Instruction-following is essential for aligning large language models (LLMs) with user intent.Yet recent reasoning-oriented models, despite their strong performance on complex mathematical problems, often fail to comply with simple natural language directives.In this work, we analyze the interaction between reasoning ability and instruction adherence in large reasoning models (LRMs).Using a controlled evaluation framework (MathIF), we uncover a persistent trade-off: as models scale reasoning capacity through long chains-of-thought or reinforcement learning on reasoning traces, their obedience to instructions degrades, particularly when generation length grows.We further show that interventions such as constraining or repeating instructions can partially restore compliance, but typically at the expense of reasoning performance.Taken together, our findings expose a dilemma between intelligence and obedience in current training paradigms and underscore the need for instruction-aware approaches to developing controllable reasoning models.

This article introduces a ground breaking approach to rotating machinery fault diagnosis by addressing the critical, yet unexplored challenge of subpopulation shift. We present the first study to consider this in the domain, introducing a novel framework combining label propagation with time–frequency consistency regularization. Motivated by limitations of existing domain adaptation methods, we propose a unique dataset partitioning strategy that models subpopulation structures within fault categories. Our approach leverages a bridging distribution to facilitate knowledge transfer across domains with different subpopulation compositions. Theoretical analysis provides performance guarantees, while experiments on real-world bearing datasets demonstrate superior performance across various transfer learning scenarios. The proposed method consistently outperforms state-of-the-art techniques in multiple adaptation settings. By pioneering subpopulation shift consideration and introducing an innovative dataset preparation method, this work significantly advances rotating machinery fault diagnosis, offering a more reliable solution for complex industrial applications. The proposed framework directly addresses critical industrial challenges by enabling robust fault diagnosis across varying operating conditions, which helps reduce maintenance costs and prevent unexpected equipment failures in manufacturing plants.

Recent LLMs exhibit limited effectiveness on molecular property prediction task due to the semantic gap between molecular representations and natural language, as well as the lack of domain-specific knowledge.To address these challenges, we propose MolRAG, a Retrieval-Augmented Generation framework integrating Chain-of-Thought reasoning for molecular property prediction.MolRAG operates by retrieving structurally analogous molecules as contextual references to guide stepwise knowledge reasoning through chemical structureproperty relationships.This dual mechanism synergizes molecular similarity analysis with structured inference, while generating humaninterpretable rationales grounded in domain knowledge.Experimental results show Mol-RAG outperforms pre-trained LLMs on four datasets, and even matches supervised methods, achieving performance gains of 1.1%-45.7%over direct prediction approaches, demonstrating versatile effectiveness.Our code is available at https://github.com/AcaciaSin/MolRAG.

Out-of-Distribution (OOD) detection is critical for safely deploying deep models in open-world environments, where inputs may lie outside the training distribution. During inference on a model trained exclusively with In-Distribution (ID) data, we observe a salient gradient phenomenon: around an ID sample, the local gradient directions for "enhancing" that sample's predicted class remain relatively consistent, whereas OOD samples--unseen in training--exhibit disorganized or conflicting gradient directions in the same neighborhood. Motivated by this observation, we propose an inference-stage technique to short-circuit those feature coordinates that spurious gradients exploit to inflate OOD confidence, while leaving ID classification largely intact. To circumvent the expense of recomputing the logits after this gradient short-circuit, we further introduce a local first-order approximation that accurately captures the post-modification outputs without a second forward pass. Experiments on standard OOD benchmarks show our approach yields substantial improvements. Moreover, the method is lightweight and requires minimal changes to the standard inference pipeline, offering a practical path toward robust OOD detection in real-world applications.

The ability to organically reason over and with both text and images is a pillar of human intelligence, yet the ability of Multimodal Large Language Models (MLLMs) to perform such multimodal reasoning remains under-explored. Existing benchmarks often emphasize text-dominant reasoning or rely on shallow visual cues, failing to adequately assess integrated visual and textual reasoning. We introduce EMMA (Enhanced MultiModal reAsoning), a benchmark targeting organic multimodal reasoning across mathematics, physics, chemistry, and coding. EMMA tasks demand advanced cross-modal reasoning that cannot be addressed by reasoning independently in each modality, offering an enhanced test suite for MLLMs' reasoning capabilities. Our evaluation of state-of-the-art MLLMs on EMMA reveals significant limitations in handling complex multimodal and multi-step reasoning tasks, even with advanced techniques like Chain-of-Thought prompting and test-time compute scaling underperforming. These findings underscore the need for improved multimodal architectures and training paradigms to close the gap between human and model reasoning in multimodality.

