Su Liu

Qiyang Xie, Jialun Wu, Xinjie He et al.

AI systems increasingly need to combine two demanding capabilities: navigating multi-session conversation history and performing deep reading comprehension within long documents. Yet no existing benchmark evaluates both simultaneously. We introduce MemoryDocDataSet, a synthetic benchmark of 50 micro-worlds and 1,000 QA pairs in which each instance comprises 3-5 personas, a temporal event graph spanning months of activity, 3-5 real long documents (20,000-50,000 tokens each sourced from the Caselaw Access Project), multi-session conversations grounded on those documents, and 20 question-answer pairs across five reasoning categories. The defining feature is the Hybrid source tag: questions requiring a system to first navigate conversation history to identify which document is relevant, then extract the answer from within that document. Hybrid questions account for 75.1% of the dataset. Dataset quality is characterised through a prompt-sensitivity self-consistency analysis using LLM-as-judge, yielding a median Cohen's $κ= 0.634$ across all 50 micro-worlds. We evaluate six baseline configurations spanning truncated context, long-context LLMs, retrieval-augmented generation (RAG), and memory systems. The best baseline (RAG-Both) achieves 0.358 overall F1 and 0.342 on Hybrid. Document-only retrieval (RAG-Doc) collapses to 0.267 on Hybrid despite achieving 0.453 on Doc-only questions, demonstrating a clear joint-retrieval gap that motivates architectures unifying conversational memory with long-document navigation. We release the dataset, generation pipeline, and all baseline implementations.

Shuai Xiao, Su Liu, Weikai Zhou et al.

Persona prompting is widely used to steer large language models, yet its practical value remains unclear. Prior work often evaluates persona prompting using aggregate scores, making it difficult to determine whether expert-role prompting consistently improves response quality or instead changes responses along different quality dimensions. We study this question through a controlled comparison of four prompting conditions across 1,140 open-ended questions spanning 38 expert roles and six domains: no role prompt, a generic domain-expert prompt, embedding-based role retrieval, and a hybrid retrieval method combining embedding search with LLM-based role selection. Aggregate results show only small overall differences between conditions. However, metric-level analysis reveals a consistent tradeoff that aggregate averages obscure: role prompting systematically increases expertise depth while reducing clarity. These effects are highly conditional rather than universal. Role prompting performs best on advisory questions and in domains such as medicine and psychology, where structured expert framing and risk communication are intrinsically valuable. In contrast, baseline prompting performs better on conceptual and explanatory questions in finance, legal, science, and technology domains, where concise plain-language explanation is more important. We further show that hybrid retrieval significantly improves over embedding-only role selection, although better role retrieval does not eliminate the broader expertise-depth versus clarity tradeoff. Overall, our findings suggest that persona prompting primarily reshapes response characteristics rather than broadly improving capability, and that multi-metric evaluation is necessary for understanding its effects.

Xinjie He, Zhiyuan Lin, Su Liu et al.

Reinforcement learning (RL) has emerged as a viable recipe for training LLM agents to reason over external memory banks in multi-session dialogue. Existing work trains exclusively on a single benchmark, leaving open how the composition of training data shapes the skills a memory agent acquires. We present a controlled empirical study that holds architecture, RL algorithm, and all hyperparameters fixed and varies only the training curriculum across three conditions: in-domain (LoCoMo), mixed-benchmark (LoCoMo + LongMemEval), and out-of-domain (LongMemEval only). Across two benchmarks and ten question types, curriculum composition acts as a fine-grained lever on specialization rather than a uniform scaling factor on performance. The mixed curriculum yields the strongest overall F1 on both evaluation sets. Training on a narrow out-of-domain set transfers a targeted skill - temporal reasoning - despite weak aggregate performance. Per-type differences substantially exceed aggregate differences, indicating that single-number benchmark comparisons systematically underreport curriculum effects. We further report two practical lessons from adapting GRPO to a single-GPU regime: cross-benchmark mixing requires filtering format-specific noise from memory banks to preserve training signal, and binary exact-match reward produces no learning signal at the small group sizes (G = 4) required on one GPU, motivating continuous reward functions in this regime.

Ran Tong, Jiaqi Liu, Tong Wang et al.

The accurate interpretation of chest radiographs using automated methods is a critical task in medical imaging. This paper presents a comparative analysis between a supervised lightweight Convolutional Neural Network (CNN) and a state-of-the-art, zero-shot medical Vision-Language Model (VLM), BiomedCLIP, across two distinct diagnostic tasks: pneumonia detection on the PneumoniaMNIST benchmark and tuberculosis detection on the Shenzhen TB dataset. Our experiments show that supervised CNNs serve as highly competitive baselines in both cases. While the default zero-shot performance of the VLM is lower, we demonstrate that its potential can be unlocked via a simple yet crucial remedy: decision threshold calibration. By optimizing the classification threshold on a validation set, the performance of BiomedCLIP is significantly boosted across both datasets. For pneumonia detection, calibration enables the zero-shot VLM to achieve a superior F1-score of 0.8841, surpassing the supervised CNN's 0.8803. For tuberculosis detection, calibration dramatically improves the F1-score from 0.4812 to 0.7684, bringing it close to the supervised baseline's 0.7834. This work highlights a key insight: proper calibration is essential for leveraging the full diagnostic power of zero-shot VLMs, enabling them to match or even outperform efficient, task-specific supervised models.

