Hong Yu

Han Liu, Xingshuo Huang, Xiaotong Zhang et al. · apple-ml

Decision-based methods have shown to be effective in black-box adversarial attacks, as they can obtain satisfactory performance and only require to access the final model prediction. Gradient estimation is a critical step in black-box adversarial attacks, as it will directly affect the query efficiency. Recent works have attempted to utilize gradient priors to facilitate score-based methods to obtain better results. However, these gradient priors still suffer from the edge gradient discrepancy issue and the successive iteration gradient direction issue, thus are difficult to simply extend to decision-based methods. In this paper, we propose a novel Decision-based Black-box Attack framework with Gradient Priors (DBA-GP), which seamlessly integrates the data-dependent gradient prior and time-dependent prior into the gradient estimation procedure. First, by leveraging the joint bilateral filter to deal with each random perturbation, DBA-GP can guarantee that the generated perturbations in edge locations are hardly smoothed, i.e., alleviating the edge gradient discrepancy, thus remaining the characteristics of the original image as much as possible. Second, by utilizing a new gradient updating strategy to automatically adjust the successive iteration gradient direction, DBA-GP can accelerate the convergence speed, thus improving the query efficiency. Extensive experiments have demonstrated that the proposed method outperforms other strong baselines significantly.

Pengshan Cai, Hui Wan, Fei Liu et al. · ibm-research

We propose novel AI-empowered chat bots for learning as conversation where a user does not read a passage but gains information and knowledge through conversation with a teacher bot. Our information-acquisition-oriented dialogue system employs a novel adaptation of reinforced self-play so that the system can be transferred to various domains without in-domain dialogue data, and can carry out conversations both informative and attentive to users. Our extensive subjective and objective evaluations on three large public data corpora demonstrate the effectiveness of our system to deliver knowledge-intensive and attentive conversations and help end users substantially gain knowledge without reading passages. Our code and datasets are publicly available for follow-up research.

Yiming Zhang, Hong Yu, Ruoyi Du et al.

Audio captioning aims at describing the content of audio clips with human language. Due to the ambiguity of audio, different people may perceive the same audio differently, resulting in caption disparities (i.e., one audio may correlate to several captions with diverse semantics). For that, general audio captioning models achieve the one-to-many training by randomly selecting a correlated caption as the ground truth for each audio. However, it leads to a significant variation in the optimization directions and weakens the model stability. To eliminate this negative effect, in this paper, we propose a two-stage framework for audio captioning: (i) in the first stage, via the contrastive learning, we construct a proxy feature space to reduce the distances between captions correlated to the same audio, and (ii) in the second stage, the proxy feature space is utilized as additional supervision to encourage the model to be optimized in the direction that benefits all the correlated captions. We conducted extensive experiments on two datasets using four commonly used encoder and decoder architectures. Experimental results demonstrate the effectiveness of the proposed method. The code is available at https://github.com/PRIS-CV/Caption-Feature-Space-Regularization.

Avijit Mitra, Richeek Pradhan, Rachel D Melamed et al.

Importance: Social determinants of health (SDOH) are known to be associated with increased risk of suicidal behaviors, but few studies utilized SDOH from unstructured electronic health record (EHR) notes. Objective: To investigate associations between suicide and recent SDOH, identified using structured and unstructured data. Design: Nested case-control study. Setting: EHR data from the US Veterans Health Administration (VHA). Participants: 6,122,785 Veterans who received care in the US VHA between October 1, 2010, and September 30, 2015. Exposures: Occurrence of SDOH over a maximum span of two years compared with no occurrence of SDOH. Main Outcomes and Measures: Cases of suicide deaths were matched with 4 controls on birth year, cohort entry date, sex, and duration of follow-up. We developed an NLP system to extract SDOH from unstructured notes. Structured data, NLP on unstructured data, and combining them yielded six, eight and nine SDOH respectively. Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) were estimated using conditional logistic regression. Results: In our cohort, 8,821 Veterans committed suicide during 23,725,382 person-years of follow-up (incidence rate 37.18/100,000 person-years). Our cohort was mostly male (92.23%) and white (76.99%). Across the five common SDOH as covariates, NLP-extracted SDOH, on average, covered 80.03% of all SDOH occurrences. All SDOH, measured by structured data and NLP, were significantly associated with increased risk of suicide. The SDOH with the largest effects was legal problems (aOR=2.66, 95% CI=.46-2.89), followed by violence (aOR=2.12, 95% CI=1.98-2.27). NLP-extracted and structured SDOH were also associated with suicide. Conclusions and Relevance: NLP-extracted SDOH were always significantly associated with increased risk of suicide among Veterans, suggesting the potential of NLP in public health studies.

Han Liu, Feng Zhang, Xiaotong Zhang et al. · apple-ml

Distribution estimation has been demonstrated as one of the most effective approaches in dealing with few-shot image classification, as the low-level patterns and underlying representations can be easily transferred across different tasks in computer vision domain. However, directly applying this approach to few-shot text classification is challenging, since leveraging the statistics of known classes with sufficient samples to calibrate the distributions of novel classes may cause negative effects due to serious category difference in text domain. To alleviate this issue, we propose two simple yet effective strategies to estimate the distributions of the novel classes by utilizing unlabeled query samples, thus avoiding the potential negative transfer issue. Specifically, we first assume a class or sample follows the Gaussian distribution, and use the original support set and the nearest few query samples to estimate the corresponding mean and covariance. Then, we augment the labeled samples by sampling from the estimated distribution, which can provide sufficient supervision for training the classification model. Extensive experiments on eight few-shot text classification datasets show that the proposed method outperforms state-of-the-art baselines significantly.

