Search

Machine learning (ML) exhibits promise in the clinical domain. However, it is constrained by data scarcity and ethical considerations, as the generation of clinical trials presents significant challenges due to stringent privacy regulations, high costs, and the extended duration required for conducting studies with human participants. Despite the advancements of large language models (LLMs) in general generation tasks, their potential in facilitating the generation of synthetic clinical trials is under-explored. To address this gap, we introduce a novel Retrieval-Reasoning few-shot framework that leverages LLMs to generate artificial yet realistic and diverse clinical trials with binary success/failure labels. Experiments conducted on real clinical trials from the \url{ClinicalTrials.gov} database demonstrate that our synthetic data can effectively augment real datasets. Furthermore, by fine-tuning a pre-trained model as a binary classifier on synthetic clinical trial datasets, we demonstrate that this augmentation enhances model training for downstream tasks such as trial outcome prediction. Our findings suggest that LLMs for synthetic clinical trial generation hold promise for accelerating clinical research and upholding ethical standards for patient privacy. The code is publicly available at https://anonymous.4open.science/r/Retrieval_Reasoning_Clinical_Trial_Generation-3EC4.

Spontaneous symmetry breaking (SSB) is crucial to the occurrence of phase transitions. Once a phase transition occurs, a quantum system presents degenerate eigenstates that lack the symmetry of the Hamiltonian. After crossing the critical point, the system is essentially evolved to a quantum superposition of these eigenstates until decoherence sets in. Despite the fundamental importance and potential applications in quantum technologies, such quantum-mechanical SSB phenomena have not been experimentally explored in many-body systems. We here present an experimental demonstration of the SSB process in the Lipkin-Meshkov-Glick model, governed by the competition between the individual driving and intra-qubit interaction. The model is realized in a circuit quantum electrodynamics system, where 6 Xmon qubits are coupled in an all-to-all manner through virtual photon exchange mediated by a resonator. The observed nonclassical correlations among these qubits in the symmetry-breaking region go beyond the conventional description of SSB, shedding new light on phase transitions for quantum many-body systems., Comment: 27 pages, 18 figures

Large language models (LLMs) have exhibited exciting progress in multiple scenarios, while the huge computational demands hinder their deployments in lots of real-world applications. As an effective means to reduce memory footprint and inference cost, quantization also faces challenges in performance degradation at low bit-widths. Understanding the impact of quantization on LLM capabilities, especially the generalization ability, is crucial. However, the community's main focus remains on the algorithms and models of quantization, with insufficient attention given to whether the quantized models can retain the strong generalization abilities of LLMs. In this work, we fill this gap by providing a comprehensive benchmark suite for this research topic, including an evaluation system, detailed analyses, and a general toolbox. Specifically, based on the dominant pipeline in LLM quantization, we primarily explore the impact of calibration data distribution on the generalization of quantized LLMs and conduct the benchmark using more than 40 datasets within two main scenarios. Based on this benchmark, we conduct extensive experiments with two well-known LLMs (English and Chinese) and four quantization algorithms to investigate this topic in-depth, yielding several counter-intuitive and valuable findings, e.g., models quantized using a calibration set with the same distribution as the test data are not necessarily optimal. Besides, to facilitate future research, we also release a modular-designed toolbox, which decouples the overall pipeline into several separate components, e.g., base LLM module, dataset module, quantizer module, etc. and allows subsequent researchers to easily assemble their methods through a simple configuration. Our benchmark suite is publicly available at https://github.com/TsingmaoAI/MI-optimize

In MCM/ICM 2023, we proposed a new result prediction model for the popular game Wordle launched by The New York Times. We first preprocessed the raw data and then established a prediction model based on ARIMA to predict the number of report results as of March 1, 2023. We selected word usage frequency, word information entropy, and the number of repeated letters contained in the word as the attributes of the word, and conducted a correlation analysis between these three attributes and the percentage of seven attempts. We also established a regression model based on the XGBoost algorithm, predicted the distribution of reported results, and predicted the correlation percentage of "EERIE". In addition, we also constructed a word classification model that classified words into "simple", "moderate", and "difficult", and explored the relationship between the three attributes and the classification results. Finally, we calculated the percentage of players in the dataset who needed 3 or more attempts for each word. The appendix provides relevant information and problems to be solved for the mathematical modeling competition., Comment: 19 pages, 19 figures, MCM/ICM 2023 Award winning Paper