This paper addresses the challenge of graph domain adaptation on evolving, multiple out-of-distribution (OOD) graphs. Conventional graph domain adaptation methods are confined to single-step adaptation, making them ineffective in handling continuous domain shifts and prone to catastrophic forgetting. This paper introduces the Graph Continual Adaptive Learning (GCAL) method, designed to enhance model sustainability and adaptability across various graph domains. GCAL employs a bilevel optimization strategy. The "adapt" phase uses an information maximization approach to fine-tune the model with new graph domains while re-adapting past memories to mitigate forgetting. Concurrently, the "generate memory" phase, guided by a theoretical lower bound derived from information bottleneck theory, involves a variational memory graph generation module to condense original graphs into memories. Extensive experimental evaluations demonstrate that GCAL substantially outperforms existing methods in terms of adaptability and knowledge retention.

Graph Neural Networks (GNNs) have shown remarkable performance across various domains, yet they often struggle with model bias, particularly in the presence of class imbalance. This bias can lead to suboptimal performance and unfair predictions, especially for underrepresented classes. We introduce NeuBM (Neutral Bias Mitigation), a novel approach to mitigate model bias in GNNs through neutral input calibration. NeuBM leverages a dynamically updated neutral graph to estimate and correct the inherent biases of the model. By subtracting the logits obtained from the neutral graph from those of the input graph, NeuBM effectively recalibrates the model's predictions, reducing bias across different classes. Our method integrates seamlessly into existing GNN architectures and training procedures, requiring minimal computational overhead. Extensive experiments on multiple benchmark datasets demonstrate that NeuBM significantly improves the balanced accuracy and recall of minority classes, while maintaining strong overall performance. The effectiveness of NeuBM is particularly pronounced in scenarios with severe class imbalance and limited labeled data, where traditional methods often struggle. We provide theoretical insights into how NeuBM achieves bias mitigation, relating it to the concept of representation balancing. Our analysis reveals that NeuBM not only adjusts the final predictions but also influences the learning of balanced feature representations throughout the network.

The task of graph-level out-of-distribution (OOD) detection is crucial for deploying graph neural networks in real-world settings. In this paper, we observe a significant difference in the relationship between the largest and second-largest eigenvalues of the Laplacian matrix for in-distribution (ID) and OOD graph samples: OOD samples often exhibit anomalous spectral gaps (the difference between the largest and second-largest eigenvalues). This observation motivates us to propose SpecGap, an effective post-hoc approach for OOD detection on graphs. SpecGap adjusts features by subtracting the component associated with the second-largest eigenvalue, scaled by the spectral gap, from the high-level features (i.e., X - (λn - λn-1) u_n-1 v_n-1^T). SpecGap achieves state-of-the-art performance across multiple benchmark datasets. We present extensive ablation studies and comprehensive theoretical analyses to support our empirical results. As a parameter-free post-hoc method, SpecGap can be easily integrated into existing graph neural network models without requiring any additional training or model modification.

Recently, Transformer has demonstrated its comparable performance in several vision tasks and multimedia domains. However, the key self-attention in Transformer computes global attention to the image primarily in the spatial dimension, which is a non-local operation. Such spatial attention lacks the ability to model the relationship among local regions of an image and the learned representations are biased with redundant channel information perturbation. To address this problem, we propose a new Multi-scale Weight-residual Transformer (MWT) for uni-and multi-modal. Specifically, we generate local and regional tokens by different convolutions and use them as query and key-value, respectively. In this way, the computed self-attention ensures both global information interaction and focuses attention on regional information, which is more relevant to the local information. It not only retains the global attention of transformers but also obtains the ability of local attention as CNNs. Moreover, we introduce a weight-residual network for channel dimension to alleviate the feature weakening in deeper layers of the network, which can improve the sensitivity of the model for key channel representations. These can build a more abstract high-level feature representation. Extensive experiments demonstrate the effectiveness of MWT on several uni- and multimodal benchmark tasks.

Transformers have been successfully applied to long-term time series forecasting (LTSF) owing to their ability to model long-term dependencies of time series by the multi-head attention mechanism. However, most existing Transformer-based forecasting models capture temporal dependency patterns, while ignoring the frequency patterns. To balance both patterns, we explore a novel approach of applying both frequency filtering and attention mechanism within the Transformer for the LTSF task. We design the frequency filtering layer that represents time series in terms of their frequency components to capture frequency features with log-linear complexity, providing deeper insights into global dependencies of the data. Then, we propose FAformer, a simple yet effective architecture built upon frequency and attention in Transformers for LTSF. In FAformer, the frequency filtering layer captures global dependencies in time series, and then the deeper attention layer further models the features. Extensive experiments demonstrate the effectiveness of our proposed method, which outperforms state-of-the-art (SOTA) baselines on nine real-world datasets. The code will be publicly available.

