Julia Ive

Ashwani Tanwar, Jingqing Zhang, Julia Ive et al.

Extracting phenotypes from clinical text has been shown to be useful for a variety of clinical use cases such as identifying patients with rare diseases. However, reasoning with numerical values remains challenging for phenotyping in clinical text, for example, temperature 102F representing Fever. Current state-of-the-art phenotyping models are able to detect general phenotypes, but perform poorly when they detect phenotypes requiring numerical reasoning. We present a novel unsupervised methodology leveraging external knowledge and contextualized word embeddings from ClinicalBERT for numerical reasoning in a variety of phenotypic contexts. Comparing against unsupervised benchmarks, it shows a substantial performance improvement with absolute gains on generalized Recall and F1 scores up to 79% and 71%, respectively. In the supervised setting, it also surpasses the performance of alternative approaches with absolute gains on generalized Recall and F1 scores up to 70% and 44%, respectively.

Damith Chamalke Senadeera, Julia Ive

Controlled text generation is a very important task in the arena of natural language processing due to its promising applications. In order to achieve this task we mainly introduce the novel soft prompt tuning method of using soft prompts at both encoder and decoder levels together in a T5 model and investigate the performance as the behaviour of an additional soft prompt related to the decoder of a T5 model in controlled text generation remained unexplored. Then we also investigate the feasibility of steering the output of this extended soft prompted T5 model at decoder level and finally analyse the utility of generated text to be used in AI related tasks such as training AI models with an interpretability analysis of the classifier trained with synthetic text, as there is a lack of proper analysis of methodologies in generating properly labelled data to be utilized in AI tasks. Through the performed in-depth intrinsic and extrinsic evaluations of this generation model along with the artificially generated data, we found that this model produced better results compared to the T5 model with a single soft prompt at encoder level and the sentiment classifier trained using this artificially generated data can produce comparable classification results to the results of a classifier trained with real labelled data and also the classifier decision is interpretable with respect to the input text content.

Heng-Yi Wu, Jingqing Zhang, Julia Ive et al.

Structured (tabular) data in the preclinical and clinical domains contains valuable information about individuals and an efficient table-to-text summarization system can drastically reduce manual efforts to condense this data into reports. However, in practice, the problem is heavily impeded by the data paucity, data sparsity and inability of the state-of-the-art natural language generation models (including T5, PEGASUS and GPT-Neo) to produce accurate and reliable outputs. In this paper, we propose a novel table-to-text approach and tackle these problems with a novel two-step architecture which is enhanced by auto-correction, copy mechanism and synthetic data augmentation. The study shows that the proposed approach selects salient biomedical entities and values from structured data with improved precision (up to 0.13 absolute increase) of copying the tabular values to generate coherent and accurate text for assay validation reports and toxicology reports. Moreover, we also demonstrate a light-weight adaptation of the proposed system to new datasets by fine-tuning with as little as 40\% training examples. The outputs of our model are validated by human experts in the Human-in-the-Loop scenario.

Hongshu Liu, Nabeel Seedat, Julia Ive

Computational models providing accurate estimates of their uncertainty are crucial for risk management associated with decision making in healthcare contexts. This is especially true since many state-of-the-art systems are trained using the data which has been labelled automatically (self-supervised mode) and tend to overfit. In this work, we investigate the quality of uncertainty estimates from a range of current state-of-the-art predictive models applied to the problem of observation detection in radiology reports. This problem remains understudied for Natural Language Processing in the healthcare domain. We demonstrate that Gaussian Processes (GPs) provide superior performance in quantifying the risks of 3 uncertainty labels based on the negative log predictive probability (NLPP) evaluation metric and mean maximum predicted confidence levels (MMPCL), whilst retaining strong predictive performance.

Amy Makawana, Abhijeet Parida, Marius George Linguraru et al.

