Same model, better performance: the impact of shuffling on DNA Language Models benchmarking

Large Language Models are increasingly popular in genomics due to their potential to decode complex biological sequences. Hence, researchers require a standardized benchmark to evaluate DNA Language Models (DNA LMs) capabilities. However, evaluating DNA LMs is a complex task that intersects genomic's domain-specific challenges and machine learning methodologies, where seemingly minor implementation details can significantly compromise benchmark validity. We demonstrate this through BEND (Benchmarking DNA Language Models), where hardware-dependent hyperparameters -- number of data loading workers and buffer sizes -- create spurious performance variations of up to 4% for identical models. The problem stems from inadequate data shuffling interacting with domain specific data characteristics. Experiments with three DNA language models (HyenaDNA, DNABERT-2, ResNet-LM) show these artifacts affect both absolute performance and relative model rankings. We propose a simple solution: pre-shuffling data before storage eliminates hardware dependencies while maintaining efficiency. This work highlights how standard ML practices can interact unexpectedly with domain-specific data characteristics, with broader implications for benchmark design in specialized domains.

Personality-Enhanced Multimodal Depression Detection in the Elderly

This paper presents our solution to the Multimodal Personality-aware Depression Detection (MPDD) challenge at ACM MM 2025. We propose a multimodal depression detection model in the Elderly that incorporates personality characteristics. We introduce a multi-feature fusion approach based on a co-attention mechanism to effectively integrate LLDs, MFCCs, and Wav2Vec features in the audio modality. For the video modality, we combine representations extracted from OpenFace, ResNet, and DenseNet to construct a comprehensive visual feature set. Recognizing the critical role of personality in depression detection, we design an interaction module that captures the relationships between personality traits and multimodal features. Experimental results from the MPDD Elderly Depression Detection track demonstrate that our method significantly enhances performance, providing valuable insights for future research in multimodal depression detection among elderly populations.

Do We Really Need Permutations? Impact of Width Expansion on Linear Mode Connectivity

Recently, Ainsworth et al. empirically demonstrated that, given two independently trained models, applying a parameter permutation that preserves the input-output behavior allows the two models to be connected by a low-loss linear path. When such a path exists, the models are said to achieve linear mode connectivity (LMC). Prior studies, including Ainsworth et al., have reported that achieving LMC requires not only an appropriate permutation search but also sufficiently wide models (e.g., a 32 $\times$ width multiplier for ResNet-20). This is broadly believed to be because increasing the model width ensures a large enough space of candidate permutations, increasing the chance of finding one that yields LMC. In this work, we empirically demonstrate that, even without any permutations, simply widening the models is sufficient for achieving LMC when using a suitable softmax temperature calibration. We further explain why this phenomenon arises by analyzing intermediate layer outputs. Specifically, we introduce layerwise exponentially weighted connectivity (LEWC), which states that the output of each layer of the merged model can be represented as an exponentially weighted sum of the outputs of the corresponding layers of the original models. Consequently the merged model's output matches that of an ensemble of the original models, which facilitates LMC. To the best of our knowledge, this work is the first to show that widening the model not only facilitates nonlinear mode connectivity, as suggested in prior research, but also significantly increases the possibility of achieving linear mode connectivity.

A class-driven hierarchical ResNet for classification of multispectral remote sensing images

This work presents a multitemporal class-driven hierarchical Residual Neural Network (ResNet) designed for modelling the classification of Time Series (TS) of multispectral images at different semantical class levels. The architecture consists of a modification of the ResNet where we introduce additional branches to perform the classification at the different hierarchy levels and leverage on hierarchy-penalty maps to discourage incoherent hierarchical transitions within the classification. In this way, we improve the discrimination capabilities of classes at different levels of semantic details and train a modular architecture that can be used as a backbone network for introducing new specific classes and additional tasks considering limited training samples available. We exploit the class-hierarchy labels to train efficiently the different layers of the architecture, allowing the first layers to train faster on the first levels of the hierarchy modeling general classes (i.e., the macro-classes) and the intermediate classes, while using the last ones to discriminate more specific classes (i.e., the micro-classes). In this way, the targets are constrained in following the hierarchy defined, improving the classification of classes at the most detailed level. The proposed modular network has intrinsic adaptation capability that can be obtained through fine tuning. The experimental results, obtained on two tiles of the Amazonian Forest on 12 monthly composites of Sentinel 2 images acquired during 2019, demonstrate the effectiveness of the hierarchical approach in both generalizing over different hierarchical levels and learning discriminant features for an accurate classification at the micro-class level on a new target area, with a better representation of the minoritarian classes.

