Large Language Models (LLMs) have become indispensable across academia, industry, and daily applications, yet current evaluation methods struggle to keep pace with their rapid development. One core challenge of evaluation in the large language model (LLM) era is the generalization issue: how to infer a model's near-unbounded abilities from inevitably bounded benchmarks. We address this challenge by proposing Model Utilization Index (MUI), a mechanism interpretability enhanced metric that complements traditional performance scores. MUI quantifies the effort a model expends on a task, defined as the proportion of activated neurons or features during inference. Intuitively, a truly capable model should achieve higher performance with lower effort. Extensive experiments across popular LLMs reveal a consistent inverse logarithmic relationship between MUI and performance, which we formulate as the Utility Law. From this law we derive four practical corollaries that (i) guide training diagnostics, (ii) expose data contamination issue, (iii) enable fairer model comparisons, and (iv) design model-specific dataset diversity. Our code can be found at https://github.com/ALEX-nlp/MUI-Eva.

Despite advances in designing personas for Large Language Models (LLM), challenges remain in aligning them with human cognitive processes and representing diverse stakeholder perspectives. We introduce a Social Cognitive Theory (SCT) agent design framework for designing, evaluating, and implementing psychologically grounded LLMs with consistent behavior. Our framework operationalizes SCT through four personal factors (cognitive, motivational, biological, and affective) for designing, six quantifiable constructs for evaluating, and a graph database-backed architecture for implementing stakeholder personas. Experiments tested agents' responses to contradicting information of varying reliability. In the highly polarized renewable energy transition discourse, we design five diverse agents with distinct ideologies, roles, and stakes to examine stakeholder representation. The evaluation of these agents in contradictory scenarios occurs through comprehensive processes that implement the SCT. Results show consistent response patterns ($R^2$ range: $0.58-0.61$) and systematic temporal development of SCT construct effects. Principal component analysis identifies two dimensions explaining $73$% of variance, validating the theoretical structure. Our framework offers improved explainability and reproducibility compared to black-box approaches. This work contributes to ongoing efforts to improve diverse stakeholder representation while maintaining psychological consistency in LLM personas.

Uncertainty Quantification (UQ) in wind speed forecasting is a critical challenge in wind power production due to the inherently volatile nature of wind. By quantifying the associated risks and returns, UQ supports more effective decision-making for grid operations and participation in the electricity market. In this paper, we design a sequence of deep learning based probabilistic forecasting methods by using the Tube loss function for wind speed forecasting. The Tube loss function is a simple and model agnostic Prediction Interval (PI) estimation approach and can obtain the narrow PI with asymptotical coverage guarantees without any distribution assumption. Our deep probabilistic forecasting models effectively incorporate popular architectures such as LSTM, GRU, and TCN within the Tube loss framework. We further design a simple yet effective heuristic for tuning the $δ$ parameter of the Tube loss function so that our deep forecasting models obtain the narrower PI without compromising its calibration ability. We have considered three wind datasets, containing the hourly recording of the wind speed, collected from three distinct location namely Jaisalmer, Los Angeles and San Fransico. Our numerical results demonstrate that the proposed deep forecasting models produce more reliable and narrower PIs compared to recently developed probabilistic wind forecasting methods.

Complex tasks are increasingly delegated to ensembles of specialized LLM-based agents that reason, communicate, and coordinate actions-both among themselves and through interactions with external tools, APIs, and databases. While persistent memory has been shown to enhance single-agent performance, most approaches assume a monolithic, single-user context-overlooking the benefits and challenges of knowledge transfer across users under dynamic, asymmetric permissions. We introduce Collaborative Memory, a framework for multi-user, multi-agent environments with asymmetric, time-evolving access controls encoded as bipartite graphs linking users, agents, and resources. Our system maintains two memory tiers: (1) private memory-private fragments visible only to their originating user; and (2) shared memory-selectively shared fragments. Each fragment carries immutable provenance attributes (contributing agents, accessed resources, and timestamps) to support retrospective permission checks. Granular read policies enforce current user-agent-resource constraints and project existing memory fragments into filtered transformed views. Write policies determine fragment retention and sharing, applying context-aware transformations to update the memory. Both policies may be designed conditioned on system, agent, and user-level information. Our framework enables safe, efficient, and interpretable cross-user knowledge sharing, with provable adherence to asymmetric, time-varying policies and full auditability of memory operations.

Bias in geospatial artificial intelligence (GeoAI) models has been documented, yet the evidence is scattered across narrowly focused studies. We synthesize this fragmented literature to provide a concise overview of bias in GeoAI and examine how the EU's Artificial Intelligence Act (EU AI Act) shapes audit obligations. We discuss recurring bias mechanisms, including representation, algorithmic and aggregation bias, and map them to specific provisions of the EU AI Act. By applying the Act's high-risk criteria, we demonstrate that widely deployed GeoAI applications qualify as high-risk systems. We then present examples of recent audits along with an outline of practical methods for detecting bias. As far as we know, this study represents the first integration of GeoAI bias evidence into the EU AI Act context, by identifying high-risk GeoAI systems and mapping bias mechanisms to the Act's Articles. Although the analysis is exploratory, it suggests that even well-curated European datasets should employ routine bias audits before 2027, when the AI Act's high-risk provisions take full effect.

