Hematoxylin and Eosin (H&E) staining is the cornerstone of histopathology but lacks molecular specificity. While Immunohistochemistry (IHC) provides molecular insights, it is costly and complex, motivating H&E-to-IHC translation as a cost-effective alternative. Existing translation methods are mainly GAN-based, often struggling with training instability and limited structural fidelity, while diffusion-based approaches remain underexplored. We propose HistDiST, a Latent Diffusion Model (LDM) based framework for high-fidelity H&E-to-IHC translation. HistDiST introduces a dual-conditioning strategy, utilizing Phikon-extracted morphological embeddings alongside VAE-encoded H&E representations to ensure pathology-relevant context and structural consistency. To overcome brightness biases, we incorporate a rescaled noise schedule, v-prediction, and trailing timesteps, enforcing a zero-SNR condition at the final timestep. During inference, DDIM inversion preserves the morphological structure, while an eta-cosine noise schedule introduces controlled stochasticity, balancing structural consistency and molecular fidelity. Moreover, we propose Molecular Retrieval Accuracy (MRA), a novel pathology-aware metric leveraging GigaPath embeddings to assess molecular relevance. Extensive evaluations on MIST and BCI datasets demonstrate that HistDiST significantly outperforms existing methods, achieving a 28% improvement in MRA on the H&E-to-Ki67 translation task, highlighting its effectiveness in capturing true IHC semantics.
Breast cancer is the most frequently diagnosed human cancer in the United States at present. Early detection is crucial for its successful treatment. X-ray mammography and digital breast tomosynthesis are currently the main methods for breast cancer screening. However, both have known limitations in terms of their sensitivity and specificity to breast cancers, while also frequently causing patient discomfort due to the requirement for breast compression. Breast computed tomography is a promising alternative, however, to obtain high-quality images, the X-ray dose needs to be sufficiently high. As the breast is highly radiosensitive, dose reduction is particularly important. Phase-contrast computed tomography (PCT) has been shown to produce higher-quality images at lower doses and has no need for breast compression. It is demonstrated in the present study that, when imaging full fresh mastectomy samples with PCT, deep learning-based image denoising can further reduce the radiation dose by a factor of 16 or more, without any loss of image quality. The image quality has been assessed both in terms of objective metrics, such as spatial resolution and contrast-to-noise ratio, as well as in an observer study by experienced medical imaging specialists and radiologists. This work was carried out in preparation for live patient PCT breast cancer imaging, initially at specialized synchrotron facilities.
The rise of Large Language Models (LLMs) has raised questions about their ability to understand climate-related contexts. Though climate change dominates social media, analyzing its multimodal expressions is understudied, and current tools have failed to determine whether LLMs amplify credible solutions or spread unsubstantiated claims. To address this, we introduce CliME (Climate Change Multimodal Evaluation), a first-of-its-kind multimodal dataset, comprising 2579 Twitter and Reddit posts. The benchmark features a diverse collection of humorous memes and skeptical posts, capturing how these formats distill complex issues into viral narratives that shape public opinion and policy discussions. To systematically evaluate LLM performance, we present the Climate Alignment Quotient (CAQ), a novel metric comprising five distinct dimensions: Articulation, Evidence, Resonance, Transition, and Specificity. Additionally, we propose three analytical lenses: Actionability, Criticality, and Justice, to guide the assessment of LLM-generated climate discourse using CAQ. Our findings, based on the CAQ metric, indicate that while most evaluated LLMs perform relatively well in Criticality and Justice, they consistently underperform on the Actionability axis. Among the models evaluated, Claude 3.7 Sonnet achieves the highest overall performance. We publicly release our CliME dataset and code to foster further research in this domain.
A key value proposition of machine learning is generalizability: the same methods and model architecture should be able to work across different domains and different contexts. While powerful, this generalization can sometimes go too far, and miss the importance of the specifics. In this work, we look at how fair machine learning has often treated as interchangeable the identity axis along which discrimination occurs. In other words, racism is measured and mitigated the same way as sexism, as ableism, as ageism. Disciplines outside of computer science have pointed out both the similarities and differences between these different forms of oppression, and in this work we draw out the implications for fair machine learning. While certainly not all aspects of fair machine learning need to be tailored to the specific form of oppression, there is a pressing need for greater attention to such specificity than is currently evident. Ultimately, context specificity can deepen our understanding of how to build more fair systems, widen our scope to include currently overlooked harms, and, almost paradoxically, also help to narrow our scope and counter the fear of an infinite number of group-specific methods of analysis.
The application of large language models (LLMs) to healthcare information extraction has emerged as a promising approach. This study evaluates the classification performance of five open-source LLMs: GEMMA-3-27B-IT, LLAMA3-70B, LLAMA4-109B, DEEPSEEK-R1-DISTILL-LLAMA-70B, and DEEPSEEK-V3-0324-UD-Q2_K_XL, across six healthcare-related classification tasks involving both social media data (breast cancer, changes in medication regimen, adverse pregnancy outcomes, potential COVID-19 cases) and clinical data (stigma labeling, medication change discussion). We report precision, recall, and F1 scores with 95% confidence intervals for all model-task combinations. Our findings reveal significant performance variability between LLMs, with DeepSeekV3 emerging as the strongest overall performer, achieving the highest F1 scores in four tasks. Notably, models generally performed better on social media tasks compared to clinical data tasks, suggesting potential domain-specific challenges. GEMMA-3-27B-IT demonstrated exceptionally high recall despite its smaller parameter count, while LLAMA4-109B showed surprisingly underwhelming performance compared to its predecessor LLAMA3-70B, indicating that larger parameter counts do not guarantee improved classification results. We observed distinct precision-recall trade-offs across models, with some favoring sensitivity over specificity and vice versa. These findings highlight the importance of task-specific model selection for healthcare applications, considering the particular data domain and precision-recall requirements rather than model size alone. As healthcare increasingly integrates AI-driven text classification tools, this comprehensive benchmarking provides valuable guidance for model selection and implementation while underscoring the need for continued evaluation and domain adaptation of LLMs in healthcare contexts.
A three-dimensional convolutional neural network was developed to classify T1-weighted brain MRI scans as healthy or Alzheimer. The network comprises 3D convolution, pooling, batch normalization, dense ReLU layers, and a sigmoid output. Using stochastic noise injection and five-fold cross-validation, the model achieved test set accuracy of 0.912 and area under the ROC curve of 0.961, an improvement of approximately 0.027 over resizing alone. Sensitivity and specificity both exceeded 0.90. These results align with prior work reporting up to 0.10 gain via synthetic augmentation. The findings demonstrate the effectiveness of simple augmentation for 3D MRI classification and motivate future exploration of advanced augmentation methods and architectures such as 3D U-Net and vision transformers.
Understanding the binding specificity between T-cell receptors (TCRs) and peptide-major histocompatibility complexes (pMHCs) is central to immunotherapy and vaccine development. However, current predictive models struggle with generalization, especially in data-scarce settings and when faced with novel epitopes. We present LANTERN (Large lAnguage model-powered TCR-Enhanced Recognition Network), a deep learning framework that combines large-scale protein language models with chemical representations of peptides. By encoding TCR \b{eta}-chain sequences using ESM-1b and transforming peptide sequences into SMILES strings processed by MolFormer, LANTERN captures rich biological and chemical features critical for TCR-peptide recognition. Through extensive benchmarking against existing models such as ChemBERTa, TITAN, and NetTCR, LANTERN demonstrates superior performance, particularly in zero-shot and few-shot learning scenarios. Our model also benefits from a robust negative sampling strategy and shows significant clustering improvements via embedding analysis. These results highlight the potential of LANTERN to advance TCR-pMHC binding prediction and support the development of personalized immunotherapies.
