Memory Reviving, Continuing Learning and Beyond: Evaluation of Pre-trained Encoders and Decoders for Multimodal Machine Translation

Multimodal Machine Translation (MMT) aims to improve translation quality by leveraging auxiliary modalities such as images alongside textual input. While recent advances in large-scale pre-trained language and vision models have significantly benefited unimodal natural language processing tasks, their effectiveness and role in MMT remain underexplored. In this work, we conduct a systematic study on the impact of pre-trained encoders and decoders in multimodal translation models. Specifically, we analyze how different training strategies, from training from scratch to using pre-trained and partially frozen components, affect translation performance under a unified MMT framework. Experiments are carried out on the Multi30K and CoMMuTE dataset across English-German and English-French translation tasks. Our results reveal that pre-training plays a crucial yet asymmetrical role in multimodal settings: pre-trained decoders consistently yield more fluent and accurate outputs, while pre-trained encoders show varied effects depending on the quality of visual-text alignment. Furthermore, we provide insights into the interplay between modality fusion and pre-trained components, offering guidance for future architecture design in multimodal translation systems.

Decoupling Multi-Contrast Super-Resolution: Pairing Unpaired Synthesis with Implicit Representations

Magnetic Resonance Imaging (MRI) is critical for clinical diagnostics but is often limited by long acquisition times and low signal-to-noise ratios, especially in modalities like diffusion and functional MRI. The multi-contrast nature of MRI presents a valuable opportunity for cross-modal enhancement, where high-resolution (HR) modalities can serve as references to boost the quality of their low-resolution (LR) counterparts-motivating the development of Multi-Contrast Super-Resolution (MCSR) techniques. Prior work has shown that leveraging complementary contrasts can improve SR performance; however, effective feature extraction and fusion across modalities with varying resolutions remains a major challenge. Moreover, existing MCSR methods often assume fixed resolution settings and all require large, perfectly paired training datasets-conditions rarely met in real-world clinical environments. To address these challenges, we propose a novel Modular Multi-Contrast Super-Resolution (MCSR) framework that eliminates the need for paired training data and supports arbitrary upscaling. Our method decouples the MCSR task into two stages: (1) Unpaired Cross-Modal Synthesis (U-CMS), which translates a high-resolution reference modality into a synthesized version of the target contrast, and (2) Unsupervised Super-Resolution (U-SR), which reconstructs the final output using implicit neural representations (INRs) conditioned on spatial coordinates. This design enables scale-agnostic and anatomically faithful reconstruction by bridging un-paired cross-modal synthesis with unsupervised resolution enhancement. Experiments show that our method achieves superior performance at 4x and 8x upscaling, with improved fidelity and anatomical consistency over existing baselines. Our framework demonstrates strong potential for scalable, subject-specific, and data-efficient MCSR in real-world clinical settings.

SAR-to-RGB Translation with Latent Diffusion for Earth Observation

Earth observation satellites like Sentinel-1 (S1) and Sentinel-2 (S2) provide complementary remote sensing (RS) data, but S2 images are often unavailable due to cloud cover or data gaps. To address this, we propose a diffusion model (DM)-based approach for SAR-to-RGB translation, generating synthetic optical images from SAR inputs. We explore three different setups: two using Standard Diffusion, which reconstruct S2 images by adding and removing noise (one without and one with class conditioning), and one using Cold Diffusion, which blends S2 with S1 before removing the SAR signal. We evaluate the generated images in downstream tasks, including land cover classification and cloud removal. While generated images may not perfectly replicate real S2 data, they still provide valuable information. Our results show that class conditioning improves classification accuracy, while cloud removal performance remains competitive despite our approach not being optimized for it. Interestingly, despite exhibiting lower perceptual quality, the Cold Diffusion setup performs well in land cover classification, suggesting that traditional quantitative evaluation metrics may not fully reflect the practical utility of generated images. Our findings highlight the potential of DMs for SAR-to-RGB translation in RS applications where RGB images are missing.