Zhichao Yang, Sunjae Kwon, Zonghai Yao et al.

Automatic International Classification of Diseases (ICD) coding aims to assign multiple ICD codes to a medical note with an average of 3,000+ tokens. This task is challenging due to the high-dimensional space of multi-label assignment (155,000+ ICD code candidates) and the long-tail challenge - Many ICD codes are infrequently assigned yet infrequent ICD codes are important clinically. This study addresses the long-tail challenge by transforming this multi-label classification task into an autoregressive generation task. Specifically, we first introduce a novel pretraining objective to generate free text diagnoses and procedure using the SOAP structure, the medical logic physicians use for note documentation. Second, instead of directly predicting the high dimensional space of ICD codes, our model generates the lower dimension of text descriptions, which then infer ICD codes. Third, we designed a novel prompt template for multi-label classification. We evaluate our Generation with Prompt model with the benchmark of all code assignment (MIMIC-III-full) and few shot ICD code assignment evaluation benchmark (MIMIC-III-few). Experiments on MIMIC-III-few show that our model performs with a marco F1 30.2, which substantially outperforms the previous MIMIC-III-full SOTA model (marco F1 4.3) and the model specifically designed for few/zero shot setting (marco F1 18.7). Finally, we design a novel ensemble learner, a cross attention reranker with prompts, to integrate previous SOTA and our best few-shot coding predictions. Experiments on MIMIC-III-full show that our ensemble learner substantially improves both macro and micro F1, from 10.4 to 14.6 and from 58.2 to 59.1, respectively.

Bhanu Pratap Singh Rawat, Samuel Kovaly, Wilfred R. Pigeon et al.

Suicide is an important public health concern and one of the leading causes of death worldwide. Suicidal behaviors, including suicide attempts (SA) and suicide ideations (SI), are leading risk factors for death by suicide. Information related to patients' previous and current SA and SI are frequently documented in the electronic health record (EHR) notes. Accurate detection of such documentation may help improve surveillance and predictions of patients' suicidal behaviors and alert medical professionals for suicide prevention efforts. In this study, we first built Suicide Attempt and Ideation Events (ScAN) dataset, a subset of the publicly available MIMIC III dataset spanning over 12k+ EHR notes with 19k+ annotated SA and SI events information. The annotations also contain attributes such as method of suicide attempt. We also provide a strong baseline model ScANER (Suicide Attempt and Ideation Events Retriever), a multi-task RoBERTa-based model with a retrieval module to extract all the relevant suicidal behavioral evidences from EHR notes of an hospital-stay and, and a prediction module to identify the type of suicidal behavior (SA and SI) concluded during the patient's stay at the hospital. ScANER achieved a macro-weighted F1-score of 0.83 for identifying suicidal behavioral evidences and a macro F1-score of 0.78 and 0.60 for classification of SA and SI for the patient's hospital-stay, respectively. ScAN and ScANER are publicly available.

Zonghai Yao, Yi Cao, Zhichao Yang et al.

Pretrained language models (PLMs) have motivated research on what kinds of knowledge these models learn. Fill-in-the-blanks problem (e.g., cloze tests) is a natural approach for gauging such knowledge. BioLAMA generates prompts for biomedical factual knowledge triples and uses the Top-k accuracy metric to evaluate different PLMs' knowledge. However, existing research has shown that such prompt-based knowledge probing methods can only probe a lower bound of knowledge. Many factors like prompt-based probing biases make the LAMA benchmark unreliable and unstable. This problem is more prominent in BioLAMA. The severe long-tailed distribution in vocabulary and large-N-M relation make the performance gap between LAMA and BioLAMA remain notable. To address these, we introduce context variance into the prompt generation and propose a new rank-change-based evaluation metric. Different from the previous known-unknown evaluation criteria, we propose the concept of "Misunderstand" in LAMA for the first time. Through experiments on 12 PLMs, our context variance prompts and Understand-Confuse-Misunderstand (UCM) metric makes BioLAMA more friendly to large-N-M relations and rare relations. We also conducted a set of control experiments to disentangle "understand" from just "read and copy".

Ye Wang, Jingbo Liao, Hong Yu et al.

Currently end-to-end deep learning based open-domain dialogue systems remain black box models, making it easy to generate irrelevant contents with data-driven models. Specifically, latent variables are highly entangled with different semantics in the latent space due to the lack of priori knowledge to guide the training. To address this problem, this paper proposes to harness the generative model with a priori knowledge through a cognitive approach involving mesoscopic scale feature disentanglement. Particularly, the model integrates the macro-level guided-category knowledge and micro-level open-domain dialogue data for the training, leveraging the priori knowledge into the latent space, which enables the model to disentangle the latent variables within the mesoscopic scale. Besides, we propose a new metric for open-domain dialogues, which can objectively evaluate the interpretability of the latent space distribution. Finally, we validate our model on different datasets and experimentally demonstrate that our model is able to generate higher quality and more interpretable dialogues than other models.

Han Liu, Feng Zhang, Xiaotong Zhang et al.