Hybrid CNN-BYOL Approach for Fault Detection in Induction Motors Using Thermal Images

Induction motors (IMs) are indispensable in industrial and daily life, but they are susceptible to various faults that can lead to overheating, wasted energy consumption, and service failure. Early detection of faults is essential to protect the motor and prolong its lifespan. This paper presents a hybrid method that integrates BYOL with CNNs for classifying thermal images of induction motors for fault detection. The thermal dataset used in this work includes different operating states of the motor, such as normal operation, overload, and faults. We employed multiple deep learning (DL) models for the BYOL technique, ranging from popular architectures such as ResNet-50, DenseNet-121, DenseNet-169, EfficientNetB0, VGG16, and MobileNetV2. Additionally, we introduced a new high-performance yet lightweight CNN model named BYOL-IMNet, which comprises four custom-designed blocks tailored for fault classification in thermal images. Our experimental results demonstrate that the proposed BYOL-IMNet achieves 99.89\% test accuracy and an inference time of 5.7 ms per image, outperforming state-of-the-art models. This study highlights the promising performance of the CNN-BYOL hybrid method in enhancing accuracy for detecting faults in induction motors, offering a robust methodology for online monitoring in industrial settings.

Accelerating Inference for Multilayer Neural Networks with Quantum Computers

Fault-tolerant Quantum Processing Units (QPUs) promise to deliver exponential speed-ups in select computational tasks, yet their integration into modern deep learning pipelines remains unclear. In this work, we take a step towards bridging this gap by presenting the first fully-coherent quantum implementation of a multilayer neural network with non-linear activation functions. Our constructions mirror widely used deep learning architectures based on ResNet, and consist of residual blocks with multi-filter 2D convolutions, sigmoid activations, skip-connections, and layer normalizations. We analyse the complexity of inference for networks under three quantum data access regimes. Without any assumptions, we establish a quadratic speedup over classical methods for shallow bilinear-style networks. With efficient quantum access to the weights, we obtain a quartic speedup over classical methods. With efficient quantum access to both the inputs and the network weights, we prove that a network with an $N$-dimensional vectorized input, $k$ residual block layers, and a final residual-linear-pooling layer can be implemented with an error of $\epsilon$ with $O(\text{polylog}(N/\epsilon)^k)$ inference cost.

High-Rate Mixout: Revisiting Mixout for Robust Domain Generalization

Ensembling fine-tuned models initialized from powerful pre-trained weights is a common strategy to improve robustness under distribution shifts, but it comes with substantial computational costs due to the need to train and store multiple models. Dropout offers a lightweight alternative by simulating ensembles through random neuron deactivation; however, when applied to pre-trained models, it tends to over-regularize and disrupt critical representations necessary for generalization. In this work, we investigate Mixout, a stochastic regularization technique that provides an alternative to Dropout for domain generalization. Rather than deactivating neurons, Mixout mitigates overfitting by probabilistically swapping a subset of fine-tuned weights with their pre-trained counterparts during training, thereby maintaining a balance between adaptation and retention of prior knowledge. Our study reveals that achieving strong performance with Mixout on domain generalization benchmarks requires a notably high masking probability of 0.9 for ViTs and 0.8 for ResNets. While this may seem like a simple adjustment, it yields two key advantages for domain generalization: (1) higher masking rates more strongly penalize deviations from the pre-trained parameters, promoting better generalization to unseen domains; and (2) high-rate masking substantially reduces computational overhead, cutting gradient computation by up to 45% and gradient memory usage by up to 90%. Experiments across five domain generalization benchmarks, PACS, VLCS, OfficeHome, TerraIncognita, and DomainNet, using ResNet and ViT architectures, show that our approach, High-rate Mixout, achieves out-of-domain accuracy comparable to ensemble-based methods while significantly reducing training costs.

