This paper introduces a novel PCA sweep procedure for Supervised Semantic Differential (SSD), a method modeling how text meaning varies with individual differences. The sweep systematically selects the optimal number of PCA components to ensure interpretable and stable semantic gradients, illustrated through a case study on AI discourse related to narcissism.

This paper introduces an AI-driven, data-driven methodology for estimating spectrum demand by leveraging both site license and crowdsourced data. The approach uses an enhanced combined proxy, validated against real-world mobile network traffic, to achieve high predictive accuracy, demonstrating its robustness across multiple major Canadian cities for improved spectrum planning.

This paper introduces a resampling technique for trained AI models that leverages invariant transformations of input data to reduce epistemic uncertainty and improve inference accuracy. By aggregating inferences from multiple transformed samples, the method offers a way to enhance performance without re-training, potentially balancing model size and effectiveness.

This paper proposes a novel three-dimensional framework—Source, Depth, and Scope—for categorizing the explanatory requirements of AI applications. This framework aims to standardize the definition of explainable AI, helping to match specific application needs with the capabilities of different machine learning techniques, thereby building trust in AI systems.

This paper introduces Vichara, a novel AI framework for predicting and explaining appellate judgments in the Indian judicial system. It utilizes decision point extraction and a structured explanation format to enhance accuracy and interpretability for legal professionals.

This paper introduces AIFL, a deterministic LSTM-based model for global daily streamflow forecasting, utilizing a novel two-stage training strategy to bridge the performance gap between historical reanalysis data and operational forecast products. It demonstrates competitive predictive skill and exceptional reliability in extreme-event detection, serving as a robust baseline for the global hydrological community.

This paper explores the application of generative AI, specifically conditional flow matching, for the inverse design of ship propellers. It aims to generate multiple propeller geometries that achieve specified performance targets, overcoming limitations of traditional forward models and addressing data scarcity through simulated data and augmentation.

This perspective article explores how modern machine learning techniques, such as convolutional neural networks and transformer architectures, can accelerate the analysis and control of chaotic dynamics. It highlights their potential to overcome the computational limitations of traditional methods in tasks like basin characterization and partial control, opening doors for real-time applications.

This paper introduces an author-assisted mechanism, based on the Isotonic Mechanism, to improve the selection of best paper awards at large machine learning and AI conferences. It relaxes the common convexity assumption for author utility functions, ensuring truthfulness even with simpler monotonicity assumptions, and empirically demonstrates its effectiveness in improving award selection quality through simulations.

This paper introduces ML-Master 2.0, an autonomous agent designed for ultra-long-horizon machine learning engineering. It leverages Hierarchical Cognitive Caching (HCC) to manage context through cognitive accumulation, achieving a state-of-the-art 56.44% medal rate on OpenAI's MLE-Bench.