Artificial intelligence (AI)-driven methods can vastly improve the historically costly drug design process, with various generative models already in widespread use. Generative models for de novo drug design, in particular, focus on the creation of novel biological compounds entirely from scratch, representing a promising future direction. Rapid development in the field, combined with the inherent complexity of the drug design process, creates a difficult landscape for new researchers to enter. In this survey, we organize de novo drug design into two overarching themes: small molecule and protein generation. Within each theme, we identify a variety of subtasks and applications, highlighting important datasets, benchmarks, and model architectures and comparing the performance of top models. We take a broad approach to AI-driven drug design, allowing for both micro-level comparisons of various methods within each subtask and macro-level observations across different fields. We discuss parallel challenges and approaches between the two applications and highlight future directions for AI-driven de novo drug design as a whole. An organized repository of all covered sources is available at https://github.com/gersteinlab/GenAI4Drug.

Semi-supervised learning (SSL) has witnessed great progress with various improvements in the self-training framework with pseudo labeling. The main challenge is how to distinguish high-quality pseudo labels against the confirmation bias. However, existing pseudo-label selection strategies are limited to pre-defined schemes or complex hand-crafted policies specially designed for classification, failing to achieve high-quality labels, fast convergence, and task versatility simultaneously. To these ends, we propose a Semi-supervised Reward framework (SemiReward) that predicts reward scores to evaluate and filter out high-quality pseudo labels, which is pluggable to mainstream SSL methods in wide task types and scenarios. To mitigate confirmation bias, SemiReward is trained online in two stages with a generator model and subsampling strategy. With classification and regression tasks on 13 standard SSL benchmarks across three modalities, extensive experiments verify that SemiReward achieves significant performance gains and faster convergence speeds upon Pseudo Label, FlexMatch, and Free/SoftMatch. Code and models are available at https://github.com/Westlake-AI/SemiReward., Comment: ICLR 2024 Camera Ready. Preprint V2 (25 pages) with the source code at https://github.com/Westlake-AI/SemiReward

The theoretical calculations of the dynamic E2-M1 polarizability at the magic wavelength of the Sr optical clock are inconsistent with experimental results. We investigate role of negative-energy states in the E2 and M1 polarizabilities. Our result for E2-M1 polarizability difference $-$7.74(3.92)$\times$10$^{-5}$ a.u. is dominated by the contribution from negative-energy states to M1 polarizability and has the same sign as and consistent with all the experimental values. In addition, we apply the present calculations to various other optical clocks, further confirming the importance of negative-energy states to the M1 polarizability., Comment: 7 pages, 3 figures, arXiv admin note: this is an extended version of arXiv:2301.06740

In the field of artificial intelligence for science, it is consistently an essential challenge to face a limited amount of labeled data for real-world problems. The prevailing approach is to pretrain a powerful task-agnostic model on a large unlabeled corpus but may struggle to transfer knowledge to downstream tasks. In this study, we propose InstructMol, a semi-supervised learning algorithm, to take better advantage of unlabeled examples. It introduces an instructor model to provide the confidence ratios as the measurement of pseudo-labels' reliability. These confidence scores then guide the target model to pay distinct attention to different data points, avoiding the over-reliance on labeled data and the negative influence of incorrect pseudo-annotations. Comprehensive experiments show that InstructBio substantially improves the generalization ability of molecular models, in not only molecular property predictions but also activity cliff estimations, demonstrating the superiority of the proposed method. Furthermore, our evidence indicates that InstructBio can be equipped with cutting-edge pretraining methods and used to establish large-scale and task-specific pseudo-labeled molecular datasets, which reduces the predictive errors and shortens the training process. Our work provides strong evidence that semi-supervised learning can be a promising tool to overcome the data scarcity limitation and advance molecular representation learning.

