Thailand

Large Language Models (LLMs) have demonstrated exceptional performances in a wide range of natural language processing tasks. However, their success does not always extend to machine translation, particularly in challenging scenarios such as translating low-resource languages. This study investigates the multilingual capability of LLMs, with a case study on Irish, an extremely low-resource language, focusing on translation tasks between English and Irish. We propose a dynamic, efficient language adaptation framework for English-centric LLMs, which involves layer-specific adjustments and subsequent fine-tuning for machine translation. Our findings highlight several key insights: (1) different layers in the LLM serve distinct functions such as language understanding and task reasoning, (2) effective translation requires extensive pre-training on both source and target languages, and (3) targeted fine-tuning for machine translation leads to significant improvements of 36.7% for English to Irish and 133.4% for Irish to English compared to the previous state-of-the-art.

ACL 2024

Irish-based Large Language Model with Extreme Low-Resource Settings in Machine Translation

large language model

language adaptation

low-resource language

machine translation

workshop paper

### Welcome!
The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. Our Virtual Poster Sessions will take place online Thursday, August 22, 2024.

You are required to register for this event. **Please register [here](https://2024.aclweb.org/registration). **

If you have already registered, please check your inbox for an email from Underline granting you access to ACL 2024 content.

Please register!

The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. More information will be announced soon.

Domain adaptation is crucial in the clinical domain since the performance of a model trained on one domain (source) degrades seriously when applied to another domain (target).
However, conventional domain adaptation methods often cannot be applied due to data sharing restrictions on source data.
Source-Free Domain Adaptation (SFDA) addresses this issue by only utilizing a source model and unlabeled target data to adapt to the target domain.
In SFDA, self-training is the most widely applied method involving retraining models with target data using predictions from the source model as pseudo-labels. Nevertheless, this approach is prone to contain substantial numbers of errors in pseudo-labeling and might limit model performance in the target domain.
In this paper, we propose a Source-Free Prototype-based Self-training (SFPS) aiming to improve the performance of self-training.
SFPS generates prototypes without accessing source data and utilizes them for prototypical learning, namely prototype-based pseudo-labeling and contrastive learning.
Also, we compare entropy-based, centroid-based, and class-weights-based prototype generation methods to identify the most effective formulation of the proposed method.
Experimental results across various datasets demonstrate the effectiveness of the proposed method, consistently outperforming vanilla self-training. 
The comparison of various prototype-generation methods identifies the most reliable generation method that improves the source model persistently. Additionally, our analysis illustrates SFPS can successfully alleviate errors in pseudo-labeling.

Improving Self-training with Prototypical Learning for Source-Free Domain Adaptation on Clinical Text

Adolescents exposed to advertisements promoting addictive substances exhibit a higher likelihood of subsequent substance use. The predominant source for youth exposure to such advertisements is through online content accessed via smartphones. Detecting these advertisements is crucial for establishing and maintaining a safer online environment for young people. In our study, we utilized Multimodal Large Language Models (MLLMs) to identify addictive substance advertisements in digital media. The performance of MLLMs depends on the quality of the prompt used to instruct the model. To optimize our prompts, an adaptive prompt engineering approach was implemented, leveraging a genetic algorithm to refine and enhance the prompts. To evaluate the model's performance, we augmented the RICO dataset, consisting of Android user interface screenshots, by superimposing alcohol ads onto them. Our results indicate that the MLLM can detect advertisements promoting alcohol with a 0.94 accuracy and a 0.94 F1 score.

Optimizing Multimodal Large Language Models for Detection of Alcohol Advertisements via Adaptive Prompting

Large Language Models (LLMs) have significantly advanced healthcare innovation on generation capabilities. However, their application in real clinical settings is challenging due to potential deviations from medical facts and inherent biases. In this work, we develop an augmented LLM framework, KG-Rank, which leverages a medical knowledge graph (KG) with ranking and re-ranking techniques, aiming to improve free-text question-answering (QA) in the medical domain. Specifically, upon receiving a question, we initially retrieve triplets from a medical KG to gather factual information. Subsequently, we innovatively apply ranking methods to refine the ordering of these triplets, aiming to yield more precise answers. To the best of our knowledge, KG-Rank is the first application of ranking models combined with KG in medical QA specifically for generating long answers. Evaluation of four selected medical QA datasets shows that KG-Rank achieves an improvement of over 18% in the ROUGE-L score. Moreover, we extend KG-Rank to open domains, where it realizes a 14% improvement in ROUGE-L, showing the effectiveness and potential of KG-Rank.

KG-Rank: Enhancing Large Language Models for Medical QA with Knowledge Graphs and Ranking Techniques

We fill a gap in scholarship by applying a generative Large Language Model (LLM) to extract information from clinical free text about the frequency of seizures experienced by people with epilepsy. Seizure frequency is difficult to determine across time from unstructured doctors' and nurses' reports of outpatients' visits that are stored in Electronic Health Records (EHRs) in the United Kingdom's National Health Service (NHS). We employ Meta's Llama 2 to mine the EHRs of people with epilepsy and determine, where possible, a person's seizure frequency at a given point in time. The results demonstrate that the new, powerful generative LLMs may improve outcomes for clinical NLP research in epilepsy and other areas.