Multi-label aspect category detection allows a given review sentence to contain multiple aspect categories, which is shown to be more practical in sentiment analysis and attracting increasing attention. As annotating large amounts of data is time-consuming and labor-intensive, data scarcity occurs frequently in real-world scenarios, which motivates multi-label few-shot aspect category detection. However, research on this problem is still in infancy and few methods are available. In this paper, we propose a novel label-enhanced prototypical network (LPN) for multi-label few-shot aspect category detection. The highlights of LPN can be summarized as follows. First, it leverages label description as auxiliary knowledge to learn more discriminative prototypes, which can retain aspect-relevant information while eliminating the harmful effect caused by irrelevant aspects. Second, it integrates with contrastive learning, which encourages that the sentences with the same aspect label are pulled together in embedding space while simultaneously pushing apart the sentences with different aspect labels. In addition, it introduces an adaptive multi-label inference module to predict the aspect count in the sentence, which is simple yet effective. Extensive experimental results on three datasets demonstrate that our proposed model LPN can consistently achieve state-of-the-art performance.

Hieu Tran, Zhichao Yang, Zonghai Yao et al.

To enhance the performance of large language models (LLMs) in biomedical natural language processing (BioNLP) by introducing a domain-specific instruction dataset and examining its impact when combined with multi-task learning principles. We created the BioInstruct, comprising 25,005 instructions to instruction-tune LLMs(LLaMA 1 & 2, 7B & 13B version). The instructions were created by prompting the GPT-4 language model with three-seed samples randomly drawn from an 80 human curated instructions. We employed Low-Rank Adaptation(LoRA) for parameter-efficient fine-tuning. We then evaluated these instruction-tuned LLMs on several BioNLP tasks, which can be grouped into three major categories: question answering(QA), information extraction(IE), and text generation(GEN). We also examined whether categories(e.g., QA, IE, and generation) of instructions impact model performance. Comparing with LLMs without instruction-tuned, our instruction-tuned LLMs demonstrated marked performance gains: 17.3% in QA, 5.7% in IE, and 96% in Generation tasks. Our 7B-parameter instruction-tuned LLaMA 1 model was competitive or even surpassed other LLMs in the biomedical domain that were also fine-tuned from LLaMA 1 with vast domain-specific data or a variety of tasks. Our results also show that the performance gain is significantly higher when instruction fine-tuning is conducted with closely related tasks. Our findings align with the observations of multi-task learning, suggesting the synergies between two tasks. The BioInstruct dataset serves as a valuable resource and instruction tuned LLMs lead to the best performing BioNLP applications.

Minhan Xu, Yu Hong

We leverage embedding duplication between aligned sub-words to extend the Parent-Child transfer learning method, so as to improve low-resource machine translation. We conduct experiments on benchmark datasets of My-En, Id-En and Tr-En translation scenarios. The test results show that our method produces substantial improvements, achieving the BLEU scores of 22.5, 28.0 and 18.1 respectively. In addition, the method is computationally efficient which reduces the consumption of training time by 63.8%, reaching the duration of 1.6 hours when training on a Tesla 16GB P100 GPU. All the models and source codes in the experiments will be made publicly available to support reproducible research.

Han Liu, Siyang Zhao, Xiaotong Zhang et al.

Zero-shot intent classification is a vital and challenging task in dialogue systems, which aims to deal with numerous fast-emerging unacquainted intents without annotated training data. To obtain more satisfactory performance, the crucial points lie in two aspects: extracting better utterance features and strengthening the model generalization ability. In this paper, we propose a simple yet effective meta-learning paradigm for zero-shot intent classification. To learn better semantic representations for utterances, we introduce a new mixture attention mechanism, which encodes the pertinent word occurrence patterns by leveraging the distributional signature attention and multi-layer perceptron attention simultaneously. To strengthen the transfer ability of the model from seen classes to unseen classes, we reformulate zero-shot intent classification with a meta-learning strategy, which trains the model by simulating multiple zero-shot classification tasks on seen categories, and promotes the model generalization ability with a meta-adapting procedure on mimic unseen categories. Extensive experiments on two real-world dialogue datasets in different languages show that our model outperforms other strong baselines on both standard and generalized zero-shot intent classification tasks.

Zhichao Yang, Weisong Liu, Dan Berlowitz et al.

Question: Can an encoder-decoder architecture pretrained on a large dataset of longitudinal electronic health records improves patient outcome predictions? Findings: In this prognostic study of 6.8 million patients, our denoising sequence-to-sequence prediction model of multiple outcomes outperformed state-of-the-art models scuh pretrained BERT on a broad range of patient outcomes, including intentional self-harm and pancreatic cancer. Meaning: Deep bidirectional and autoregressive representation improves patient outcome prediction.

Cecilia Aas, Hisham Abdelsalam, Irina Belousova et al.

It has recently become feasible to run personal digital assistants on phones and other personal devices. In this paper we describe a design for a natural language understanding system that runs on device. In comparison to a server-based assistant, this system is more private, more reliable, faster, more expressive, and more accurate. We describe what led to key choices about architecture and technologies. For example, some approaches in the dialog systems literature are difficult to maintain over time in a deployment setting. We hope that sharing learnings from our practical experiences may help inform future work in the research community.

Zihao Zhang, Zonghai Yao, Huixue Zhou et al.

Large language models have shown success as a tutor in education in various fields. Educating patients about their clinical visits plays a pivotal role in patients' adherence to their treatment plans post-discharge. This paper presents EHRTutor, an innovative multi-component framework leveraging the Large Language Model (LLM) for patient education through conversational question-answering. EHRTutor first formulates questions pertaining to the electronic health record discharge instructions. It then educates the patient through conversation by administering each question as a test. Finally, it generates a summary at the end of the conversation. Evaluation results using LLMs and domain experts have shown a clear preference for EHRTutor over the baseline. Moreover, EHRTutor also offers a framework for generating synthetic patient education dialogues that can be used for future in-house system training.

Junda Wang, Zhichao Yang, Zonghai Yao et al.

Large Language Models (LLMs) have demonstrated a remarkable potential in medical knowledge acquisition and question-answering. However, LLMs can potentially hallucinate and yield factually incorrect outcomes, even with domain-specific pretraining. Previously, retrieval augmented generation (RAG) has limited success in addressing hallucinations. Unlike previous methods in RAG where the retrieval model was trained separately from the LLM, we introduce JMLR (for Jointly trains LLM and information Retrieval) during the fine-tuning phase. The synchronized training mechanism enhances JMLR's ability to retrieve clinical guidelines and leverage medical knowledge to reason and answer questions and reduces the demand for computational resources. We evaluated JMLR on the important medical question-answering application. Our experimental results demonstrate that JMLR-13B (70.5%) outperforms a previous state-of-the-art open-source model using conventional pre-training and fine-tuning Meditron-70B (68.9%) and Llama2-13B with RAG (67.7%) on a medical question-answering dataset. Comprehensive evaluations reveal JMLR-13B enhances reasoning quality and reduces hallucinations better than Claude3-Opus. Additionally, JMLR-13B (148 GPU hours) also trains much faster than Meditron-70B (42630 GPU hours). Through this work, we provide a new and efficient knowledge enhancement method for healthcare, demonstrating the potential of integrating retrieval and LLM training for medical question-answering systems.

Sunjae Kwon, Zhichao Yang, Hong Yu

In this paper, we introduce a comprehensive framework for developing a machine learning-based SOAP (Subjective, Objective, Assessment, and Plan) classification system without manually SOAP annotated training data or with less manually SOAP annotated training data. The system is composed of the following two parts: 1) Data construction, 2) A neural network-based SOAP classifier, and 3) Transfer learning framework. In data construction, since a manual construction of a large size training dataset is expensive, we propose a rule-based weak labeling method utilizing the structured information of an EHR note. Then, we present a SOAP classifier composed of a pre-trained language model and bi-directional long-short term memory with conditional random field (Bi-LSTM-CRF). Finally, we propose a transfer learning framework that re-uses the trained parameters of the SOAP classifier trained with the weakly labeled dataset for datasets collected from another hospital. The proposed weakly label-based learning model successfully performed SOAP classification (89.99 F1-score) on the notes collected from the target hospital. Otherwise, in the notes collected from other hospitals and departments, the performance dramatically decreased. Meanwhile, we verified that the transfer learning framework is advantageous for inter-hospital adaptation of the model increasing the models' performance in every cases. In particular, the transfer learning approach was more efficient when the manually annotated data size was smaller. We showed that SOAP classification models trained with our weakly labeling algorithm can perform SOAP classification without manually annotated data on the EHR notes from the same hospital. The transfer learning framework helps SOAP classification model's inter-hospital migration with a minimal size of the manually annotated dataset.

Shruti Bhargava, Anand Dhoot, Ing-Marie Jonsson et al.

Voice assistants help users make phone calls, send messages, create events, navigate, and do a lot more. However, assistants have limited capacity to understand their users' context. In this work, we aim to take a step in this direction. Our work dives into a new experience for users to refer to phone numbers, addresses, email addresses, URLs, and dates on their phone screens. Our focus lies in reference understanding, which becomes particularly interesting when multiple similar texts are present on screen, similar to visual grounding. We collect a dataset and propose a lightweight general-purpose model for this novel experience. Due to the high cost of consuming pixels directly, our system is designed to rely on the extracted text from the UI. Our model is modular, thus offering flexibility, improved interpretability, and efficient runtime memory utilization.

Hieu Tran, Zonghai Yao, Junda Wang et al.

This work introduces RARE (Retrieval-Augmented Reasoning Enhancement), a versatile extension to the mutual reasoning framework (rStar), aimed at enhancing reasoning accuracy and factual integrity across large language models (LLMs) for complex, knowledge-intensive tasks such as commonsense and medical reasoning. RARE incorporates two innovative actions within the Monte Carlo Tree Search (MCTS) framework: A6, which generates search queries based on the initial problem statement, performs information retrieval using those queries, and augments reasoning with the retrieved data to formulate the final answer; and A7, which leverages information retrieval specifically for generated sub-questions and re-answers these sub-questions with the relevant contextual information. Additionally, a Retrieval-Augmented Factuality Scorer is proposed to replace the original discriminator, prioritizing reasoning paths that meet high standards of factuality. Experimental results with LLaMA 3.1 show that RARE enables open-source LLMs to achieve competitive performance with top open-source models like GPT-4 and GPT-4o. This research establishes RARE as a scalable solution for improving LLMs in domains where logical coherence and factual integrity are critical.

Yu Hong, Qian Liu, Huayuan Cheng et al.