Eye-tracking data has been shown to correlate with a user's knowledge level and query formulation behaviour. While previous work has focused primarily on eye gaze fixations for attention analysis, often requiring additional contextual information, our study investigates the memory-related cognitive dimension by relying solely on pupil dilation and gaze velocity to infer users' topic familiarity and query specificity without needing any contextual information. Using eye-tracking data collected via a lab user study (N=18), we achieved a Macro F1 score of 71.25% for predicting topic familiarity with a Gradient Boosting classifier, and a Macro F1 score of 60.54% with a k-nearest neighbours (KNN) classifier for query specificity. Furthermore, we developed a novel annotation guideline -- specifically tailored for question answering -- to manually classify queries as Specific or Non-specific. This study demonstrates the feasibility of eye-tracking to better understand topic familiarity and query specificity in search.
Area under the receiving operator characteristic curve (AUC) is commonly reported alongside binary prediction models. However, there are concerns that AUC might be a misleading measure of prediction performance in the rare event setting. This setting is common since many events of clinical importance are rare events. We conducted a simulation study to determine when or whether AUC is unstable in the rare event setting. Specifically, we aimed to determine whether the bias and variance of AUC are driven by the number of events or the event rate. We also investigated the behavior of other commonly used measures of prediction performance, including positive predictive value, accuracy, sensitivity, and specificity. Our results indicate that poor AUC behavior -- as measured by empirical bias, variability of cross-validated AUC estimates, and empirical coverage of confidence intervals -- is driven by the minimum class size, not event rate. Performance of sensitivity is driven by the number of events, while that of specificity is driven by the number of non-events. Other measures, including positive predictive value and accuracy, depend on the event rate even in large samples. AUC is reliable in the rare event setting provided that the total number of events is moderately large.
LLMs trained on massive datasets may inadvertently acquire sensitive information such as personal details and potentially harmful content. This risk is further heightened in multimodal LLMs as they integrate information from multiple modalities (image and text). Adversaries can exploit this knowledge through multimodal prompts to extract sensitive details. Evaluating how effectively MLLMs can forget such information (targeted unlearning) necessitates the creation of high-quality, well-annotated image-text pairs. While prior work on unlearning has focused on text, multimodal unlearning remains underexplored. To address this gap, we first introduce a multimodal unlearning benchmark, UnLOK-VQA (Unlearning Outside Knowledge VQA), as well as an attack-and-defense framework to evaluate methods for deleting specific multimodal knowledge from MLLMs. We extend a visual question-answering dataset using an automated pipeline that generates varying-proximity samples for testing generalization and specificity, followed by manual filtering for maintaining high quality. We then evaluate six defense objectives against seven attacks (four whitebox, three blackbox), including a novel whitebox method leveraging interpretability of hidden states. Our results show multimodal attacks outperform text- or image-only ones, and that the most effective defense removes answer information from internal model states. Additionally, larger models exhibit greater post-editing robustness, suggesting that scale enhances safety. UnLOK-VQA provides a rigorous benchmark for advancing unlearning in MLLMs.
Mechanistic interpretability of deep learning models has emerged as a crucial research direction for understanding the functioning of neural networks. While significant progress has been made in interpreting discriminative models like transformers, understanding generative models such as Variational Autoencoders (VAEs) remains challenging. This paper introduces a comprehensive causal intervention framework for mechanistic interpretability of VAEs. We develop techniques to identify and analyze "circuit motifs" in VAEs, examining how semantic factors are encoded, processed, and disentangled through the network layers. Our approach uses targeted interventions at different levels: input manipulations, latent space perturbations, activation patching, and causal mediation analysis. We apply our framework to both synthetic datasets with known causal relationships and standard disentanglement benchmarks. Results show that our interventions can successfully isolate functional circuits, map computational graphs to causal graphs of semantic factors, and distinguish between polysemantic and monosemantic units. Furthermore, we introduce metrics for causal effect strength, intervention specificity, and circuit modularity that quantify the interpretability of VAE components. Experimental results demonstrate clear differences between VAE variants, with FactorVAE achieving higher disentanglement scores (0.084) and effect strengths (mean 4.59) compared to standard VAE (0.064, 3.99) and Beta-VAE (0.051, 3.43). Our framework advances the mechanistic understanding of generative models and provides tools for more transparent and controllable VAE architectures.
Non-adherence to medications is a critical concern since nearly half of patients with chronic illnesses do not follow their prescribed medication regimens, leading to increased mortality, costs, and preventable human distress. Amongst stage 0-3 breast cancer survivors, adherence to long-term adjuvant endocrine therapy (i.e., Tamoxifen and aromatase inhibitors) is associated with a significant increase in recurrence-free survival. This work aims to develop multi-scale models of medication adherence to understand the significance of different factors influencing adherence across varying time frames. We introduce a computational framework guided by Social Cognitive Theory for multi-scale (daily and weekly) modeling of longitudinal medication adherence. Our models employ both dynamic medication-taking patterns in the recent past (dynamic factors) as well as less frequently changing factors (static factors) for adherence prediction. Additionally, we assess the significance of various factors in influencing adherence behavior across different time scales. Our models outperform traditional machine learning counterparts in both daily and weekly tasks in terms of both accuracy and specificity. Daily models achieved an accuracy of 87.25%, and weekly models, an accuracy of 76.04%. Notably, dynamic past medication-taking patterns prove most valuable for predicting daily adherence, while a combination of dynamic and static factors is significant for macro-level weekly adherence patterns.
Text-based prompting remains the predominant interaction paradigm in generative AI, yet it often introduces friction for novice users such as small business owners (SBOs), who struggle to articulate creative goals in domain-specific contexts like advertising. Through a formative study with six SBOs in the United Kingdom, we identify three key challenges: difficulties in expressing brand intuition through prompts, limited opportunities for fine-grained adjustment and refinement during and after content generation, and the frequent production of generic content that lacks brand specificity. In response, we present ACAI (AI Co-Creation for Advertising and Inspiration), a multimodal generative AI tool designed to support novice designers by moving beyond traditional prompt interfaces. ACAI features a structured input system composed of three panels: Branding, Audience and Goals, and the Inspiration Board. These inputs allow users to convey brand-relevant context and visual preferences. This work contributes to HCI research on generative systems by showing how structured interfaces can foreground user-defined context, improve alignment, and enhance co-creative control in novice creative workflows.
Multimodal deep learning harnesses diverse imaging modalities, such as MRI sequences, to enhance diagnostic accuracy in medical imaging. A key challenge is determining the optimal timing for integrating these modalities-specifically, identifying the network layers where fusion modules should be inserted. Current approaches often rely on manual tuning or exhaustive search, which are computationally expensive without any guarantee of converging to optimal results. We propose a sequential forward search algorithm that incrementally activates and evaluates candidate fusion modules at different layers of a multimodal network. At each step, the algorithm retrains from previously learned weights and compares validation loss to identify the best-performing configuration. This process systematically reduces the search space, enabling efficient identification of the optimal fusion timing without exhaustively testing all possible module placements. The approach is validated on two multimodal MRI datasets, each addressing different classification tasks. Our algorithm consistently identified configurations that outperformed unimodal baselines, late fusion, and a brute-force ensemble of all potential fusion placements. These architectures demonstrated superior accuracy, F-score, and specificity while maintaining competitive or improved AUC values. Furthermore, the sequential nature of the search significantly reduced computational overhead, making the optimization process more practical. By systematically determining the optimal timing to fuse imaging modalities, our method advances multimodal deep learning for medical imaging. It provides an efficient and robust framework for fusion optimization, paving the way for improved clinical decision-making and more adaptable, scalable architectures in medical AI applications.
Backdoor attacks pose a significant threat to deep neural networks, as backdoored models would misclassify poisoned samples with specific triggers into target classes while maintaining normal performance on clean samples. Among these, multi-target backdoor attacks can simultaneously target multiple classes. However, existing multi-target backdoor attacks all follow the dirty-label paradigm, where poisoned samples are mislabeled, and most of them require an extremely high poisoning rate. This makes them easily detectable by manual inspection. In contrast, clean-label attacks are more stealthy, as they avoid modifying the labels of poisoned samples. However, they generally struggle to achieve stable and satisfactory attack performance and often fail to scale effectively to multi-target attacks. To address this issue, we propose the Feature-based Full-target Clean-label Backdoor Attacks (FFCBA) which consists of two paradigms: Feature-Spanning Backdoor Attacks (FSBA) and Feature-Migrating Backdoor Attacks (FMBA). FSBA leverages class-conditional autoencoders to generate noise triggers that align perturbed in-class samples with the original category's features, ensuring the effectiveness, intra-class consistency, inter-class specificity and natural-feature correlation of triggers. While FSBA supports swift and efficient attacks, its cross-model attack capability is relatively weak. FMBA employs a two-stage class-conditional autoencoder training process that alternates between using out-of-class samples and in-class samples. This allows FMBA to generate triggers with strong target-class features, making it highly effective for cross-model attacks. We conduct experiments on multiple datasets and models, the results show that FFCBA achieves outstanding attack performance and maintains desirable robustness against the state-of-the-art backdoor defenses.
Multi-target backdoor attacks pose significant security threats to deep neural networks, as they can preset multiple target classes through a single backdoor injection. This allows attackers to control the model to misclassify poisoned samples with triggers into any desired target class during inference, exhibiting superior attack performance compared with conventional backdoor attacks. However, existing multi-target backdoor attacks fail to guarantee trigger specificity and stealthiness in black-box settings, resulting in two main issues. First, they are unable to simultaneously target all classes when only training data can be manipulated, limiting their effectiveness in realistic attack scenarios. Second, the triggers often lack visual imperceptibility, making poisoned samples easy to detect. To address these problems, we propose a Spatial-based Full-target Invisible Backdoor Attack, called SFIBA. It restricts triggers for different classes to specific local spatial regions and morphologies in the pixel space to ensure specificity, while employing a frequency-domain-based trigger injection method to guarantee stealthiness. Specifically, for injection of each trigger, we first apply fast fourier transform to obtain the amplitude spectrum of clean samples in local spatial regions. Then, we employ discrete wavelet transform to extract the features from the amplitude spectrum and use singular value decomposition to integrate the trigger. Subsequently, we selectively filter parts of the trigger in pixel space to implement trigger morphology constraints and adjust injection coefficients based on visual effects. We conduct experiments on multiple datasets and models. The results demonstrate that SFIBA can achieve excellent attack performance and stealthiness, while preserving the model's performance on benign samples, and can also bypass existing backdoor defenses.
Motivated by the emerging demand in the financial industry for the automatic analysis of unstructured and structured data at scale, Question Answering (QA) systems can provide lucrative and competitive advantages to companies by facilitating the decision making of financial advisers. Consequently, we propose a novel financial QA system using the transformer-based pre-trained BERT language model to address the limitations of data scarcity and language specificity in the financial domain. Our system focuses on financial non-factoid answer selection, which retrieves a set of passage-level texts and selects the most relevant as the answer. To increase efficiency, we formulate the answer selection task as a re-ranking problem, in which our system consists of an Answer Retriever using BM25, a simple information retrieval approach, to first return a list of candidate answers, and an Answer Re-ranker built with variants of pre-trained BERT language models to re-rank and select the most relevant answers. We investigate various learning, further pre-training, and fine-tuning approaches for BERT. Our experiments suggest that FinBERT-QA, a model built from applying the Transfer and Adapt further fine-tuning and pointwise learning approach, is the most effective, improving the state-of-the-art results of task 2 of the FiQA dataset by 16% on MRR, 17% on NDCG, and 21% on Precision@1.