Self-supervised learning (SSL) has advanced medical image analysis be enabling learning form large unlabelled data. However, in brain magnetic resonance imaging (MRI), most 3D models remain specialized for either segmentation of classification, limiting their ability to generalize across datasets, imaging protocols,, and downstream tasks. This lack of transferability constrains the clinical utility of 3D MRI models, despite the availability of unlabeled volumetric data. We present Volta-3D, a self-supervised 3D Vision Transformer framework designed to learn transferable volumetric representations. Volta-3D jointly aligns global class-style tokens and local patch tokens within a student-teacher paradigm and enforces fine-grained structural reconstruction. This combined global-local alignment addresses the limited semantic diversity and subtle anatomical characteristics of brain MRI, which challenges existing SSL approaches. We evaluate Volta-3D on multiple out-of-distribution downstream tasks, including hippocampal segmentation and classification of sex and Alzheimer's disease versus healthy controls. Across all tasks, representations learned by Volta-3D outperform randomly initialized baselines, demonstrating improved transferability and robustness under domain shift. Hence jointly enforcing global semantic consistency and local structural learning during pretraining enables broader concept learning from unlabeled brain MRI data. Overall VolTA-3D supports effective multi-task downstream performance with task-specific pertaining, a step towards generalizable and clinically viable 3D models.

Eric Hua Qing Zhang, Julia Ive

Mental health illness represents a substantial global socioeconomic burden, with COVID-19 further exacerbating accessibility challenges and driving increased demand for telehealth mental health support. While large language models (LLMs) offer promising solutions through 24/7 availability and non-judgmental interactions, pre-trained models often lack the contextual and emotional awareness necessary for appropriate therapeutic responses. This paper investigated the application of supervised fine-tuning (SFT) and reinforcement learning (RL) techniques to enhance GPT-2's capacity for therapeutic dialogue generation. The methodology restructured input formats to enable simultaneous processing of contextual information and emotional states alongside user input, employing a multi-component reward function that aligned model outputs with professional therapist responses and annotated emotions. Results demonstrated improvements through reinforcement learning over baseline GPT-2 across multiple evaluation metrics: BLEU (0.0111), ROUGE-1 (0.1397), ROUGE-2 (0.0213), ROUGE-L (0.1317), and METEOR (0.0581). LLM evaluation confirmed high contextual relevance and professionalism, while reinforcement learning achieved 99.34% emotion accuracy compared to 66.96% for baseline GPT-2. These findings demonstrate reinforcement learning's effectiveness in developing therapeutic dialogue systems that can serve as valuable assistive tools for therapists while maintaining essential human clinical oversight.

Qing Ding, Eric Hua Qing Zhang, Felix Jozsa et al.

Large language models (LLMs) are increasingly used in healthcare, yet standardised benchmarks for evaluating guideline-based clinical reasoning are missing. This study introduces a validated dataset derived from publicly available guidelines across multiple diagnoses. The dataset was created with the help of GPT and contains realistic patient scenarios, as well as clinical questions. We benchmark a range of recent popular LLMs to showcase the validity of our dataset. The framework supports systematic evaluation of LLMs' clinical utility and guideline adherence.

Vjosa Preniqi, Iacopo Ghinassi, Julia Ive et al.

Moral values play a fundamental role in how we evaluate information, make decisions, and form judgements around important social issues. Controversial topics, including vaccination, abortion, racism, and sexual orientation, often elicit opinions and attitudes that are not solely based on evidence but rather reflect moral worldviews. Recent advances in Natural Language Processing (NLP) show that moral values can be gauged in human-generated textual content. Building on the Moral Foundations Theory (MFT), this paper introduces MoralBERT, a range of language representation models fine-tuned to capture moral sentiment in social discourse. We describe a framework for both aggregated and domain-adversarial training on multiple heterogeneous MFT human-annotated datasets sourced from Twitter (now X), Reddit, and Facebook that broaden textual content diversity in terms of social media audience interests, content presentation and style, and spreading patterns. We show that the proposed framework achieves an average F1 score that is between 11% and 32% higher than lexicon-based approaches, Word2Vec embeddings, and zero-shot classification with large language models such as GPT-4 for in-domain inference. Domain-adversarial training yields better out-of domain predictions than aggregate training while achieving comparable performance to zero-shot learning. Our approach contributes to annotation-free and effective morality learning, and provides useful insights towards a more comprehensive understanding of moral narratives in controversial social debates using NLP.