Out-of-distribution (OOD) detection is crucial for safe ML deployment, yet neural networks often exhibit overconfidence on unseen data. While activation norms provide useful OOD signals, they remain vulnerable---OOD inputs can artificially inflate norms through sparse, high-magnitude activations, while valid in-distribution samples with moderate norms may be misclassified. We propose that activation distributional shape, not just magnitude, is essential for robust detection. Our method, Activation Norm and Entropy Weighting (ANEW), combines L2-norm (strength) with Shannon entropy (spread) to distinguish between genuine in-distribution patterns and OOD samples, including adversarial examples mimicking high norms via low-entropy spikes. ANEW requires only a single forward pass without retraining, making it highly practical. Extensive experiments across diverse architectures and benchmarks show ANEW significantly outperforms norm-only baselines, reducing both false positives and false negatives in challenging scenarios. Code available upon acceptance.

Graph Neural Networks (GNNs) are increasingly deployed in mission-critical tasks, yet they often encounter inputs that lie outside their training distribution, leading to unreliable or overconfident predictions. To address this limitation, we present RAGNOR (Robust Aggregation Graph Norm for Outlier Recognition), a post-hoc approach that leverages embedding norms for robust out-of-distribution (OOD) detection on both node-level and graph-level tasks. Unlike previous methods designed primarily for image domains, RAGNOR directly tackles the relational challenges intrinsic to graphs: local contamination by anomalous neighbors, disparate norm scales across classes or roles, and insufficient references for boundary or low-degree nodes. By combining global Z-score normalization, median-based local aggregation, and multi-hop blending, RAGNOR effectively refines raw norm signals into robust OOD scores while incurring minimal overhead and requiring no retraining of the original GNN. Experimental evaluations on multiple benchmarks demonstrate that RAGNOR not only achieves competitive or superior detection performance compared to alternative techniques, but also provides an intuitive, modular design that can be readily integrated into existing graph pipelines.

Global Covariance Pooling (GCP) has garnered increasing attention in visual recognition tasks, where second-order statistics frequently yield stronger representations than first-order approaches. However, two main streams of GCP---Newton--Schulz-based iSQRT-COV and exact or near-exact SVD methods---struggle at opposite ends of the training spectrum. While iSQRT-COV stabilizes early learning by avoiding large gradient explosions, it over-compresses significant eigenvalues in later stages, causing an \emph{over-flattening} phenomenon that stalls final accuracy. In contrast, SVD-based methods excel at preserving the high-eigenvalue structure essential for deep networks but suffer from sensitivity to small eigenvalue gaps early on. We propose \textbf{Halley-SVD}, a high-order iterative method that unites the smooth gradient advantages of iSQRT-COV with the late-stage fidelity of SVD. Grounded in Halley's iteration, our approach obviates explicit divisions by $(\lambda_i - \lambda_j)$ and forgoes threshold- or polynomial-based heuristics. As a result, it prevents both early gradient explosions and the excessive compression of large eigenvalues. Extensive experiments on CNNs and transformer architectures show that Halley-SVD consistently and robustly outperforms iSQRT-COV at large model scales and batch sizes, achieving higher overall accuracy without mid-training switches or custom truncations. This work provides a new solution to the long-standing dichotomy in GCP, illustrating how high-order methods can balance robustness and spectral precision to fully harness the representational power of modern deep networks.

Pseudo-label-based Semi-Supervised Learning (SSL) often suffers from classifier bias, particularly under class imbalance, as inaccurate pseudo-labels tend to exacerbate existing biases towards majority classes. Existing methods, such as \textit{CDMAD}\cite{cdmad}, utilize simplistic reference inputs—typically uniform or blank-colored images—to estimate and correct this bias. However, such simplistic references fundamentally ignore realistic statistical information inherent to real datasets, specifically typical color distributions, texture details, and frequency characteristics. This lack of \emph{statistical representativeness} can lead the model to inaccurately estimate its inherent bias, limiting the effectiveness of bias correction, particularly under severe class imbalance or substantial distribution mismatches between labeled and unlabeled datasets. To overcome these limitations, we introduce the \textbf{FARAD} (Fourier-Adapted Reference for Accurate Debiasing) System. This system utilizes random-phase images, constructed by preserving the amplitude spectrum of real data while randomizing the phase spectrum. This strategy ensures two critical properties: (1) \textbf{Semantic Irrelevance}, as randomizing phase removes any structural or recognizable semantic cues, and (2) \textbf{Statistical Representativeness}, as preserving the amplitude spectrum maintains realistic textures, color distributions, and frequency characteristics. Grounded theoretically in classical Fourier analysis, the FARAD System provides a robust, accurate estimation of per-class biases. Furthermore, computational efficiency is enhanced through optimized real-to-complex (R2C) batched Fast Fourier Transforms (FFTs). Comprehensive experiments demonstrate that our approach, significantly improves minority-class accuracy and overall SSL performance, particularly under challenging imbalance scenarios, compared with existing reference-based bias correction methods.