Extracting Epilepsy Patient Data with Llama 2

This paper introduces MedExQA, a novel benchmark in medical question-answering, to evaluate large language models' (LLMs) understanding of medical knowledge through explanations. By constructing datasets across five distinct medical specialties that are underrepresented in current datasets and further incorporating multiple explanations for each question-answer pair, we address a major gap in current medical QA benchmarks which is the absence of comprehensive assessments of LLMs' ability to generate nuanced medical explanations. Our work highlights the importance of explainability in medical LLMs, proposes an effective methodology for evaluating models beyond classification accuracy, and sheds light on one specific domain, speech language pathology, where current LLMs including GPT4 lack good understanding. Our results show generation evaluation with multiple explanations aligns better with human assessment, highlighting an opportunity for a more robust automated comprehension assessment for LLMs. To diversify open-source medical LLMs (currently mostly based on Llama2), this work also proposes a new medical model, MedPhi-2, based on Phi-2 (2.7B). The model outperformed medical LLMs based on Llama2-70B in generating explanations, showing its effectiveness in the resource-constrained medical domain. We will share our benchmark datasets and the trained model.

MedExQA: Medical Question Answering Benchmark with Multiple Explanations

We present a novel approach to automating the identification of risk factors for diseases from medical literature, leveraging pre-trained models in the bio-medical domain, while tuning them for the specific task. Faced with the challenges of the diverse and unstructured nature of medical articles, our study introduces a multi-step system to first identify relevant articles, then classify them based on the presence of risk factor discussions and, finally, extract specific risk factor information for a disease through a question-answering model. 

Our contributions include the development of a comprehensive pipeline for the automated extraction of risk factors and the compilation of several datasets, which can serve as valuable resources for further research in this area. These datasets encompass a wide range of diseases, as well as their associated risk factors, meticulously identified and validated through a fine-grained evaluation scheme. We conducted both automatic and thorough manual evaluation, demonstrating encouraging results. We also highlight the importance of improving models and expanding dataset comprehensiveness to keep pace with the rapidly evolving field of medical research.

Automatic Extraction of Disease Risk Factors from Medical Publications

Biomedical question-answering systems remain popular for biomedical experts interacting with the literature to answer their medical questions. However, these systems are difficult to evaluate in the absence of costly human experts. Therefore, automatic evaluation metrics are often used in this space.
Traditional automatic metrics such as ROUGE or BLEU, which rely on token overlap, have shown a low correlation with humans. 
We present a study that uses large language models (LLMs) to automatically evaluate systems from an international challenge on biomedical semantic indexing and question answering, called BioASQ. We measure the agreement of LLM-produced scores against human judgements. We show that LLMs correlate similarly to lexical methods when using basic prompting techniques. However, by aggregating evaluators with LLMs or by fine-tuning, we find that our methods outperform the baselines by a large margin, achieving a Spearman correlation of 0.501 and 0.511, respectively.

Using Large Language Models to Evaluate Biomedical Query-Focused Summarisation

Large language models (LLMs) have recently become the leading source of answers for users' questions online. Despite their ability to offer eloquent answers, their accuracy and reliability can pose a significant challenge. This is especially true for sensitive domains such as biomedicine, where there is a higher need for factually correct answers. This paper introduces a biomedical retrieval-augmented generation (RAG) system designed to enhance the reliability of generated responses. The system is based on a fine-tuned LLM for the referenced question-answering, where retrieved relevant abstracts from PubMed are passed to LLM's context as input through a prompt. Its output is an answer based on PubMed abstracts, where each statement is referenced accordingly, allowing the users to verify the answer. Our retrieval system achieves an absolute improvement of 23% compared to the PubMed search engine. Based on the manual evaluation on a small sample, our fine-tuned LLM component achieves comparable results to GPT-4 Turbo in referencing relevant abstracts. We make the dataset used to fine-tune the models and the fine-tuned models based on Mistral-7B-instruct-v0.1 and v0.2 publicly available.

How do you know that? Teaching Generative Language Models to Reference Answers to Biomedical Questions

Domain adaptation of Large Language Models (LLMs) leads to models better suited for a particular domain by capturing patterns from domain text which leads to improvements in downstream tasks. To the naked eye, these improvements are visible; however, the patterns are not so. How can we know which patterns and how much they contribute to changes in downstream scores? Through a Multilevel Analysis we discover and quantify the effect of text patterns on downstream scores of domain-adapted Llama 2 for the task of sentence similarity (BIOSSES dataset). We show that text patterns from PubMed abstracts such as clear writing and simplicity, as well as the amount of biomedical information, are the key for improving downstream scores. Also, we show how another factor not usually quantified contributes equally to downstream scores: choice of hyperparameters for both domain adaptation and fine-tuning.

Multilevel Analysis of Biomedical Domain Adaptation of Llama 2: What Matters the Most? A Case Study

The automatic construction of knowledge graphs (KGs) is an important research area in medicine, with far-reaching applications spanning drug discovery and clinical trial design. These applications hinge on the accurate identification of interactions among medical and biological entities. In this study, we propose an end-to-end machine learning solution based on large language models (LLMs) that utilize electronic medical record notes to construct KGs. The entities used in the KG construction process are diseases, factors, treatments, as well as manifestations that coexist with the patient while experiencing the disease. Given the critical need for high-quality performance in medical applications, we embark on a comprehensive assessment of 12 LLMs of various architectures, evaluating their performance and safety attributes. To gauge the quantitative efficacy of our approach by assessing both precision and recall, we manually annotate a dataset provided by the Macula and Retina Institute. We also assess the qualitative performance of LLMs, such as the ability to generate structured outputs or the tendency to hallucinate. The results illustrate that in contrast to encoder-only and encoder-decoder, decoder-only LLMs require further investigation. Additionally, we provide guided prompt design to utilize such LLMs. The application of the proposed methodology is demonstrated on age-related macular degeneration.

Premium content

Irish-based Large Language Model with Extreme Low-Resource Settings in Machine Translation

Downloads

Next from ACL 2024

Improving Self-training with Prototypical Learning for Source-Free Domain Adaptation on Clinical Text

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES