AINeutralarXiv – CS AI · Jun 96/10
🧠Researchers introduce SAILS, a model-agnostic framework that goes beyond detecting feature interactions in machine learning models to reveal their functional forms and characteristics. Using surrogate generalized additive models, SAILS categorizes interactions as linear, product-separable, or non-product-separable and provides tailored visualizations, advancing the field of explainable AI.
AINeutralarXiv – CS AI · Jun 96/10
🧠Researchers propose a fine-tuned speech language model that provides both multi-level L2 English proficiency assessment and natural-language explanations for its predictions. The model demonstrates competitive performance on standard benchmarks while offering improved interpretability, though generated rationales show lower reliability at granular word-level assessments.
AINeutralarXiv – CS AI · Jun 96/10
🧠Researchers propose a methodology for validating attention-head circuits in large language models by combining co-activation clustering with causal ablation testing. Their findings reveal that while clustering signals identify circuit proposals, true circuit validation requires closure tests that measure functional impact through ablation—a distinction that challenges current interpretability approaches.
AINeutralarXiv – CS AI · Jun 96/10
🧠Researchers demonstrate that Concept Bottleneck Models and Sparse Autoencoders, two distinct interpretability approaches in machine learning, share an underlying geometric structure based on concept cones. This unification enables quantitative evaluation of how well unsupervised concept discovery aligns with human-defined concepts, advancing AI interpretability standards.
AINeutralarXiv – CS AI · Jun 96/10
🧠Researchers introduce Hyperflux, a novel L0 pruning method that models neural network pruning as a dynamically evolving system driven by flux and pressure mechanisms. The approach provides interpretability at multiple scales while achieving competitive sparsity results on standard vision benchmarks, advancing understanding of how neural networks can be efficiently compressed.
AINeutralarXiv – CS AI · Jun 96/10
🧠Researchers propose a method to improve NLP benchmark understanding by extracting executable representations (computables) that provide operational evidence of semantic adequacy beyond traditional text-based reasoning. The approach demonstrates consistent improvements over baseline methods across mathematical reasoning, legal, and biomedical benchmarks while offering inspectable semantic evidence.
AINeutralarXiv – CS AI · Jun 96/10
🧠Researchers introduce an information-theoretic framework to measure representational ambiguity in neural networks, demonstrating that network connectivity structures can encode unambiguous content independent of behavioral performance. Using MNIST classification experiments, they achieve 100% accuracy in identifying output neuron class identity from relational structure alone in dropout-trained networks, suggesting neural systems can exhibit the low-ambiguity representations theorized as necessary for consciousness.
AINeutralarXiv – CS AI · Jun 85/10
🧠Researchers demonstrate that instruction-following audio language models can effectively utilize explicit acoustic cues for speech emotion recognition, with aligned acoustic tokens improving performance on standard benchmarks while remaining grounded in the underlying audio signal.
AINeutralarXiv – CS AI · Jun 86/10
🧠Researchers demonstrate that large language models exhibit Endogenous Steering Resistance (ESR), the ability to detect and recover from activation-space steering attempts mid-generation, with Llama-3.3-70B showing explicit resistance in over half of cases. The discovery reveals both a potential safety feature against adversarial manipulation and a complication for beneficial steering-based interventions, since models cannot distinguish between malicious and helpful steering.
🧠 Llama
AINeutralarXiv – CS AI · Jun 86/10
🧠Researchers developed an AI-enhanced diagnostic system for traditional Chinese medicine that combines Neo4j knowledge graphs, large language models, and multimodal visualization to improve diagnostic transparency and treatment planning. The system demonstrated a 32% reduction in non-standard outputs and significantly improved diagnostic trust and credibility compared to existing tools.
AINeutralarXiv – CS AI · Jun 86/10
🧠Researchers propose the Glassbox Framework, a new AI architecture that replaces post-hoc explainability with ante-hoc probabilistic mediation using Bayesian networks as transparent reasoning layers for large language models. This approach aims to make AI systems fundamentally accountable in high-stakes domains like healthcare, law, and public administration by encoding domain knowledge and causal assumptions before inference occurs.
AINeutralarXiv – CS AI · Jun 86/10
🧠Researchers propose a novel attention-guided encoder-decoder architecture for longitudinal medical visual question answering using chest X-rays, incorporating affine registration and vision foundation models (DINO) to identify anatomical changes over time. The approach combines saliency masking with multimodal transformer decoding and auxiliary learning objectives, achieving strong benchmark performance while providing interpretable visual explanations for clinical reasoning.
AINeutralarXiv – CS AI · Jun 86/10
🧠Researchers present MSAIC-Net, a deep learning framework that improves ECG-based detection of myocardial substrate abnormalities like scarring and heart attacks. The model combines multi-scale attention mechanisms with contrastive learning to address class imbalance and interpretability challenges, demonstrating strong performance on both institutional and public datasets.