Jiayu Song, Jenny Chim, Adam Tsakalidis et al.

We introduce a hybrid abstractive summarisation approach combining hierarchical VAE with LLMs (LlaMA-2) to produce clinically meaningful summaries from social media user timelines, appropriate for mental health monitoring. The summaries combine two different narrative points of view: clinical insights in third person useful for a clinician are generated by feeding into an LLM specialised clinical prompts, and importantly, a temporally sensitive abstractive summary of the user's timeline in first person, generated by a novel hierarchical variational autoencoder, TH-VAE. We assess the generated summaries via automatic evaluation against expert summaries and via human evaluation with clinical experts, showing that timeline summarisation by TH-VAE results in more factual and logically coherent summaries rich in clinical utility and superior to LLM-only approaches in capturing changes over time.

Julia Ive, Felix Jozsa, Nick Jackson et al.

Background: Clinical guidelines are central to safe evidence-based medicine in modern healthcare, providing diagnostic criteria, treatment options and monitoring advice for a wide range of illnesses. LLM-empowered chatbots have shown great promise in Healthcare Q&A tasks, offering the potential to provide quick and accurate responses to medical inquiries. Our main objective was the development and preliminary assessment of an LLM-empowered chatbot software capable of reliably answering clinical guideline questions using University College London Hospital (UCLH) clinical guidelines. Methods: We used the open-weight Llama-3.1-8B LLM to extract relevant information from the UCLH guidelines to answer questions. Our approach highlights the safety and reliability of referencing information over its interpretation and response generation. Seven doctors from the ward assessed the chatbot's performance by comparing its answers to the gold standard. Results: Our chatbot demonstrates promising performance in terms of relevance, with ~73% of its responses rated as very relevant, showcasing a strong understanding of the clinical context. Importantly, our chatbot achieves a recall of 1.00 for extracted guideline lines, substantially minimising the risk of missing critical information. Approximately 78% of responses were rated satisfactory in terms of completeness. A small portion (~14.5%) contained minor unnecessary information, indicating occasional lapses in precision. The chatbot' showed high efficiency, with an average completion time of 10 seconds, compared to 30 seconds for human respondents. Evaluation of clinical reasoning showed that 72% of the chatbot's responses were without flaws. Our chatbot demonstrates significant potential to speed up and improve the process of accessing locally relevant clinical information for healthcare professionals.

Mariia Ignashina, Paulina Bondaronek, Dan Santel et al.

Traditional depression screening methods, such as the PHQ-9, are particularly challenging for children in pediatric primary care due to practical limitations. AI has the potential to help, but the scarcity of annotated datasets in mental health, combined with the computational costs of training, highlights the need for efficient, zero-shot approaches. In this work, we investigate the feasibility of state-of-the-art LLMs for depressive symptom extraction in pediatric settings (ages 6-24). This approach aims to complement traditional screening and minimize diagnostic errors. Our findings show that all LLMs are 60% more efficient than word match, with Flan leading in precision (average F1: 0.65, precision: 0.78), excelling in the extraction of more rare symptoms like "sleep problems" (F1: 0.92) and "self-loathing" (F1: 0.8). Phi strikes a balance between precision (0.44) and recall (0.60), performing well in categories like "Feeling depressed" (0.69) and "Weight change" (0.78). Llama 3, with the highest recall (0.90), overgeneralizes symptoms, making it less suitable for this type of analysis. Challenges include the complexity of clinical notes and overgeneralization from PHQ-9 scores. The main challenges faced by LLMs include navigating the complex structure of clinical notes with content from different times in the patient trajectory, as well as misinterpreting elevated PHQ-9 scores. We finally demonstrate the utility of symptom annotations provided by Flan as features in an ML algorithm, which differentiates depression cases from controls with high precision of 0.78, showing a major performance boost compared to a baseline that does not use these features.