In intelligent fault diagnosis for construction machinery, robust and precise detection of out-of-distribution (OOD) data is crucial for enhancing operational efficiency and reducing downtime. This article introduces the energy-driven graph neural OOD (EGN-OOD) detector, a novel framework designed to address the complexities of OOD data in dynamic Internet of Things (IoT) environments. By integrating graph neural networks with energy-based models, our approach captures intricate fault correlations and improves the accuracy of fault diagnosis. The EGN-OOD framework uses the maximal information coefficient to transform sensor-acquired vibration data, typical in IoT applications, into graph representations. This conversion produces an adjacency matrix that outlines the nonlinear interactions among different fault types. Additionally, the framework includes an energy score-based OOD detection module that redefines classifier logits to create an energy function, enabling precise differentiation between in-distribution (ID) and OOD data. To enhance model robustness in semi-supervised settings, a propagation mechanism-based energy score update scheme is implemented, iteratively refining energy values within the graph. Empirical validation on a framework for monitoring mechanical equipment bearing wear demonstrates the EGN-OOD framework’s exceptional ability to detect and diagnose various fault conditions. This validation confirms the framework’s robust generalization capabilities and precision in fault detection and underscores its integration within IoT infrastructures, facilitating smarter diagnostic processes. The results provide substantial technical support for the intelligent diagnosis of construction machinery, advancing IoT-driven solutions for sustainable and intelligent construction practices.

Large Language Models (LLMs) excel in diverse tasks but often underperform in specialized fields due to limited domain-specific or proprietary corpus.Continual pre-training (CPT) enhances LLM capabilities by imbuing new domain-specific or proprietary knowledge while replaying general corpus to prevent catastrophic forgetting.The data mixture ratio of general corpus and domain-specific corpus, however, has been chosen heuristically, leading to sub-optimal training efficiency in practice.In this context, we attempt to re-visit the scaling behavior of LLMs under the hood of CPT, and discover a power-law relationship between loss, mixture ratio, and training tokens scale.We formalize the trade-off between general and domain-specific capabilities, leading to a well-defined Critical Mixture Ratio (CMR) of general and domain data.By striking the balance, CMR maintains the model's general ability and achieves the desired domain transfer, ensuring the highest utilization of available resources.Considering the balance between efficiency and effectiveness, CMR can be regarded as the optimal mixture ratio.Through extensive experiments, we ascertain the predictability of CMR, propose CMR scaling law and have substantiated its generalization.These findings offer practical guidelines for optimizing LLM training in specialized domains, ensuring both general and domain-specific performance while efficiently managing training resources.

While deep learning has revolutionized computer-aided drug discovery, the AI community has predominantly focused on model innovation and placed less emphasis on establishing best benchmarking practices. We posit that without a sound model evaluation framework, the AI community's efforts cannot reach their full potential, thereby slowing the progress and transfer of innovation into real-world drug discovery.Thus, in this paper, we seek to establish a new gold standard for small molecule drug discovery benchmarking, WelQrate. Specifically, our contributions are threefold: WelQrate dataset collection - we introduce a meticulously curated collection of 9 datasets spanning 5 therapeutic target classes. Our hierarchical curation pipelines, designed by drug discovery experts, go beyond the primary high-throughput screen by leveraging additional confirmatory and counter screens along with rigorous domain-driven preprocessing, such as Pan-Assay Interference Compounds (PAINS) filtering, to ensure the high-quality data in the datasets; WelQrate Evaluation Framework - we propose a standardized model evaluation framework considering high-quality datasets, featurization, 3D conformation generation, evaluation metrics, and data splits, which provides a reliable benchmarking for drug discovery experts conducting real-world virtual screening; Benchmarking - we evaluate model performance through various research questions using the WelQrate dataset collection, exploring the effects of different models, dataset quality, featurization methods, and data splitting strategies on the results.In summary, we recommend adopting our proposed WelQrate as the gold standard in small molecule drug discovery benchmarking. The WelQrate dataset collection, along with the curation codes, and experimental scripts are all publicly available at www.WelQrate.org.