AINeutralarXiv – CS AI · Jun 86/10
🧠Researchers introduce Product-Unit Residual Networks (PURe), a neural architecture that explicitly models nonlinear feature interactions through multiplicative units combined with residual connections. The approach demonstrates improved interpretability, robustness to noise, and sample efficiency compared to standard MLPs across synthetic and real-world datasets.
AINeutralarXiv – CS AI · Jun 56/10
🧠Researchers demonstrate that large language models can reliably self-recognize their own outputs through implicit signals encoded in generated text, and this capability can be amplified through targeted steering of internal activation patterns. By injecting sparse random vectors into a model's residual stream during generation, they create detectable fingerprints enabling attribution to specific LLMs with over 98% accuracy while maintaining text quality. This approach offers a practical alternative to traditional AI-generated content detection by leveraging models' natural representation structures.
AINeutralarXiv – CS AI · Jun 56/10
🧠Researchers propose a framework combining SHAP explainability with LLM-generated rationales to improve transparency in automated rubric-based scoring systems for educational assessment. Testing on classroom transcripts reveals fine-tuned language models outperform LLMs in accuracy, but SHAP attributions provide more faithful and transferable explanations than LLM rationales across different model architectures.
AINeutralarXiv – CS AI · Jun 56/10
🧠ReasoningFlow is a framework that maps the complex, non-linear reasoning traces of large reasoning models into directed acyclic graphs, enabling better understanding and monitoring of AI reasoning processes. Through analysis of 1,260 traces across multiple models and tasks, researchers discovered that LRMs exhibit structurally similar reasoning patterns despite different training origins, while most erroneous steps don't influence final answers.
AINeutralarXiv – CS AI · Jun 56/10
🧠Researchers propose HPME, a novel framework for explaining Graph Neural Network decisions using hard-perturbation mixup strategies instead of soft masks. The method addresses out-of-distribution issues in GNN explainability by extracting discrete subgraphs and employing structure-level replacement, achieving improved explanation fidelity across synthetic and real-world datasets.
AINeutralarXiv – CS AI · Jun 56/10
🧠Researchers introduce CausalPhys, a benchmark with over 3,000 curated video and image questions designed to evaluate how well vision-language models understand causal physical reasoning. The work includes expert-annotated causal graphs and proposes Causal Rationale-informed Fine-Tuning (CRFT) to improve VLM performance on physical world reasoning tasks.
AINeutralarXiv – CS AI · Jun 56/10
🧠Researchers propose a metamorphic testing framework to evaluate the trustworthiness of machine learning model explanations by identifying inconsistencies between model predictions and feature attributions, addressing the Rashomon effect where multiple models achieve similar performance but yield conflicting explanations.
AINeutralarXiv – CS AI · Jun 56/10
🧠Researchers propose using emergent language in multi-agent reinforcement learning as a methodology to study artificial consciousness, where agents develop communication from minimal constraints to reveal whether consciousness-relevant structures arise from task demands rather than human language biases. A proof-of-concept demonstrates agents spontaneously develop self-referential communication and an echo-mismatch detection mechanism, suggesting genuine cognitive emergence rather than inherited patterns.
AINeutralarXiv – CS AI · Jun 56/10
🧠Researchers introduce FreqX, a novel interpretability method for machine learning models that leverages signal processing and information theory to address challenges in personalized federated learning. The approach achieves 10x faster performance than existing methods while providing both attribution and concept information while maintaining privacy.
AINeutralarXiv – CS AI · Jun 56/10
🧠Researchers introduce Learning-to-Theorize, a new AI paradigm that builds explicit explanatory theories of the world from observations rather than simply predicting future states. The Neural Theorizer (NEO) model represents understanding as executable, compositional programs whose learned primitives can be recombined to explain novel phenomena, enabling explanation-driven generalization.
AINeutralarXiv – CS AI · Jun 46/10
🧠Researchers introduce an affinity-based reinforcement learning approach tested in the board game Fog of Love, demonstrating that localized affinities enable AI agents to balance competitive and cooperative objectives simultaneously. This advancement moves virtuous AI behavior engineering from simplified toy environments to more complex multi-agent scenarios, improving agent interpretability and performance in nuanced social settings.
AINeutralarXiv – CS AI · Jun 46/10
🧠Researchers propose a counterfactual explanation framework for deep two-sample testing that generates interpretable edits to show which data features drive statistical differences between groups. The method combines diffusion autoencoders with deep learning models to produce plausible sample transformations that reduce distributional discrepancies, validated on synthetic data and MRI cohorts.