Juan Jose Rubio Jan, Jack Wu, Julia Ive

This study applies Large Language Models (LLMs) to two foundational Electronic Health Record (EHR) data science tasks: structured data querying (using programmatic languages, Python/Pandas) and information extraction from unstructured clinical text via a Retrieval Augmented Generation (RAG) pipeline. We test the ability of LLMs to interact accurately with large structured datasets for analytics and the reliability of LLMs in extracting semantically correct information from free text health records when supported by RAG. To this end, we presented a flexible evaluation framework that automatically generates synthetic question and answer pairs tailored to the characteristics of each dataset or task. Experiments were conducted on a curated subset of MIMIC III, (four structured tables and one clinical note type), using a mix of locally hosted and API-based LLMs. Evaluation combined exact-match metrics, semantic similarity, and human judgment. Our findings demonstrate the potential of LLMs to support precise querying and accurate information extraction in clinical workflows.

Zhijun Guo, Alvina Lai, Julia Ive et al.

Static tools like the Patient Health Questionnaire-9 (PHQ-9) effectively screen depression but lack interactivity and adaptability. We developed HopeBot, a chatbot powered by a large language model (LLM) that administers the PHQ-9 using retrieval-augmented generation and real-time clarification. In a within-subject study, 132 adults in the United Kingdom and China completed both self-administered and chatbot versions. Scores demonstrated strong agreement (ICC = 0.91; 45% identical). Among 75 participants providing comparative feedback, 71% reported greater trust in the chatbot, highlighting clearer structure, interpretive guidance, and a supportive tone. Mean ratings (0-10) were 8.4 for comfort, 7.7 for voice clarity, 7.6 for handling sensitive topics, and 7.4 for recommendation helpfulness; the latter varied significantly by employment status and prior mental-health service use (p < 0.05). Overall, 87.1% expressed willingness to reuse or recommend HopeBot. These findings demonstrate voice-based LLM chatbots can feasibly serve as scalable, low-burden adjuncts for routine depression screening.

Gianluca Manzo, Julia Ive

Automating chest radiograph interpretation using Deep Learning (DL) models has the potential to significantly improve clinical workflows, decision-making, and large-scale health screening. However, in medical settings, merely optimising predictive performance is insufficient, as the quantification of uncertainty is equally crucial. This paper investigates the relationship between predictive uncertainty, derived from Bayesian Deep Learning approximations, and human/linguistic uncertainty, as estimated from free-text radiology reports labelled by rule-based labellers. Utilising BERT as the model of choice, this study evaluates different binarisation methods for uncertainty labels and explores the efficacy of Monte Carlo Dropout and Deep Ensembles in estimating predictive uncertainty. The results demonstrate good model performance, but also a modest correlation between predictive and linguistic uncertainty, highlighting the challenges in aligning machine uncertainty with human interpretation nuances. Our findings suggest that while Bayesian approximations provide valuable uncertainty estimates, further refinement is necessary to fully capture and utilise the subtleties of human uncertainty in clinical applications.

Julia Ive, Olatomiwa Olukoya, Jonathan P. Funnell et al.

Introduction: Timely care in a specialised neuro-intensive therapy unit (ITU) reduces mortality and hospital stays, with planned admissions being safer than unplanned ones. However, post-operative care decisions remain subjective. This study used artificial intelligence (AI), specifically natural language processing (NLP) to analyse electronic health records (EHRs) and predict ITU admissions for elective surgery patients. Methods: This study analysed the EHRs of elective neurosurgery patients from University College London Hospital (UCLH) using NLP. Patients were categorised into planned high dependency unit (HDU) or ITU admission; unplanned HDU or ITU admission; or ward / overnight recovery (ONR). The Medical Concept Annotation Tool (MedCAT) was used to identify SNOMED-CT concepts within the clinical notes. We then explored the utility of these identified concepts for a range of AI algorithms trained to predict ITU admission. Results: The CogStack-MedCAT NLP model, initially trained on hospital-wide EHRs, underwent two refinements: first with data from patients with Normal Pressure Hydrocephalus (NPH) and then with data from Vestibular Schwannoma (VS) patients, achieving a concept detection F1-score of 0.93. This refined model was then used to extract concepts from EHR notes of 2,268 eligible neurosurgical patients. We integrated the extracted concepts into AI models, including a decision tree model and a neural time-series model. Using the simpler decision tree model, we achieved a recall of 0.87 (CI 0.82 - 0.91) for ITU admissions, reducing the proportion of unplanned ITU cases missed by human experts from 36% to 4%. Conclusion: The NLP model, refined for accuracy, has proven its efficiency in extracting relevant concepts, providing a reliable basis for predictive AI models to use in clinically valid applications.