Ran Tong, Jiaqi Liu, Su Liu et al.

We present a compact, strictly causal benchmark for streaming clinical time series on the MIT--BIH Arrhythmia Database using per-second heart rate. Two tasks are studied under record-level, non-overlapping splits: near-term tachycardia risk (next ten seconds) and one-step heart rate forecasting. We compare a GRU-D (RNN) and a Transformer under matched training budgets against strong non-learned baselines. Evaluation is calibration-aware for classification and proper for forecasting, with temperature scaling and grouped bootstrap confidence intervals. On MIT-BIH, GRU-D slightly surpasses the Transformer for tachycardia risk, while the Transformer clearly lowers forecasting error relative to GRU-D and persistence. Our results show that, in longitudinal monitoring, model choice is task-dependent: compact RNNs remain competitive for short-horizon risk scoring, whereas compact Transformers deliver clearer gains for point forecasting.

Jiaqi Liu, Tong Wang, Su Liu et al.

The research evaluates lightweight medical abstract classification methods to establish their maximum performance capabilities under financial budget restrictions. On the public medical abstracts corpus, we finetune BERT base and Distil BERT with three objectives cross entropy (CE), class weighted CE, and focal loss under identical tokenization, sequence length, optimizer, and schedule. DistilBERT with plain CE gives the strongest raw argmax trade off, while a post hoc operating point selection (validation calibrated, classwise thresholds) sub stantially improves deployed performance; under this tuned regime, focal benefits most. We report Accuracy, Macro F1, and WeightedF1, release evaluation artifacts, and include confusion analyses to clarify error structure. The practical takeaway is to start with a compact encoder and CE, then add lightweight calibration or thresholding when deployment requires higher macro balance.

Jiexi Xu, Jiaqi Liu, Lanruo Wang et al.

Large language model (LLM) agents are increasingly capable of orchestrating complex tasks in low-code environments. However, these agents often exhibit hallucinations and logical inconsistencies because their inherent reasoning mechanisms rely on probabilistic associations rather than genuine causal understanding. This paper introduces a new programming paradigm: Causal-Visual Programming (CVP), designed to address this fundamental issue by explicitly introducing causal structures into the workflow design. CVP allows users to define a simple "world model" for workflow modules through an intuitive low-code interface, effectively creating a Directed Acyclic Graph (DAG) that explicitly defines the causal relationships between modules. This causal graph acts as a crucial constraint during the agent's reasoning process, anchoring its decisions to a user-defined causal structure and significantly reducing logical errors and hallucinations by preventing reliance on spurious correlations. To validate the effectiveness of CVP, we designed a synthetic experiment that simulates a common real-world problem: a distribution shift between the training and test environments. Our results show that a causally anchored model maintained stable accuracy in the face of this shift, whereas a purely associative baseline model that relied on probabilistic correlations experienced a significant performance drop. The primary contributions of this study are: a formal definition of causal structures for workflow modules; the proposal and implementation of a CVP framework that anchors agent reasoning to a user-defined causal graph; and empirical evidence demonstrating the framework's effectiveness in enhancing agent robustness and reducing errors caused by causal confusion in dynamic environments. CVP offers a viable path toward building more interpretable, reliable, and trustworthy AI agents.

Su Liu, Xin Hu, Shurong Wen et al.

Physiologically Based Pharmacokinetic (PBPK) modeling plays a critical role in drug development by predicting drug concentration dynamics across organs. Traditional approaches rely on ordinary differential equations with strong simplifying assumptions, which limit their adaptability to nonlinear physiological interactions. In this study, we explore data-driven alternatives for PBPK prediction using deep learning. Two baseline architectures - a multilayer perceptron (MLP) and a long short-term memory (LSTM) network - are implemented to capture molecular and temporal dependencies, respectively. To incorporate inter-organ interactions, we propose a Dynamic Graph Neural Network (Dynamic GNN) that models physiological connections as recurrent message-passing processes between organs. Experimental results demonstrate that the proposed Dynamic GNN achieves the highest predictive performance among all models, with an R^2 of 0.9342, an RMSE of 0.0159, and an MAE of 0.0116. In comparison, the MLP baseline obtains an R^2 of 0.8705 and the LSTM achieves 0.8059. These results highlight that explicitly modeling the spatial and temporal dependencies of organ interactions enables more accurate and generalizable drug concentration prediction. The Dynamic GNN provides a scalable, equation-free alternative to traditional PBPK formulations and demonstrates strong potential for data-driven pharmacokinetic modeling in preclinical and clinical research.