RASMD: RGB And SWIR Multispectral Driving Dataset for Robust Perception in Adverse Conditions

Current autonomous driving algorithms heavily rely on the visible spectrum, which is prone to performance degradation in adverse conditions like fog, rain, snow, glare, and high contrast. Although other spectral bands like near-infrared (NIR) and long-wave infrared (LWIR) can enhance vision perception in such situations, they have limitations and lack large-scale datasets and benchmarks. Short-wave infrared (SWIR) imaging offers several advantages over NIR and LWIR. However, no publicly available large-scale datasets currently incorporate SWIR data for autonomous driving. To address this gap, we introduce the RGB and SWIR Multispectral Driving (RASMD) dataset, which comprises 100,000 synchronized and spatially aligned RGB-SWIR image pairs collected across diverse locations, lighting, and weather conditions. In addition, we provide a subset for RGB-SWIR translation and object detection annotations for a subset of challenging traffic scenarios to demonstrate the utility of SWIR imaging through experiments on both object detection and RGB-to-SWIR image translation. Our experiments show that combining RGB and SWIR data in an ensemble framework significantly improves detection accuracy compared to RGB-only approaches, particularly in conditions where visible-spectrum sensors struggle. We anticipate that the RASMD dataset will advance research in multispectral imaging for autonomous driving and robust perception systems.

Deterministic Medical Image Translation via High-fidelity Brownian Bridges

Recent studies have shown that diffusion models produce superior synthetic images when compared to Generative Adversarial Networks (GANs). However, their outputs are often non-deterministic and lack high fidelity to the ground truth due to the inherent randomness. In this paper, we propose a novel High-fidelity Brownian bridge model (HiFi-BBrg) for deterministic medical image translations. Our model comprises two distinct yet mutually beneficial mappings: a generation mapping and a reconstruction mapping. The Brownian bridge training process is guided by the fidelity loss and adversarial training in the reconstruction mapping. This ensures that translated images can be accurately reversed to their original forms, thereby achieving consistent translations with high fidelity to the ground truth. Our extensive experiments on multiple datasets show HiFi-BBrg outperforms state-of-the-art methods in multi-modal image translation and multi-image super-resolution.

Synthetic CT Generation from Time-of-Flight Non-Attenutaion-Corrected PET for Whole-Body PET Attenuation Correction

Positron Emission Tomography (PET) imaging requires accurate attenuation correction (AC) to account for photon loss due to tissue density variations. In PET/MR systems, computed tomography (CT), which offers a straightforward estimation of AC is not available. This study presents a deep learning approach to generate synthetic CT (sCT) images directly from Time-of-Flight (TOF) non-attenuation corrected (NAC) PET images, enhancing AC for PET/MR. We first evaluated models pre-trained on large-scale natural image datasets for a CT-to-CT reconstruction task, finding that the pre-trained model outperformed those trained solely on medical datasets. The pre-trained model was then fine-tuned using an institutional dataset of 35 TOF NAC PET and CT volume pairs, achieving the lowest mean absolute error (MAE) of 74.49 HU and highest peak signal-to-noise ratio (PSNR) of 28.66 dB within the body contour region. Visual assessments demonstrated improved reconstruction of both bone and soft tissue structures from TOF NAC PET images. This work highlights the effectiveness of using pre-trained deep learning models for medical image translation tasks. Future work will assess the impact of sCT on PET attenuation correction and explore additional neural network architectures and datasets to further enhance performance and practical applications in PET imaging.

The Structural Safety Generalization Problem

LLM jailbreaks are a widespread safety challenge. Given this problem has not yet been tractable, we suggest targeting a key failure mechanism: the failure of safety to generalize across semantically equivalent inputs. We further focus the target by requiring desirable tractability properties of attacks to study: explainability, transferability between models, and transferability between goals. We perform red-teaming within this framework by uncovering new vulnerabilities to multi-turn, multi-image, and translation-based attacks. These attacks are semantically equivalent by our design to their single-turn, single-image, or untranslated counterparts, enabling systematic comparisons; we show that the different structures yield different safety outcomes. We then demonstrate the potential for this framework to enable new defenses by proposing a Structure Rewriting Guardrail, which converts an input to a structure more conducive to safety assessment. This guardrail significantly improves refusal of harmful inputs, without over-refusing benign ones. Thus, by framing this intermediate challenge - more tractable than universal defenses but essential for long-term safety - we highlight a critical milestone for AI safety research.