Julia Ive, Paulina Bondaronek, Vishal Yadav et al.

Introduction: Healthcare AI models often inherit biases from their training data. While efforts have primarily targeted bias in structured data, mental health heavily depends on unstructured data. This study aims to detect and mitigate linguistic differences related to non-biological differences in the training data of AI models designed to assist in pediatric mental health screening. Our objectives are: (1) to assess the presence of bias by evaluating outcome parity across sex subgroups, (2) to identify bias sources through textual distribution analysis, and (3) to develop a de-biasing method for mental health text data. Methods: We examined classification parity across demographic groups and assessed how gendered language influences model predictions. A data-centric de-biasing method was applied, focusing on neutralizing biased terms while retaining salient clinical information. This methodology was tested on a model for automatic anxiety detection in pediatric patients. Results: Our findings revealed a systematic under-diagnosis of female adolescent patients, with a 4% lower accuracy and a 9% higher False Negative Rate (FNR) compared to male patients, likely due to disparities in information density and linguistic differences in patient notes. Notes for male patients were on average 500 words longer, and linguistic similarity metrics indicated distinct word distributions between genders. Implementing our de-biasing approach reduced diagnostic bias by up to 27%, demonstrating its effectiveness in enhancing equity across demographic groups. Discussion: We developed a data-centric de-biasing framework to address gender-based content disparities within clinical text. By neutralizing biased language and enhancing focus on clinically essential information, our approach demonstrates an effective strategy for mitigating bias in AI healthcare models trained on text.

Mariia Ignashina, Julia Ive

Current text generation models are trained using real data which can potentially contain sensitive information, such as confidential patient information and the like. Under certain conditions output of the training data which they have memorised can be triggered, exposing sensitive data. To mitigate against this risk we propose a safer alternative which sees fragmented data in the form of domain-specific short phrases randomly grouped together shared instead of full texts. Thus, text fragments that could re-identify an individual cannot be reproduced by the model in one sequence, giving significant protection against linkage attacks. We fine-tune several state-of-the-art LLMs using meaningful syntactic chunks to explore their utility. In particular, we fine-tune BERT-based models to predict two cardiovascular diagnoses. Our results demonstrate the capacity of LLMs to benefit from the pre-trained knowledge and deliver classification results when fine-tuned with fragmented data comparable to fine-tuning with full training data.

Mohammed Ataaur Rahaman, Julia Ive

Source code clone detection is the task of finding code fragments that have the same or similar functionality, but may differ in syntax or structure. This task is important for software maintenance, reuse, and quality assurance (Roy et al. 2009). However, code clone detection is challenging, as source code can be written in different languages, domains, and styles. In this paper, we argue that source code is inherently a graph, not a sequence, and that graph-based methods are more suitable for code clone detection than sequence-based methods. We compare the performance of two state-of-the-art models: CodeBERT (Feng et al. 2020), a sequence-based model, and CodeGraph (Yu et al. 2023), a graph-based model, on two benchmark data-sets: BCB (Svajlenko et al. 2014) and PoolC (PoolC no date). We show that CodeGraph outperforms CodeBERT on both data-sets, especially on cross-lingual code clones. To the best of our knowledge, this is the first work to demonstrate the superiority of graph-based methods over sequence-based methods on cross-lingual code clone detection.