The past few years have witnessed the rapid development of vision-centric 3D perception in autonomous driving. Although the 3D perception models share many structural and conceptual similarities, there still exist gaps in their feature representations, data formats, and objectives, posing challenges for unified and efficient 3D perception framework design. In this paper, we present UniVision, a simple and efficient framework that unifies two major tasks in vision-centric 3D perception, \ie, occupancy prediction and object detection. Specifically, we propose an explicit-implicit view transform module for complementary 2D-3D feature transformation. We propose a local-global feature extraction and fusion module for efficient and adaptive voxel and BEV feature extraction, enhancement, and interaction. Further, we propose a joint occupancy-detection data augmentation strategy and a progressive loss weight adjustment strategy which enables the efficiency and stability of the multi-task framework training. We conduct extensive experiments for different perception tasks on four public benchmarks, including nuScenes LiDAR segmentation, nuScenes detection, OpenOccupancy, and Occ3D. UniVision achieves state-of-the-art results with +1.5 mIoU, +1.8 NDS, +1.5 mIoU, and +1.8 mIoU gains on each benchmark, respectively. We believe that the UniVision framework can serve as a high-performance baseline for the unified vision-centric 3D perception task. The code will be available at \url{https://github.com/Cc-Hy/UniVision}.

Jiangyun Li, Wenxuan Wang, Chen Chen et al.

Transformer, benefiting from global (long-range) information modeling using self-attention mechanism, has been successful in natural language processing and computer vision recently. Convolutional Neural Networks, capable of capturing local features, are difficult to model explicit long-distance dependencies from global feature space. However, both local and global features are crucial for dense prediction tasks, especially for 3D medical image segmentation. In this paper, we present the further attempt to exploit Transformer in 3D CNN for 3D medical image volumetric segmentation and propose a novel network named TransBTSV2 based on the encoder-decoder structure. Different from TransBTS, the proposed TransBTSV2 is not limited to brain tumor segmentation (BTS) but focuses on general medical image segmentation, providing a stronger and more efficient 3D baseline for volumetric segmentation of medical images. As a hybrid CNN-Transformer architecture, TransBTSV2 can achieve accurate segmentation of medical images without any pre-training, possessing the strong inductive bias as CNNs and powerful global context modeling ability as Transformer. With the proposed insight to redesign the internal structure of Transformer block and the introduced Deformable Bottleneck Module to capture shape-aware local details, a highly efficient architecture is achieved with superior performance. Extensive experimental results on four medical image datasets (BraTS 2019, BraTS 2020, LiTS 2017 and KiTS 2019) demonstrate that TransBTSV2 achieves comparable or better results compared to the state-of-the-art methods for the segmentation of brain tumor, liver tumor as well as kidney tumor. Code will be publicly available at https://github.com/Wenxuan-1119/TransBTS.

Han Liu, Zhi Xu, Xiaotong Zhang et al.

Black-box hard-label adversarial attack on text is a practical and challenging task, as the text data space is inherently discrete and non-differentiable, and only the predicted label is accessible. Research on this problem is still in the embryonic stage and only a few methods are available. Nevertheless, existing methods rely on the complex heuristic algorithm or unreliable gradient estimation strategy, which probably fall into the local optimum and inevitably consume numerous queries, thus are difficult to craft satisfactory adversarial examples with high semantic similarity and low perturbation rate in a limited query budget. To alleviate above issues, we propose a simple yet effective framework to generate high quality textual adversarial examples under the black-box hard-label attack scenarios, named HQA-Attack. Specifically, after initializing an adversarial example randomly, HQA-attack first constantly substitutes original words back as many as possible, thus shrinking the perturbation rate. Then it leverages the synonym set of the remaining changed words to further optimize the adversarial example with the direction which can improve the semantic similarity and satisfy the adversarial condition simultaneously. In addition, during the optimizing procedure, it searches a transition synonym word for each changed word, thus avoiding traversing the whole synonym set and reducing the query number to some extent. Extensive experimental results on five text classification datasets, three natural language inference datasets and two real-world APIs have shown that the proposed HQA-Attack method outperforms other strong baselines significantly.

Rumeng Li, Xun Wang, Hong Yu

To address the limitations of Large Language Models (LLMs) in the International Classification of Diseases (ICD) coding task, where they often produce inaccurate and incomplete prediction results due to the high-dimensional and skewed distribution of the ICD codes, and often lack interpretability and reliability as well. We introduce an innovative multi-agent approach for ICD coding which mimics the ICD coding assignment procedure in real-world settings, comprising five distinct agents: the patient, physician, coder, reviewer, and adjuster. Each agent utilizes an LLM-based model tailored to their specific role within the coding process. We also integrate the system with Electronic Health Record (HER)'s SOAP (subjective, objective, assessment and plan) structure to boost the performances. We compare our method with a system of agents designed solely by LLMs and other strong baselines and evaluate it using the Medical Information Mart for Intensive Care III (MIMIC-III) dataset. Our multi-agent coding framework significantly outperforms Zero-shot Chain of Thought (CoT) prompting and self-consistency with CoT (CoT-SC) in coding common and rare ICD codes. An ablation study validates the effectiveness of the designated agent roles. it also outperforms the LLM-designed agent system. Moreover, our method achieves comparable results to state-of-the-art ICD coding methods that require extensive pre-training or fine-tuning, and outperforms them in rare code accuracy, and explainability. Additionally, we demonstrate the method's practical applicability by presenting its performance in scenarios not limited by the common or rare ICD code constraints.The proposed multi-agent method for ICD coding effectively mimics the real-world coding process and improves performance on both common and rare codes.