The learning rate is a critical hyperparameter for deep learning tasks since it determines the extent to which the model parameters are updated during the learning course. However, the choice of learning rates typically depends on empirical judgment, which may not result in satisfactory outcomes without intensive try-and-error experiments. In this study, we propose a novel learning rate adaptation scheme called QLABGrad. Without any user-specified hyperparameter, QLABGrad automatically determines the learning rate by optimizing the Quadratic Loss Approximation-Based (QLAB) function for a given gradient descent direction, where only one extra forward propagation is required. We theoretically prove the convergence of QLABGrad with a smooth Lipschitz condition on the loss function. Experiment results on multiple architectures, including MLP, CNN, and ResNet, on MNIST, CIFAR10, and ImageNet datasets, demonstrate that QLABGrad outperforms various competing schemes for deep learning., Comment: Accepted by AAAI 2024

With the improvement of high-precision optical clock, the higher-order multipolar interaction between atoms and light needs quantitative evaluation. However for the Sr clock, the differential dynamic E2-M1 polarizability at the magic wavelength has contradictions among available theoretical and experimental results. Recently, the new experimental measurement of S. D\"{o}rscher {\em et al.} [arXiv: 2210. 14727] is consistent with measurement of Ushijima {\em et al.}, which poses new challenges to theory and urgently calls for theoretical explanations. In present work, we investigate contributions of negative-energy states to the E2 and M1 polarizabilities. We find that for the M1 polarizability, the contribution from negative-energy states is crucial and dominant. Our new theoretical result for E2-M1 polarizability difference is $-7.74(3.92)\times 10^{-5}$ a.u., which is in good agreement with the recent experiment of S. D\"{o}rscher et al., so the inconsistency problem of E2-M1 polarizability in the Sr clock between theory and experiment is eliminated., Comment: 5 pages, 1 figure

The success of graph neural networks (GNNs) provokes the question about explainability: ``Which fraction of the input graph is the most determinant of the prediction?'' Particularly, parametric explainers prevail in existing approaches because of their more robust capability to decipher the black-box (i.e., target GNNs). In this paper, based on the observation that graphs typically share some common motif patterns, we propose a novel non-parametric subgraph matching framework, dubbed MatchExplainer, to explore explanatory subgraphs. It couples the target graph with other counterpart instances and identifies the most crucial joint substructure by minimizing the node corresponding-based distance. Moreover, we note that present graph sampling or node-dropping methods usually suffer from the false positive sampling problem. To alleviate this issue, we designed a new augmentation paradigm named MatchDrop. It takes advantage of MatchExplainer to fix the most informative portion of the graph and merely operates graph augmentations on the rest less informative part. Extensive experiments on synthetic and real-world datasets show the effectiveness of our MatchExplainer by outperforming all state-of-the-art parametric baselines with significant margins. Results also demonstrate that MatchDrop is a general scheme to be equipped with GNNs for enhanced performance. The code is available at: https://github.com/smiles724/MatchExplainer.

Geometric deep learning has recently achieved great success in non-Euclidean domains, and learning on 3D structures of large biomolecules is emerging as a distinct research area. However, its efficacy is largely constrained due to the limited quantity of structural data. Meanwhile, protein language models trained on substantial 1D sequences have shown burgeoning capabilities with scale in a broad range of applications. Several previous studies consider combining these different protein modalities to promote the representation power of geometric neural networks, but fail to present a comprehensive understanding of their benefits. In this work, we integrate the knowledge learned by well-trained protein language models into several state-of-the-art geometric networks and evaluate a variety of protein representation learning benchmarks, including protein-protein interface prediction, model quality assessment, protein-protein rigid-body docking, and binding affinity prediction. Our findings show an overall improvement of 20% over baselines. Strong evidence indicates that the incorporation of protein language models' knowledge enhances geometric networks' capacity by a significant margin and can be generalized to complex tasks.