Leveraging Generalizability of Image-to-Image Translation for Enhanced Adversarial Defense

In the rapidly evolving field of artificial intelligence, machine learning emerges as a key technology characterized by its vast potential and inherent risks. The stability and reliability of these models are important, as they are frequent targets of security threats. Adversarial attacks, first rigorously defined by Ian Goodfellow et al. in 2013, highlight a critical vulnerability: they can trick machine learning models into making incorrect predictions by applying nearly invisible perturbations to images. Although many studies have focused on constructing sophisticated defensive mechanisms to mitigate such attacks, they often overlook the substantial time and computational costs of training and maintaining these models. Ideally, a defense method should be able to generalize across various, even unseen, adversarial attacks with minimal overhead. Building on our previous work on image-to-image translation-based defenses, this study introduces an improved model that incorporates residual blocks to enhance generalizability. The proposed method requires training only a single model, effectively defends against diverse attack types, and is well-transferable between different target models. Experiments show that our model can restore the classification accuracy from near zero to an average of 72\% while maintaining competitive performance compared to state-of-the-art methods.

Rotational ultrasound and photoacoustic tomography of the human body

Imaging the human body's morphological and angiographic information is essential for diagnosing, monitoring, and treating medical conditions. Ultrasonography performs the morphological assessment of the soft tissue based on acoustic impedance variations, whereas photoacoustic tomography (PAT) can visualize blood vessels based on intrinsic hemoglobin absorption. Three-dimensional (3D) panoramic imaging of the vasculature is generally not practical in conventional ultrasonography with limited field-of-view (FOV) probes, and PAT does not provide sufficient scattering-based soft tissue morphological contrast. Complementing each other, fast panoramic rotational ultrasound tomography (RUST) and PAT are integrated for hybrid rotational ultrasound and photoacoustic tomography (RUS-PAT), which obtains 3D ultrasound structural and PAT angiographic images of the human body quasi-simultaneously. The RUST functionality is achieved in a cost-effective manner using a single-element ultrasonic transducer for ultrasound transmission and rotating arc-shaped arrays for 3D panoramic detection. RUST is superior to conventional ultrasonography, which either has a limited FOV with a linear array or is high-cost with a hemispherical array that requires both transmission and receiving. By switching the acoustic source to a light source, the system is conveniently converted to PAT mode to acquire angiographic images in the same region. Using RUS-PAT, we have successfully imaged the human head, breast, hand, and foot with a 10 cm diameter FOV, submillimeter isotropic resolution, and 10 s imaging time for each modality. The 3D RUS-PAT is a powerful tool for high-speed, 3D, dual-contrast imaging of the human body with potential for rapid clinical translation.

MuTri: Multi-view Tri-alignment for OCT to OCTA 3D Image Translation

Optical coherence tomography angiography (OCTA) shows its great importance in imaging microvascular networks by providing accurate 3D imaging of blood vessels, but it relies upon specialized sensors and expensive devices. For this reason, previous works show the potential to translate the readily available 3D Optical Coherence Tomography (OCT) images into 3D OCTA images. However, existing OCTA translation methods directly learn the mapping from the OCT domain to the OCTA domain in continuous and infinite space with guidance from only a single view, i.e., the OCTA project map, resulting in suboptimal results. To this end, we propose the multi-view Tri-alignment framework for OCT to OCTA 3D image translation in discrete and finite space, named MuTri. In the first stage, we pre-train two vector-quantized variational auto-encoder (VQ- VAE) by reconstructing 3D OCT and 3D OCTA data, providing semantic prior for subsequent multi-view guidances. In the second stage, our multi-view tri-alignment facilitates another VQVAE model to learn the mapping from the OCT domain to the OCTA domain in discrete and finite space. Specifically, a contrastive-inspired semantic alignment is proposed to maximize the mutual information with the pre-trained models from OCT and OCTA views, to facilitate codebook learning. Meanwhile, a vessel structure alignment is proposed to minimize the structure discrepancy with the pre-trained models from the OCTA project map view, benefiting from learning the detailed vessel structure information. We also collect the first large-scale dataset, namely, OCTA2024, which contains a pair of OCT and OCTA volumes from 846 subjects.