While deep learning has shown strong performance in musculoskeletal imaging, existing work has largely focused on pathologies where diagnosis is not a clinical challenge, leaving more difficult problems underexplored, such as detecting Bankart lesions (anterior-inferior glenoid labral tears) on standard MRIs. Diagnosing these lesions is challenging due to their subtle imaging features, often leading to reliance on invasive MRI arthrograms (MRAs). This study introduces ScopeMRI, the first publicly available, expert-annotated dataset for shoulder pathologies, and presents a deep learning (DL) framework for detecting Bankart lesions on both standard MRIs and MRAs. ScopeMRI includes 586 shoulder MRIs (335 standard, 251 MRAs) from 558 patients who underwent arthroscopy. Ground truth labels were derived from intraoperative findings, the gold standard for diagnosis. Separate DL models for MRAs and standard MRIs were trained using a combination of CNNs and transformers. Predictions from sagittal, axial, and coronal views were ensembled to optimize performance. The models were evaluated on a 20% hold-out test set (117 MRIs: 46 MRAs, 71 standard MRIs). The models achieved an AUC of 0.91 and 0.93, sensitivity of 83% and 94%, and specificity of 91% and 86% for standard MRIs and MRAs, respectively. Notably, model performance on non-invasive standard MRIs matched or surpassed radiologists interpreting MRAs. External validation demonstrated initial generalizability across imaging protocols. This study demonstrates that DL models can achieve radiologist-level diagnostic performance on standard MRIs, reducing the need for invasive MRAs. By releasing ScopeMRI and a modular codebase for training and evaluating deep learning models on 3D medical imaging data, we aim to accelerate research in musculoskeletal imaging and support the development of new datasets for clinically challenging diagnostic tasks.
The current study describes a cost-effective method for adapting large language models (LLMs) for academic advising with study-abroad contexts in mind and for application in low-resource methods for acculturation. With the Mistral-7B-Instruct model applied with a Low-Rank Adaptation (LoRA) method and a 4-bit quantization method, the model underwent training in two distinct stages related to this study's purpose to enhance domain specificity while maintaining computational efficiency. In Phase 1, the model was conditioned with a synthetic dataset via the Gemini Pro API, and in Phase 2, it was trained with manually curated datasets from the StudyAbroadGPT project to achieve enhanced, contextualized responses. Technical innovations entailed memory-efficient quantization, parameter-efficient adaptation, and continuous training analytics via Weights & Biases. After training, this study demonstrated a reduction in training loss by 52.7%, 92% accuracy in domain-specific recommendations, achieved 95% markdown-based formatting support, and a median run-rate of 100 samples per second on off-the-shelf GPU equipment. These findings support the effective application of instruction-tuned LLMs within educational advisers, especially in low-resource institutional scenarios. Limitations included decreased generalizability and the application of a synthetically generated dataset, but this framework is scalable for adding new multilingual-augmented and real-time academic advising processes. Future directions may include plans for the integration of retrieval-augmented generation, applying dynamic quantization routines, and connecting to real-time academic databases to increase adaptability and accuracy.
Zero-shot anomaly detection (ZSAD) aims to identify anomalies in unseen categories by leveraging CLIP's zero-shot capabilities to match text prompts with visual features. A key challenge in ZSAD is learning general prompts stably and utilizing them effectively, while maintaining both generalizability and category specificity. Although general prompts have been explored in prior works, achieving their stable optimization and effective deployment remains a significant challenge. In this work, we propose GenCLIP, a novel framework that learns and leverages general prompts more effectively through multi-layer prompting and dual-branch inference. Multi-layer prompting integrates category-specific visual cues from different CLIP layers, enriching general prompts with more comprehensive and robust feature representations. By combining general prompts with multi-layer visual features, our method further enhances its generalization capability. To balance specificity and generalization, we introduce a dual-branch inference strategy, where a vision-enhanced branch captures fine-grained category-specific features, while a query-only branch prioritizes generalization. The complementary outputs from both branches improve the stability and reliability of anomaly detection across unseen categories. Additionally, we propose an adaptive text prompt filtering mechanism, which removes irrelevant or atypical class names not encountered during CLIP's training, ensuring that only meaningful textual inputs contribute to the final vision-language alignment.