-automation systems. The approach combines active automata learning -- to derive coarse, finite-state abstractions of human behaviour from simulations -- with game-theoretic reactive synthesis to determine whether a controller can guarantee safety when interacting with these models. If no such strategy exists, the framework supports iterative refinement of the human model or adjustment of the automation's controllable actions. A driving case study, integrating automata learning with reactive synthesis in UPPAAL, illustrates the applicability of the framework on a simplified driving scenario and its potential for analysing shared-control strategies in human-centred cyber-physical systems.
The steadily increasing level of automation in human-centred systems demands rigorous design methods for analysing and controlling interactions between humans and automated components, especially in safety-critical applications. The variability of human behaviour poses particular challenges for formal verification and synthesis. We present a model-based framework that enables design-time exploration of safe shared-control strategies in human-automation systems. The approach combines active automata learning -- to derive coarse, finite-state abstractions of human behaviour from simulations -- with game-theoretic reactive synthesis to determine whether a controller can guarantee safety when interacting with these models. If no such strategy exists, the framework supports iterative refinement of the human model or adjustment of the automation's controllable actions. A driving case study, integrating automata learning with reactive synthesis in UPPAAL, illustrates the applicability of the framework on a simplified driving scenario and its potential for analysing shared-control strategies in human-centred cyber-physical systems.
Machine Learning Lifecycle
University of Tartu
Abstract
Contemporary machine learning models, including large language models, exhibit remarkable capabilities in static tasks yet falter in non-stationary environments due to rigid architectures that hinder continual adaptation and lifelong learning. Building upon the nested learning paradigm, which decomposes models into multi-level optimization problems with fixed update frequencies, this work proposes dynamic nested hierarchies as the next evolutionary step in advancing artificial intelligence and machine learning. Dynamic nested hierarchies empower models to autonomously adjust the number of optimization levels, their nesting structures, and update frequencies during training or inference, inspired by neuroplasticity to enable self-evolution without predefined constraints. This innovation addresses the anterograde amnesia in existing models, facilitating true lifelong learning by dynamically compressing context flows and adapting to distribution shifts. Through rigorous mathematical formulations, theoretical proofs of convergence, expressivity bounds, and sublinear regret in varying regimes, alongside empirical demonstrations of superior performance in language modeling, continual learning, and long-context reasoning, dynamic nested hierarchies establish a foundational advancement toward adaptive, general-purpose intelligence.
Contemporary machine learning models, including large language models, exhibit remarkable capabilities in static tasks yet falter in non-stationary environments due to rigid architectures that hinder continual adaptation and lifelong learning. Building upon the nested learning paradigm, which decomposes models into multi-level optimization problems with fixed update frequencies, this work proposes dynamic nested hierarchies as the next evolutionary step in advancing artificial intelligence and machine learning. Dynamic nested hierarchies empower models to autonomously adjust the number of optimization levels, their nesting structures, and update frequencies during training or inference, inspired by neuroplasticity to enable self-evolution without predefined constraints. This innovation addresses the anterograde amnesia in existing models, facilitating true lifelong learning by dynamically compressing context flows and adapting to distribution shifts. Through rigorous mathematical formulations, theoretical proofs of convergence, expressivity bounds, and sublinear regret in varying regimes, alongside empirical demonstrations of superior performance in language modeling, continual learning, and long-context reasoning, dynamic nested hierarchies establish a foundational advancement toward adaptive, general-purpose intelligence.
Interests not found
We did not find any papers that match the below interests.
Try other terms also consider if the content exists in arxiv.org.
- MLOps
You can edit or add more interests any time.
Help us improve your experience!
This project is on its early stages your feedback can be pivotal on the future of the project.
Let us know what you think about this week's papers and suggestions!
Give Feedback