Atharva Kulkarni, Yuan Zhang, Joel Ruben Antony Moniz et al. · cmu

Hallucinations pose a significant obstacle to the reliability and widespread adoption of language models, yet their accurate measurement remains a persistent challenge. While many task- and domain-specific metrics have been proposed to assess faithfulness and factuality concerns, the robustness and generalization of these metrics are still untested. In this paper, we conduct a large-scale empirical evaluation of 6 diverse sets of hallucination detection metrics across 4 datasets, 37 language models from 5 families, and 5 decoding methods. Our extensive investigation reveals concerning gaps in current hallucination evaluation: metrics often fail to align with human judgments, take an overtly myopic view of the problem, and show inconsistent gains with parameter scaling. Encouragingly, LLM-based evaluation, particularly with GPT-4, yields the best overall results, and mode-seeking decoding methods seem to reduce hallucinations, especially in knowledge-grounded settings. These findings underscore the need for more robust metrics to understand and quantify hallucinations, and better strategies to mitigate them.

Hieu Tran, Zonghai Yao, Hong Yu

Reinforcement learning improves LLM reasoning, yet sparse delayed reward over long sequences makes token-level credit assignment the key bottleneck. We study the verifiable-reward setting, where the final answer is checkable and multiple responses can be drawn per prompt. Reasoning tasks in math and medical QA align with this setup, where only a few decision tokens significantly impact the outcome. PPO offers token-level advantages with a learned value model, but it is complex to train both the actor and critic models simultaneously, and it is not easily generalizable, as the token-level values from the critic model can make training prone to overfitting. GRPO is critic-free and supports verifiable rewards, but spreads a single sequence-level return across tokens and ignores branching. We introduce \textbf{Prefix-to-Tree (P2T)}, a simple procedure that converts a group of responses into a prefix tree and computes \emph{nonparametric} prefix values \(V(s)\) by aggregating descendant outcomes. Built on P2T, we propose \textbf{TEMPO} (\emph{\textbf{T}ree-\textbf{E}stimated \textbf{M}ean Prefix Value for \textbf{P}olicy \textbf{O}ptimization}), a critic-free algorithm that augments the group-relative outcome signal of GRPO with \emph{branch-gated} temporal-difference corrections derived from the tree. At non-branch tokens, the temporal-difference (TD) term is zero, so TEMPO reduces to GRPO; at branching tokens, it supplies precise token-level credit without a learned value network or extra judges/teachers. On Qwen3-1.7B/4B, TEMPO outperforms PPO and GRPO on in-distribution (MATH, MedQA) and out-of-distribution (GSM-HARD, AMC23, MedMCQA, MMLU-Medical) benchmarks, and reaches higher validation accuracy with roughly the same wall-clock time.

Vijeta Deshpande, Minhwa Lee, Zonghai Yao et al.

Prior research on Twitter (now X) data has provided positive evidence of its utility in developing supplementary health surveillance systems. In this study, we present a new framework to surveil public health, focusing on mental health (MH) outcomes. We hypothesize that locally posted tweets are indicative of local MH outcomes and collect tweets posted from 765 neighborhoods (census block groups) in the USA. We pair these tweets from each neighborhood with the corresponding MH outcome reported by the Center for Disease Control (CDC) to create a benchmark dataset, LocalTweets. With LocalTweets, we present the first population-level evaluation task for Twitter-based MH surveillance systems. We then develop an efficient and effective method, LocalHealth, for predicting MH outcomes based on LocalTweets. When used with GPT3.5, LocalHealth achieves the highest F1-score and accuracy of 0.7429 and 79.78\%, respectively, a 59\% improvement in F1-score over the GPT3.5 in zero-shot setting. We also utilize LocalHealth to extrapolate CDC's estimates to proxy unreported neighborhoods, achieving an F1-score of 0.7291. Our work suggests that Twitter data can be effectively leveraged to simulate neighborhood-level MH outcomes.

Won Seok Jang, Hieu Tran, Manav Mistry et al.

Patients must possess the knowledge necessary to actively participate in their care. We present NoteAid-Chatbot, a conversational AI that promotes patient understanding via a novel 'learning as conversation' framework, built on a multi-agent large language model (LLM) and reinforcement learning (RL) setup without human-labeled data. NoteAid-Chatbot was built on a lightweight LLaMA 3.2 3B model trained in two stages: initial supervised fine-tuning on conversational data synthetically generated using medical conversation strategies, followed by RL with rewards derived from patient understanding assessments in simulated hospital discharge scenarios. Our evaluation, which includes comprehensive human-aligned assessments and case studies, demonstrates that NoteAid-Chatbot exhibits key emergent behaviors critical for patient education, such as clarity, relevance, and structured dialogue, even though it received no explicit supervision for these attributes. Our results show that even simple Proximal Policy Optimization (PPO)-based reward modeling can successfully train lightweight, domain-specific chatbots to handle multi-turn interactions, incorporate diverse educational strategies, and meet nuanced communication objectives. Our Turing test demonstrates that NoteAid-Chatbot surpasses non-expert human. Although our current focus is on healthcare, the framework we present illustrates the feasibility and promise of applying low-cost, PPO-based RL to realistic, open-ended conversational domains, broadening the applicability of RL-based alignment methods.