Software products have brought convenience to modern society but also pose significant security risks due to various types of vulnerabilities. Identifying vulnerability names is vital for program repair and software maintenance, but the lack of training data presents a challenge. Big data analytics and machine learning can help overcome this challenge by processing large amounts of data and improving the accuracy of vulnerability name prediction. Considering that the data is often from datasets composed of multiple sources, a feature-based or attention-based multi-source domain adaptation (MSDA) approach is required. In this paper, we propose an MSDA method based on both feature and attention to accomplish the task of predicting vulnerability names, called Multi-Source Domain Adaptation for Vulnerability Name Prediction (MSDA-VNP). First, MSDA-VNP reduces domain divergence by adversarial training and then uses domain-invariant features to obtain feature correlations between individual source and target domains. In combination with the obtained domain correlations, Weighted multi-kernel Maximum Mean Discrepancy (WMK-MMD) is proposed as the attention mechanism. Second, a data augmentation strategy is employed to enhance MSDA-VNP to identify privacy-related vulnerabilities. To evaluate our approach, we conducted experiments on eight Java real-world projects in the Software Assurance Reference Dataset (SARD). The experimental results show that the proposed method MSDA-VNP performed efficiently and stably for the 44 types of vulnerabilities involved. The data augmentation strategy has also been proved to be effective as an enhancement for the proposed method MSDA-VNP.

Protein secondary structure prediction is still a challenging task in bioinformatics, especially for 8-state (Q8) classification. To address this problem, we have proposed a deep learning based model by integrating graph convolutional network(GCN) and bidirectional long short-term memory (Bi-LSTM) network in this paper. In the model, GCN is utilized to synthesize the information of amino acids and their interactions, while Bi-LSTM has strong ability to capture the long-range dependencies of amino acids. For sequence representation, a new protein embedding derived by ProtTrans is used instead of the traditional amino acid one-hot encoding, together with evolutionary features of PSSM and HHM profiles. Amino acid contact potential derived from SPOTContact-Helical is used to construct amino acid graph. To verify the effectiveness of our proposed model, it is applied to several benchmark datasets, and obtained 78.05%, 76.81% 72.84%, 74.46% and 76.04% Q8 accuracy on CASP10, CASP11, CASP12, CB513 and TS115 datasets, respectively. Compared with 8 state-of-the-art competitions, our model obtained the best performance in most of datasets.

High Efficiency Video Coding (HEVC) standard has a superior video compression rate compared with previous H.264/AVC. At the same time, Ultra-high-definition (UHD) video applications are becoming a reality under the development of the display technology. In this paper, a hardware-friendly multi-layer HEVC motion estimation (ME) algorithm for UHD applications are proposed. To keep the computational regularity of the traditional full-search (FS) ME algorithm as well as reduce the computational complexity of ME in a large search range (SR), the basic layer of the proposed algorithm is to combine FS scheme in a core area with downsampling search scheme in a large peripheral area. Moreover, the finer layer of the algorithm employs a hexagon search scheme to perform further ME around the optimal match point generated by the basic layer. Integrating the proposed algorithm into the HM 15.0, experimental results show that our hardware-friendly algorithm can achieve 97.8% of computations reduction while only 0.77% of BD-rate loss on average. Consequently, the proposed algorithm is feasible for HEVC ME hardware design for UHD applications.

We are proposing a new system to enhance the tactile experience of digital painting hat includes multi-strokes for different painting needs. In this paper, we describe how FlexStroke is used as a Chinese brush, an oil brush, and a crayon by changing the jamming tip. This tip has different levels of stiffness based on its jamming structure. Visual simulations on PixelSense[3] jointly enhance the intuitive painting process with realistic display results.

This paper presents RoCuModel, an iterative tangible modeling system that helps users build 3D models in a tangible way for personal fabrication. It consists mainly of a special tangible curve and an infrared camera. Users can create 3D objects by creating sketchy low-fidelity shapes with the hand. By rotating the curve along a fixed axis, users can visualize the volumetric model in a 3D space in real time. RoCuModel provides a new way for people to design and create a rotationally symmetric 3D model. This is our first step towards eliminating the gap between specialists and non-specialist users in personal fabrication.