A Comparative Study of Vision Transformers and CNNs for Few-Shot Rigid Transformation and Fundamental Matrix Estimation

Vision-transformers (ViTs) and large-scale convolution-neural-networks (CNNs) have reshaped computer vision through pretrained feature representations that enable strong transfer learning for diverse tasks. However, their efficiency as backbone architectures for geometric estimation tasks involving image deformations in low-data regimes remains an open question. This work considers two such tasks: 1) estimating 2D rigid transformations between pairs of images and 2) predicting the fundamental matrix for stereo image pairs, an important problem in various applications, such as autonomous mobility, robotics, and 3D scene reconstruction. Addressing this intriguing question, this work systematically compares large-scale CNNs (ResNet, EfficientNet, CLIP-ResNet) with ViT-based foundation models (CLIP-ViT variants and DINO) in various data size settings, including few-shot scenarios. These pretrained models are optimized for classification or contrastive learning, encouraging them to focus mostly on high-level semantics. The considered tasks require balancing local and global features differently, challenging the straightforward adoption of these models as the backbone. Empirical comparative analysis shows that, similar to training from scratch, ViTs outperform CNNs during refinement in large downstream-data scenarios. However, in small data scenarios, the inductive bias and smaller capacity of CNNs improve their performance, allowing them to match that of a ViT. Moreover, ViTs exhibit stronger generalization in cross-domain evaluation where the data distribution changes. These results emphasize the importance of carefully selecting model architectures for refinement, motivating future research towards hybrid architectures that balance local and global representations.

Neuroplastic Modular Framework: Cross-Domain Image Classification of Garbage and Industrial Surfaces

Efficient and accurate classification of waste and industrial surface defects is essential for ensuring sustainable waste management and maintaining high standards in quality control. This paper introduces the Neuroplastic Modular Classifier, a novel hybrid architecture designed for robust and adaptive image classification in dynamic environments. The model combines a ResNet-50 backbone for localized feature extraction with a Vision Transformer (ViT) to capture global semantic context. Additionally, FAISS-based similarity retrieval is incorporated to provide a memory-like reference to previously encountered data, enriching the model's feature space. A key innovation of our architecture is the neuroplastic modular design composed of expandable, learnable blocks that dynamically grow during training when performance plateaus. Inspired by biological learning systems, this mechanism allows the model to adapt to data complexity over time, improving generalization. Beyond garbage classification, we validate the model on the Kolektor Surface Defect Dataset 2 (KolektorSDD2), which involves industrial defect detection on metal surfaces. Experimental results across domains show that the proposed architecture outperforms traditional static models in both accuracy and adaptability. The Neuroplastic Modular Classifier offers a scalable, high-performance solution for real-world image classification, with strong applicability in both environmental and industrial domains.

Arithmetic-Mean $μ$P for Modern Architectures: A Unified Learning-Rate Scale for CNNs and ResNets

Choosing an appropriate learning rate remains a key challenge in scaling depth of modern deep networks. The classical maximal update parameterization ($\mu$P) enforces a fixed per-layer update magnitude, which is well suited to homogeneous multilayer perceptrons (MLPs) but becomes ill-posed in heterogeneous architectures where residual accumulation and convolutions introduce imbalance across layers. We introduce Arithmetic-Mean $\mu$P (AM-$\mu$P), which constrains not each individual layer but the network-wide average one-step pre-activation second moment to a constant scale. Combined with a residual-aware He fan-in initialization - scaling residual-branch weights by the number of blocks ($\mathrm{Var}[W]=c/(K\cdot \mathrm{fan\text{-}in})$) - AM-$\mu$P yields width-robust depth laws that transfer consistently across depths. We prove that, for one- and two-dimensional convolutional networks, the maximal-update learning rate satisfies $\eta^\star(L)\propto L^{-3/2}$; with zero padding, boundary effects are constant-level as $N\gg k$. For standard residual networks with general conv+MLP blocks, we establish $\eta^\star(L)=\Theta(L^{-3/2})$, with $L$ the minimal depth. Empirical results across a range of depths confirm the $-3/2$ scaling law and enable zero-shot learning-rate transfer, providing a unified and practical LR principle for convolutional and deep residual networks without additional tuning overhead.