Zhifeng Li, Yifan Fan, Bowei Zou et al.

Leveraging external knowledge to enhance the reasoning ability is crucial for commonsense question answering. However, the existing knowledge bases heavily rely on manual annotation which unavoidably causes deficiency in coverage of world-wide commonsense knowledge. Accordingly, the knowledge bases fail to be flexible enough to support the reasoning over diverse questions. Recently, large-scale language models (LLMs) have dramatically improved the intelligence in capturing and leveraging knowledge, which opens up a new way to address the issue of eliciting knowledge from language models. We propose a Unified Facts Obtaining (UFO) approach. UFO turns LLMs into knowledge sources and produces relevant facts (knowledge statements) for the given question. We first develop a unified prompt consisting of demonstrations that cover different aspects of commonsense and different question styles. On this basis, we instruct the LLMs to generate question-related supporting facts for various commonsense questions via prompting. After facts generation, we apply a dense retrieval-based fact selection strategy to choose the best-matched fact. This kind of facts will be fed into the answer inference model along with the question. Notably, due to the design of unified prompts, UFO can support reasoning in various commonsense aspects (including general commonsense, scientific commonsense, and social commonsense). Extensive experiments on CommonsenseQA 2.0, OpenBookQA, QASC, and Social IQA benchmarks show that UFO significantly improves the performance of the inference model and outperforms manually constructed knowledge sources.

Shilin Zhou, Zhenghua Li, Yu Hong et al.

End-to-end automatic speech recognition (ASR) systems have made significant progress in general scenarios. However, it remains challenging to transcribe contextual named entities (NEs) in the contextual ASR scenario. Previous approaches have attempted to address this by utilizing the NE dictionary. These approaches treat entities as individual tokens and generate them token-by-token, which may result in incomplete transcriptions of entities. In this paper, we treat entities as indivisible wholes and introduce the idea of copying into ASR. We design a systematic mechanism called CopyNE, which can copy entities from the NE dictionary. By copying all tokens of an entity at once, we can reduce errors during entity transcription, ensuring the completeness of the entity. Experiments demonstrate that CopyNE consistently improves the accuracy of transcribing entities compared to previous approaches. Even when based on the strong Whisper, CopyNE still achieves notable improvements.

Haw-Shiuan Chang, Zonghai Yao, Alolika Gon et al.

Is the output softmax layer, which is adopted by most language models (LMs), always the best way to compute the next word probability? Given so many attention layers in a modern transformer-based LM, are the pointer networks redundant nowadays? In this study, we discover that the answers to both questions are no. This is because the softmax bottleneck sometimes prevents the LMs from predicting the desired distribution and the pointer networks can be used to break the bottleneck efficiently. Based on the finding, we propose several softmax alternatives by simplifying the pointer networks and accelerating the word-by-word rerankers. In GPT-2, our proposals are significantly better and more efficient than mixture of softmax, a state-of-the-art softmax alternative. In summarization experiments, without significantly decreasing its training/testing speed, our best method based on T5-Small improves factCC score by 2 points in CNN/DM and XSUM dataset, and improves MAUVE scores by 30% in BookSum paragraph-level dataset.

Yanling Li, Bowei Zou, Yifan Fan et al.

We tackle Multi-party Dialogue Reading Comprehension (abbr., MDRC). MDRC stands for an extractive reading comprehension task grounded on a batch of dialogues among multiple interlocutors. It is challenging due to the requirement of understanding cross-utterance contexts and relationships in a multi-turn multi-party conversation. Previous studies have made great efforts on the utterance profiling of a single interlocutor and graph-based interaction modeling. The corresponding solutions contribute to the answer-oriented reasoning on a series of well-organized and thread-aware conversational contexts. However, the current MDRC models still suffer from two bottlenecks. On the one hand, a pronoun like "it" most probably produces multi-skip reasoning throughout the utterances of different interlocutors. On the other hand, an MDRC encoder is potentially puzzled by fuzzy features, i.e., the mixture of inner linguistic features in utterances and external interactive features among utterances. To overcome the bottlenecks, we propose a coreference-aware attention modeling method to strengthen the reasoning ability. In addition, we construct a two-channel encoding network. It separately encodes utterance profiles and interactive relationships, so as to relieve the confusion among heterogeneous features. We experiment on the benchmark corpora Molweni and FriendsQA. Experimental results demonstrate that our approach yields substantial improvements on both corpora, compared to the fine-tuned BERT and ELECTRA baselines. The maximum performance gain is about 2.5\% F1-score. Besides, our MDRC models outperform the state-of-the-art in most cases.

Zhifeng Li, Bowei Zou, Yifan Fan et al.

