SailCompass: Towards Reproducible and Robust Evaluation for Southeast Asian Languages

In this paper, we introduce SailCompass, a reproducible and robust evaluation benchmark for assessing Large Language Models (LLMs) on Southeast Asian Languages (SEA). SailCompass encompasses three main SEA languages, eight primary tasks including 14 datasets covering three task types (generation, multiple-choice questions, and classification). To improve the robustness of the evaluation approach, we explore different prompt configurations for multiple-choice questions and leverage calibrations to improve the faithfulness of classification tasks. With SailCompass, we derive the following findings: (1) SEA-specialized LLMs still outperform general LLMs, although the gap has narrowed; (2) A balanced language distribution is important for developing better SEA-specialized LLMs; (3) Advanced prompting techniques (e.g., calibration, perplexity-based ranking) are necessary to better utilize LLMs. All datasets and evaluation scripts are public.

Enhancing Brain Age Estimation with a Multimodal 3D CNN Approach Combining Structural MRI and AI-Synthesized Cerebral Blood Volume Data

The growing global aging population necessitates enhanced methods for assessing brain aging and related neurodegenerative changes. Brain Age Gap Estimation (BrainAGE) offers a neuroimaging biomarker for understanding these changes by predicting brain age from MRI scans. Current approaches primarily use T1-weighted magnetic resonance imaging (T1w MRI) data, capturing only structural brain information. To address the lack of functional data, we integrated AI-generated Cerebral Blood Volume (AICBV) with T1w MRI, combining both structural and functional metrics. We developed a deep learning model using a VGG-based architecture to predict brain age. Our model achieved a mean absolute error (MAE) of 3.95 years and a correlation of \(R^2 = 0.94\) on the test set (\(n = 288\)), outperforming existing models trained on similar data. We have further created gradient-based class activation maps (Grad-CAM) to visualize the regions of the brain that most influenced the model's predictions, providing interpretable insights into the structural and functional contributors to brain aging.

Deep Learning for Beamforming in Multi-User Continuous Aperture Array (CAPA) Systems

A DeepCAPA (Deep Learning for Continuous Aperture Array (CAPA)) framework is proposed to learn beamforming in CAPA systems. The beamforming optimization problem is firstly formulated, and it is mathematically proved that the optimal beamforming lies in the subspace spanned by users' conjugate channel responses. Two challenges are encountered when directly applying deep neural networks (DNNs) for solving the formulated problem, i) both the input and output spaces are infinite-dimensional, which are not compatible with DNNs. The finite-dimensional representations of inputs and outputs are derived to address this challenge. ii) A closed-form loss function is unavailable for training the DNN. To tackle this challenge, two additional DNNs are trained to approximate the operations without closed-form expressions for expediting gradient back-propagation. To improve learning performance and reduce training complexity, the permutation equivariance properties of the mappings to be learned are mathematically proved. As a further advance, the DNNs are designed as graph neural networks to leverage the properties. Numerical results demonstrate that: i) the proposed DeepCAPA framework achieves higher spectral efficiency and lower inference complexity compared to match-filtering and state-of-art Fourier-based discretization method, and ii) DeepCAPA approaches the performance upper bound of optimizing beamforming in the spatially discrete array-based system as the number of antennas in a fixed-sized area tends toward infinity.

Advancing Efficient Brain Tumor Multi-Class Classification -- New Insights from the Vision Mamba Model in Transfer Learning

Early and accurate diagnosis of brain tumors is crucial for improving patient survival rates. However, the detection and classification of brain tumors are challenging due to their diverse types and complex morphological characteristics. This study investigates the application of pre-trained models for brain tumor classification, with a particular focus on deploying the Mamba model. We fine-tuned several mainstream transfer learning models and applied them to the multi-class classification of brain tumors. By comparing these models to those trained from scratch, we demonstrated the significant advantages of transfer learning, especially in the medical imaging field, where annotated data is often limited. Notably, we introduced the Vision Mamba (Vim), a novel network architecture, and applied it for the first time in brain tumor classification, achieving exceptional classification accuracy. Experimental results indicate that the Vim model achieved 100% classification accuracy on an independent test set, emphasizing its potential for tumor classification tasks. These findings underscore the effectiveness of transfer learning in brain tumor classification and reveal that, compared to existing state-of-the-art models, the Vim model is lightweight, efficient, and highly accurate, offering a new perspective for clinical applications. Furthermore, the framework proposed in this study for brain tumor classification, based on transfer learning and the Vision Mamba model, is broadly applicable to other medical imaging classification problems.

Q-MRS: A Deep Learning Framework for Quantitative Magnetic Resonance Spectra Analysis

Magnetic resonance spectroscopy (MRS) is an established technique for studying tissue metabolism, particularly in central nervous system disorders. While powerful and versatile, MRS is often limited by challenges associated with data quality, processing, and quantification. Existing MRS quantification methods face difficulties in balancing model complexity and reproducibility during spectral modeling, often falling into the trap of either oversimplification or over-parameterization. To address these limitations, this study introduces a deep learning (DL) framework that employs transfer learning, in which the model is pre-trained on simulated datasets before it undergoes fine-tuning on in vivo data. The proposed framework showed promising performance when applied to the Philips dataset from the BIG GABA repository and represents an exciting advancement in MRS data analysis.

