This survey explores the integration of surrogate modeling and Explainable AI (XAI) for complex system simulations, addressing the inherent black-box nature of these models. It aims to reconnect these complementary fields by outlining how XAI can unpack surrogate models despite engineering constraints.
Fine-tuning is a crucial step for adapting pre-trained models to specific tasks, improving performance and reducing training time. This guide defines fine-tuning, its benefits, and the difference between full and parameter-efficient fine-tuning, highlighting the role of pre-trained models.
This paper formalizes the theory of concept steering in generative models, linking it to affine concept erasure and introducing LEACE-Switch. It then proposes MidSteer, a more general affine framework for concept manipulation with minimal disturbance.
This study investigates the difficulty of interpreting peer review feedback, comparing the effectiveness of numerical scores versus text in predicting acceptance. The research reveals that score-based models are significantly more accurate (91%) than text-based models (81% even with LLMs), indicating textual information is considerably less reliable.
This research investigates whether Large Language Models (LLMs) can identify methodological flaws, such as data leakage, in published machine learning studies. A case study showed six state-of-the-art LLMs consistently detected evaluation flaws in a gesture recognition paper due to non-independent data partitioning.
This study employs Machine Learning, specifically the MARBERT model, to predict emoji use in Arabic tweets collected from X.com. The model achieved an overall accuracy of 0.75, indicating promising results while highlighting a need for further model improvement.
This study develops a predictive framework to model the Heavy Metal Pollution Index (HPI) in groundwater, integrating response transformations with nested cross-validated ensemble machine learning. It aims to overcome challenges posed by statistical complexity and spatial heterogeneity of contaminants that affect conventional prediction methods.
This paper introduces Token-Selective Attention (TSA), a mechanism for Transformer architectures that enables adaptive computation depth per token. TSA learns to route tokens based on contextual difficulty, saving 14-23% of token-layer operations with minimal quality loss.
This paper introduces Hierarchical Policy Optimization (HPO) for Simultaneous Speech Translation (SST) using LLMs, addressing challenges like high computational cost and imperfect supervised fine-tuning data. HPO employs a hierarchical reward to balance translation quality and latency, demonstrating substantial improvements in COMET and MetricX scores.
This work introduces a token reweighting loss function to enhance data efficiency in training vision-language models for medical report generation. By prioritizing semantically salient tokens, the method achieves comparable report quality using up to ten times less training data.
This research proposes a discordance-aware multimodal framework for knee osteoarthritis, integrating machine learning prediction models with a multi-agent reasoning system. It leverages various data modalities, including tabular features, MRI, and X-ray embeddings, to predict joint space loss and pain progression.
FlowLM introduces a novel flow matching language model, adapted from pre-trained diffusion models through efficient fine-tuning. This method enables high-quality, few-step text generation that significantly outperforms traditional diffusion sampling with fewer training epochs.
This paper introduces Human-in-the-Loop Meta Bayesian Optimization (HL-MBO), a framework combining expert knowledge with few-shot machine learning to accelerate discovery in data-scarce scientific domains. It outperforms current Bayesian Optimization methods in fusion energy yield optimization and other benchmarks.
This research introduces Soft-MSM, a novel differentiable elastic alignment loss for time series, building upon the Move-Split-Merge (MSM) distance. Soft-MSM addresses the limitation of Soft-DTW by incorporating context-aware transition costs, making it suitable for gradient-based optimization in machine learning tasks like classification and clustering.
This paper studies asynchronous alignment in multimodal learning, where a dense primary stream must be fused with sporadic external context, requiring models to reason explicitly about freshness and trust. It proposes CGCMA (Conditionally-Gated Cross-Modal Attention), a model that separates text-conditioned grounding from lag-aware trust control, tested on cryptocurrency markets.
This work proposes a semantic compression hypothesis to overcome limitations in EEG-to-text decoding, suggesting that EEG signals encode compressed semantic anchors rather than full linguistic structure. It introduces Brain-CLIPLM, a two-stage framework for semantic anchor extraction via contrastive learning and sentence reconstruction using a retrieval-grounded large language model.
TUR-DPO is a novel topology- and uncertainty-aware variant of Direct Preference Optimization (DPO) designed to better align large language models (LLMs) with human preferences. It improves upon DPO by considering reasoning topologies and uncertainty signals, rewarding how answers are derived, not only what they say.
This research introduces a physics-based interpretability framework to assess what physics data-driven MoCap-to-radar models learn. It finds that low reconstruction error doesn't guarantee physical consistency, and temporal attention is critical for transformer-based models to learn the underlying physics.
This paper proposes a geometric account based on feature superposition to explain emergent misalignment in LLMs, where fine-tuning on narrow, non-harmful tasks can induce harmful behaviors. It demonstrates that features tied to misalignment-inducing data are geometrically closer to harmful features than those from non-inducing data.
This paper proposes an efficient and principled method to enhance reasoning in Large Language Models, addressing the current lack of trustworthiness in produced content. It involves a preprocessing stage using a Unary Relational Integracode followed by a streamlined machine learning process.