Atijit Anuchitanukul, Julia Ive, Lucia Specia

Understanding toxicity in user conversations is undoubtedly an important problem. Addressing "covert" or implicit cases of toxicity is particularly hard and requires context. Very few previous studies have analysed the influence of conversational context in human perception or in automated detection models. We dive deeper into both these directions. We start by analysing existing contextual datasets and come to the conclusion that toxicity labelling by humans is in general influenced by the conversational structure, polarity and topic of the context. We then propose to bring these findings into computational detection models by introducing and evaluating (a) neural architectures for contextual toxicity detection that are aware of the conversational structure, and (b) data augmentation strategies that can help model contextual toxicity detection. Our results have shown the encouraging potential of neural architectures that are aware of the conversation structure. We have also demonstrated that such models can benefit from synthetic data, especially in the social media domain.

Jingqing Zhang, Luis Bolanos, Tong Li et al.

Contextualised word embeddings is a powerful tool to detect contextual synonyms. However, most of the current state-of-the-art (SOTA) deep learning concept extraction methods remain supervised and underexploit the potential of the context. In this paper, we propose a self-supervised pre-training approach which is able to detect contextual synonyms of concepts being training on the data created by shallow matching. We apply our methodology in the sparse multi-class setting (over 15,000 concepts) to extract phenotype information from electronic health records. We further investigate data augmentation techniques to address the problem of the class sparsity. Our approach achieves a new SOTA for the unsupervised phenotype concept annotation on clinical text on F1 and Recall outperforming the previous SOTA with a gain of up to 4.5 and 4.0 absolute points, respectively. After fine-tuning with as little as 20\% of the labelled data, we also outperform BioBERT and ClinicalBERT. The extrinsic evaluation on three ICU benchmarks also shows the benefit of using the phenotypes annotated by our model as features.

Jingqing Zhang, Luis Bolanos, Ashwani Tanwar et al.

Objective: Clinical notes contain information not present elsewhere, including drug response and symptoms, all of which are highly important when predicting key outcomes in acute care patients. We propose the automatic annotation of phenotypes from clinical notes as a method to capture essential information, which is complementary to typically used vital signs and laboratory test results, to predict outcomes in the Intensive Care Unit (ICU). Methods: We develop a novel phenotype annotation model to annotate phenotypic features of patients which are then used as input features of predictive models to predict ICU patient outcomes. We demonstrate and validate our approach conducting experiments on three ICU prediction tasks including in-hospital mortality, physiological decompensation and length of stay for over 24,000 patients by using MIMIC-III dataset. Results: The predictive models incorporating phenotypic information achieve 0.845 (AUC-ROC) to predict in-hospital mortality, 0.839 (AUC-ROC) for physiological decompensation and 0.430 (Kappa) for length of stay, all of which consistently outperform the baseline models leveraging only vital signs and laboratory test results. Moreover, we conduct a thorough interpretability study, showing that phenotypes provide valuable insights at the patient and cohort levels. Conclusion: The proposed approach demonstrates phenotypic information complements traditionally used vital signs and laboratory test results, improving significantly forecast of outcomes in the ICU.

Julia Ive, Zixu Wang, Marina Fomicheva et al.

Reinforcement Learning (RL) is a powerful framework to address the discrepancy between loss functions used during training and the final evaluation metrics to be used at test time. When applied to neural Machine Translation (MT), it minimises the mismatch between the cross-entropy loss and non-differentiable evaluation metrics like BLEU. However, the suitability of these metrics as reward function at training time is questionable: they tend to be sparse and biased towards the specific words used in the reference texts. We propose to address this problem by making models less reliant on such metrics in two ways: (a) with an entropy-regularised RL method that does not only maximise a reward function but also explore the action space to avoid peaky distributions; (b) with a novel RL method that explores a dynamic unsupervised reward function to balance between exploration and exploitation. We base our proposals on the Soft Actor-Critic (SAC) framework, adapting the off-policy maximum entropy model for language generation applications such as MT. We demonstrate that SAC with BLEU reward tends to overfit less to the training data and performs better on out-of-domain data. We also show that our dynamic unsupervised reward can lead to better translation of ambiguous words.

Julia Ive, Andy Mingren Li, Yishu Miao et al.