ViLReF: A Chinese Vision-Language Retinal Foundation Model

Subtle semantic differences in retinal image and text data present great challenges for pre-training visual-language models. Moreover, false negative samples, i.e., image-text pairs having the same semantics but incorrectly regarded as negatives, disrupt the visual-language pre-training process and affect the model's learning ability. This work aims to develop a retinal foundation model, called ViLReF, by pre-training on a paired dataset comprising 451,956 retinal images and corresponding diagnostic text reports. In our vision-language pre-training strategy, we leverage expert knowledge to facilitate the extraction of labels and propose a novel constraint, the Weighted Similarity Coupling Loss, to adjust the speed of pushing sample pairs further apart dynamically within the feature space. Furthermore, we employ a batch expansion module with dynamic memory queues, maintained by momentum encoders, to supply extra samples and compensate for the vacancies caused by eliminating false negatives. Extensive experiments are conducted on multiple datasets for downstream classification and segmentation tasks. The experimental results demonstrate the powerful zero-shot and transfer learning capabilities of ViLReF, verifying the effectiveness of our pre-training strategy. Our ViLReF model is available at: https://github.com/T6Yang/ViLReF.

Multi-User Continuous-Aperture Array Communications: How to Learn Current Distribution?

The continuous aperture array (CAPA) can provide higher degree-of-freedom and spatial resolution than the spatially discrete array (SDPA), where optimizing multi-user current distributions in CAPA systems is crucial but challenging. The challenge arises from solving non-convex functional optimization problems without closed-form objective functions and constraints. In this paper, we propose a deep learning framework called L-CAPA to learn current distribution policies. In the framework, we find finite-dimensional representations of channel functions and current distributions, allowing them to be inputted into and outputted from a deep neural network (DNN) for learning the policy. To address the issue that the integrals in the loss function without closed-form expressions hinder training the DNN in an unsupervised manner, we propose to design another two DNNs for learning the integrals. The DNNs are designed as graph neural networks to incorporate with the permutation properties of the mappings to be learned, thereby improving learning performance. Simulation results show that L-CAPA can achieve the performance upper-bound of optimizing precoding in the SDPA system as the number of antennas approaches infinity, and it is with low inference complexity.

High-Fidelity Facial Albedo Estimation via Texture Quantization

Recent 3D face reconstruction methods have made significant progress in shape estimation, but high-fidelity facial albedo reconstruction remains challenging. Existing methods depend on expensive light-stage captured data to learn facial albedo maps. However, a lack of diversity in subjects limits their ability to recover high-fidelity results. In this paper, we present a novel facial albedo reconstruction model, HiFiAlbedo, which recovers the albedo map directly from a single image without the need for captured albedo data. Our key insight is that the albedo map is the illumination invariant texture map, which enables us to use inexpensive texture data to derive an albedo estimation by eliminating illumination. To achieve this, we first collect large-scale ultra-high-resolution facial images and train a high-fidelity facial texture codebook. By using the FFHQ dataset and limited UV textures, we then fine-tune the encoder for texture reconstruction from the input image with adversarial supervision in both image and UV space. Finally, we train a cross-attention module and utilize group identity loss to learn the adaptation from facial texture to the albedo domain. Extensive experimentation has demonstrated that our method exhibits excellent generalizability and is capable of achieving high-fidelity results for in-the-wild facial albedo recovery. Our code, pre-trained weights, and training data will be made publicly available at https://hifialbedo.github.io/.

RET-CLIP: A Retinal Image Foundation Model Pre-trained with Clinical Diagnostic Reports

The Vision-Language Foundation model is increasingly investigated in the fields of computer vision and natural language processing, yet its exploration in ophthalmology and broader medical applications remains limited. The challenge is the lack of labeled data for the training of foundation model. To handle this issue, a CLIP-style retinal image foundation model is developed in this paper. Our foundation model, RET-CLIP, is specifically trained on a dataset of 193,865 patients to extract general features of color fundus photographs (CFPs), employing a tripartite optimization strategy to focus on left eye, right eye, and patient level to reflect real-world clinical scenarios. Extensive experiments demonstrate that RET-CLIP outperforms existing benchmarks across eight diverse datasets spanning four critical diagnostic categories: diabetic retinopathy, glaucoma, multiple disease diagnosis, and multi-label classification of multiple diseases, which demonstrate the performance and generality of our foundation model. The sourse code and pre-trained model are available at https://github.com/sStonemason/RET-CLIP.

Dinomaly: The Less Is More Philosophy in Multi-Class Unsupervised Anomaly Detection

Recent studies highlighted a practical setting of unsupervised anomaly detection (UAD) that builds a unified model for multi-class images, serving as an alternative to the conventional one-class-one-model setup. Despite various advancements addressing this challenging task, the detection performance under the multi-class setting still lags far behind state-of-the-art class-separated models. Our research aims to bridge this substantial performance gap. In this paper, we introduce a minimalistic reconstruction-based anomaly detection framework, namely Dinomaly, which leverages pure Transformer architectures without relying on complex designs, additional modules, or specialized tricks. Given this powerful framework consisted of only Attentions and MLPs, we found four simple components that are essential to multi-class anomaly detection: (1) Foundation Transformers that extracts universal and discriminative features, (2) Noisy Bottleneck where pre-existing Dropouts do all the noise injection tricks, (3) Linear Attention that naturally cannot focus, and (4) Loose Reconstruction that does not force layer-to-layer and point-by-point reconstruction. Extensive experiments are conducted across three popular anomaly detection benchmarks including MVTec-AD, VisA, and the recently released Real-IAD. Our proposed Dinomaly achieves impressive image AUROC of 99.6%, 98.7%, and 89.3% on the three datasets respectively, which is not only superior to state-of-the-art multi-class UAD methods, but also surpasses the most advanced class-separated UAD records.