Electric energy is difficult to store, requiring stricter control over its generation, transmission, and distribution. A persistent challenge in power systems is maintaining real-time equilibrium between electricity demand and supply. Oversupply contributes to resource wastage, while undersupply can strain the grid, increase operational costs, and potentially impact service reliability. To maintain grid stability, load forecasting is needed. Accurate load forecasting balances generation and demand by striving to predict future electricity consumption. This thesis examines and evaluates four machine learning frameworks for short term load forecasting, including gradient boosting decision tree methods such as Extreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (LightGBM). A hybrid framework is also developed. In addition, two recurrent neural network architectures, Long Short Term Memory (LSTM) networks and Gated Recurrent Units (GRU), are designed and implemented. Pearson Correlation Coefficient is applied to assess the relationships between electricity demand and exogenous variables. The experimental results show that, for the specific dataset and forecasting task in this study, machine learning-based models achieved improved forecasting performance compared to a classical ARIMA baseline.

The performance of battery materials is determined by their composition and the processing conditions employed during commercial-scale fabrication, where raw materials undergo complex processing steps with various additives to yield final products. As the complexity of these parameters expands with the development of industry, conventional one-factor-at-a-time (OFAT) experiment becomes old fashioned. While domain expertise aids in parameter optimization, this traditional approach becomes increasingly vulnerable to cognitive limitations and anthropogenic biases as the complexity of factors grows. Herein, we introduce an iterative machine learning (ML) framework that integrates active learning to guide targeted experimentation and facilitate incremental model refinement. This method systematically leverages comprehensive experimental observations, including both successful and unsuccessful results, effectively mitigating human-induced biases and alleviating data scarcity. Consequently, it significantly accelerates exploration within the high-dimensional design space. Our results demonstrate that active-learning-driven experimentation markedly reduces the total number of experimental cycles necessary, underscoring the transformative potential of ML-based strategies in expediting battery material optimization.

This paper investigates the application of machine learning regression algorithms Kernel Ridge Regression (KRR), Huber Regressor (HR), and Gaussian Process Regression (GPR) for predicting sound power levels of gensets, offering significant value for marketing and sales teams during the early bidding process. When engine sizes and genset enclosure dimensions are tentative, and measured noise data is unavailable, these algorithms enable reliable noise level estimation for unbuilt gensets. The study utilizes high fidelity datasets from over 100 experiments conducted at Cummins Acoustics Technology Center (ATC) in a hemi-anechoic chamber, adhering to ISO 3744 standards. By using readily available information from the bidding and initial design stages, KRR predicts sound power with an average accuracy of within 5 dBA. While HR and GPR show slightly higher prediction errors, all models effectively capture the overall noise trends across various genset configurations. These findings present a promising method for early-stage noise estimation in genset design.

Time series forecasting plays a critical role in domains such as energy, finance, and healthcare, where accurate predictions inform decision-making under uncertainty. Although Transformer-based models have demonstrated success in sequential modeling, their adoption for time series remains limited by challenges such as noise sensitivity, long-range dependencies, and a lack of inductive bias for temporal structure. In this work, we present a unified and principled framework for benchmarking three prominent Transformer forecasting architectures-Autoformer, Informer, and Patchtst-each evaluated through three architectural variants: Minimal, Standard, and Full, representing increasing levels of complexity and modeling capacity. We conduct over 1500 controlled experiments on a suite of ten synthetic signals, spanning five patch lengths and five forecast horizons under both clean and noisy conditions. Our analysis reveals consistent patterns across model families. To advance this landscape further, we introduce the Koopman-enhanced Transformer framework, Deep Koopformer, which integrates operator-theoretic latent state modeling to improve stability and interpretability. We demonstrate its efficacy on nonlinear and chaotic dynamical systems. Our results highlight Koopman based Transformer as a promising hybrid approach for robust, interpretable, and theoretically grounded time series forecasting in noisy and complex real-world conditions.

Ensuring safe operation of safety-critical complex systems interacting with their environment poses significant challenges, particularly when the system's world model relies on machine learning algorithms to process the perception input. A comprehensive safety argumentation requires knowledge of how faults or functional insufficiencies propagate through the system and interact with external factors, to manage their safety impact. While statistical analysis approaches can support the safety assessment, associative reasoning alone is neither sufficient for the safety argumentation nor for the identification and investigation of safety measures. A causal understanding of the system and its interaction with the environment is crucial for safeguarding safety-critical complex systems. It allows to transfer and generalize knowledge, such as insights gained from testing, and facilitates the identification of potential improvements. This work explores using causal Bayesian networks to model the system's causalities for safety analysis, and proposes measures to assess causal influences based on Pearl's framework of causal inference. We compare the approach of causal Bayesian networks to the well-established fault tree analysis, outlining advantages and limitations. In particular, we examine importance metrics typically employed in fault tree analysis as foundation to discuss suitable causal metrics. An evaluation is performed on the example of a perception system for automated driving. Overall, this work presents an approach for causal reasoning in safety analysis that enables the integration of data-driven and expert-based knowledge to account for uncertainties arising from complex systems operating in open environments.