This paper addresses the problem of simultaneous machine translation (SiMT) by exploring two main concepts: (a) adaptive policies to learn a good trade-off between high translation quality and low latency; and (b) visual information to support this process by providing additional (visual) contextual information which may be available before the textual input is produced. For that, we propose a multimodal approach to simultaneous machine translation using reinforcement learning, with strategies to integrate visual and textual information in both the agent and the environment. We provide an exploration on how different types of visual information and integration strategies affect the quality and latency of simultaneous translation models, and demonstrate that visual cues lead to higher quality while keeping the latency low.

Ozan Caglayan, Julia Ive, Veneta Haralampieva et al.

Simultaneous machine translation (SiMT) aims to translate a continuous input text stream into another language with the lowest latency and highest quality possible. The translation thus has to start with an incomplete source text, which is read progressively, creating the need for anticipation. In this paper, we seek to understand whether the addition of visual information can compensate for the missing source context. To this end, we analyse the impact of different multimodal approaches and visual features on state-of-the-art SiMT frameworks. Our results show that visual context is helpful and that visually-grounded models based on explicit object region information are much better than commonly used global features, reaching up to 3 BLEU points improvement under low latency scenarios. Our qualitative analysis illustrates cases where only the multimodal systems are able to translate correctly from English into gender-marked languages, as well as deal with differences in word order, such as adjective-noun placement between English and French.

Zixiu Wu, Ozan Caglayan, Julia Ive et al.

This paper describes the cascaded multimodal speech translation systems developed by Imperial College London for the IWSLT 2019 evaluation campaign. The architecture consists of an automatic speech recognition (ASR) system followed by a Transformer-based multimodal machine translation (MMT) system. While the ASR component is identical across the experiments, the MMT model varies in terms of the way of integrating the visual context (simple conditioning vs. attention), the type of visual features exploited (pooled, convolutional, action categories) and the underlying architecture. For the latter, we explore both the canonical transformer and its deliberation version with additive and cascade variants which differ in how they integrate the textual attention. Upon conducting extensive experiments, we found that (i) the explored visual integration schemes often harm the translation performance for the transformer and additive deliberation, but considerably improve the cascade deliberation; (ii) the transformer and cascade deliberation integrate the visual modality better than the additive deliberation, as shown by the incongruence analysis.

Zixiu Wu, Julia Ive, Josiah Wang et al.

We address the task of text translation on the How2 dataset using a state of the art transformer-based multimodal approach. The question we ask ourselves is whether visual features can support the translation process, in particular, given that this is a dataset extracted from videos, we focus on the translation of actions, which we believe are poorly captured in current static image-text datasets currently used for multimodal translation. For that purpose, we extract different types of action features from the videos and carefully investigate how helpful this visual information is by testing whether it can increase translation quality when used in conjunction with (i) the original text and (ii) the original text where action-related words (or all verbs) are masked out. The latter is a simulation that helps us assess the utility of the image in cases where the text does not provide enough context about the action, or in the presence of noise in the input text.

Zixu Wang, Julia Ive, Sumithra Velupillai et al.

A major obstacle to the development of Natural Language Processing (NLP) methods in the biomedical domain is data accessibility. This problem can be addressed by generating medical data artificially. Most previous studies have focused on the generation of short clinical text, and evaluation of the data utility has been limited. We propose a generic methodology to guide the generation of clinical text with key phrases. We use the artificial data as additional training data in two key biomedical NLP tasks: text classification and temporal relation extraction. We show that artificially generated training data used in conjunction with real training data can lead to performance boosts for data-greedy neural network algorithms. We also demonstrate the usefulness of the generated data for NLP setups where it fully replaces real training data.

Julia Ive, Pranava Madhyastha, Lucia Specia

Previous work on multimodal machine translation has shown that visual information is only needed in very specific cases, for example in the presence of ambiguous words where the textual context is not sufficient. As a consequence, models tend to learn to ignore this information. We propose a translate-and-refine approach to this problem where images are only used by a second stage decoder. This approach is trained jointly to generate a good first draft translation and to improve over this draft by (i) making better use of the target language textual context (both left and right-side contexts) and (ii) making use of visual context. This approach leads to the state of the art results. Additionally, we show that it has the ability to recover from erroneous or missing words in the source language.