Generative commonsense question answering (GenCQA) is a task of automatically generating a list of answers given a question. The answer list is required to cover all reasonable answers. This presents the considerable challenges of producing diverse answers and ranking them properly. Incorporating a variety of closely-related background knowledge into the encoding of questions enables the generation of different answers. Meanwhile, learning to distinguish positive answers from negative ones potentially enhances the probabilistic estimation of plausibility, and accordingly, the plausibility-based ranking. Therefore, we propose a Knowledge Enhancement and Plausibility Ranking (KEPR) approach grounded on the Generate-Then-Rank pipeline architecture. Specifically, we expand questions in terms of Wiktionary commonsense knowledge of keywords, and reformulate them with normalized patterns. Dense passage retrieval is utilized for capturing relevant knowledge, and different PLM-based (BART, GPT2 and T5) networks are used for generating answers. On the other hand, we develop an ELECTRA-based answer ranking model, where logistic regression is conducted during training, with the aim of approximating different levels of plausibility in a polar classification scenario. Extensive experiments on the benchmark ProtoQA show that KEPR obtains substantial improvements, compared to the strong baselines. Within the experimental models, the T5-based GenCQA with KEPR obtains the best performance, which is up to 60.91% at the primary canonical metric Inc@3. It outperforms the existing GenCQA models on the current leaderboard of ProtoQA.

Sunjae Kwon, Rishabh Garodia, Minhwa Lee et al.

Visual Word Sense Disambiguation (VWSD) is a task to find the image that most accurately depicts the correct sense of the target word for the given context. Previously, image-text matching models often suffered from recognizing polysemous words. This paper introduces an unsupervised VWSD approach that uses gloss information of an external lexical knowledge-base, especially the sense definitions. Specifically, we suggest employing Bayesian inference to incorporate the sense definitions when sense information of the answer is not provided. In addition, to ameliorate the out-of-dictionary (OOD) issue, we propose a context-aware definition generation with GPT-3. Experimental results show that the VWSD performance significantly increased with our Bayesian inference-based approach. In addition, our context-aware definition generation achieved prominent performance improvement in OOD examples exhibiting better performance than the existing definition generation method.

Hongyu Zhu, Yan Chen, Jing Yan et al.

In this paper, we focus on studying robustness evaluation of Chinese question matching. Most of the previous work on analyzing robustness issue focus on just one or a few types of artificial adversarial examples. Instead, we argue that it is necessary to formulate a comprehensive evaluation about the linguistic capabilities of models on natural texts. For this purpose, we create a Chinese dataset namely DuQM which contains natural questions with linguistic perturbations to evaluate the robustness of question matching models. DuQM contains 3 categories and 13 subcategories with 32 linguistic perturbations. The extensive experiments demonstrate that DuQM has a better ability to distinguish different models. Importantly, the detailed breakdown of evaluation by linguistic phenomenon in DuQM helps us easily diagnose the strength and weakness of different models. Additionally, our experiment results show that the effect of artificial adversarial examples does not work on the natural texts.

Shilin Zhou, Qingrong Xia, Zhenghua Li et al.

This paper proposes to cast end-to-end span-based SRL as a word-based graph parsing task. The major challenge is how to represent spans at the word level. Borrowing ideas from research on Chinese word segmentation and named entity recognition, we propose and compare four different schemata of graph representation, i.e., BES, BE, BIES, and BII, among which we find that the BES schema performs the best. We further gain interesting insights through detailed analysis. Moreover, we propose a simple constrained Viterbi procedure to ensure the legality of the output graph according to the constraints of the SRL structure. We conduct experiments on two widely used benchmark datasets, i.e., CoNLL05 and CoNLL12. Results show that our word-based graph parsing approach achieves consistently better performance than previous results, under all settings of end-to-end and predicate-given, without and with pre-trained language models (PLMs). More importantly, our model can parse 669/252 sentences per second, without and with PLMs respectively.

Tsendsuren Munkhdalai, Hong Yu

Neural networks have been successfully applied in applications with a large amount of labeled data. However, the task of rapid generalization on new concepts with small training data while preserving performances on previously learned ones still presents a significant challenge to neural network models. In this work, we introduce a novel meta learning method, Meta Networks (MetaNet), that learns a meta-level knowledge across tasks and shifts its inductive biases via fast parameterization for rapid generalization. When evaluated on Omniglot and Mini-ImageNet benchmarks, our MetaNet models achieve a near human-level performance and outperform the baseline approaches by up to 6% accuracy. We demonstrate several appealing properties of MetaNet relating to generalization and continual learning.

Xi'ao Ma, Guoyin Wang, Hong Yu

Attribute reduction is one of the most important topics in rough set theory. Heuristic attribute reduction algorithms have been presented to solve the attribute reduction problem. It is generally known that fitness functions play a key role in developing heuristic attribute reduction algorithms. The monotonicity of fitness functions can guarantee the validity of heuristic attribute reduction algorithms. In probabilistic rough set model, distribution reducts can ensure the decision rules derived from the reducts are compatible with those derived from the original decision table. However, there are few studies on developing heuristic attribute reduction algorithms for finding distribution reducts. This is partly due to the fact that there are no monotonic fitness functions that are used to design heuristic attribute reduction algorithms in probabilistic rough set model. The main objective of this paper is to develop heuristic attribute reduction algorithms for finding distribution reducts in probabilistic rough set model. For one thing, two monotonic fitness functions are constructed, from which equivalence definitions of distribution reducts can be obtained. For another, two modified monotonic fitness functions are proposed to evaluate the significance of attributes more effectively. On this basis, two heuristic attribute reduction algorithms for finding distribution reducts are developed based on addition-deletion method and deletion method. In particular, the monotonicity of fitness functions guarantees the rationality of the proposed heuristic attribute reduction algorithms. Results of experimental analysis are included to quantify the effectiveness of the proposed fitness functions and distribution reducts.