arXiv subjects
Publications and source records attributed to Ziyu Li.
In late 1990's Tsujii proved monotonicity of topological entropy of real quadratic family $f_c(x)=x^2+c$ on parameter $c$ by proving an inequality concerning orbital information of the critical point. In this paper, we consider a weak analog of such inequality for the general family $f_{c,r}(x)=|x|^r+c$ with rational $r>1$, by following an algebraic approach.
Multimodal large language models (MLLMs) are increasingly being applied in the medical field, particularly in medical imaging. However, developing MLLMs for ECG signals, which are crucial in clinical settings, has been a significant challenge beyond medical imaging. Previous studies have attempted to address this by converting ECGs into several text tags using an external classifier in a training-free manner. However, this approach significantly compresses the information in ECGs and underutilizes the reasoning capabilities of LLMs. In this work, we directly feed the embeddings of ECGs into the LLM through a projection layer, retaining more information about ECGs and better leveraging the reasoning abilities of LLMs. Our method can also effectively handle a common situation in clinical practice where it is necessary to compare two ECGs taken at different times. Recent studies found that MLLMs may rely solely on text input to provide answers, ignoring inputs from other modalities. We analyzed this phenomenon from a causal perspective in the context of ECG MLLMs and discovered that the confounder, severity of illness, introduces a spurious correlation between the question and answer, leading the model to rely on this spurious correlation and ignore the ECG input. Such models do not comprehend the ECG input and perform poorly in adversarial tests where different expressions of the same question are used in the training and testing sets. We designed a de-biased pre-training method to eliminate the confounder's effect according to the theory of backdoor adjustment. Our model performed well on the ECG-QA task under adversarial testing and demonstrated zero-shot capabilities. An interesting random ECG test further validated that our model effectively understands and utilizes the input ECG signal.
The advent of Large Language Models (LLMs) provides an opportunity to change the way queries are processed, moving beyond the constraints of conventional SQL-based database systems. However, using an LLM to answer a prediction query is still challenging, since an external ML model has to be employed and inference has to be performed in order to provide an answer. This paper introduces LLM-PQA, a novel tool that addresses prediction queries formulated in natural language. LLM-PQA is the first to combine the capabilities of LLMs and retrieval-augmented mechanism for the needs of prediction queries by integrating data lakes and model zoos. This integration provides users with access to a vast spectrum of heterogeneous data and diverse ML models, facilitating dynamic prediction query answering. In addition, LLM-PQA can dynamically train models on demand, based on specific query requirements, ensuring reliable and relevant results even when no pre-trained model in a model zoo, available for the task.
This chapter provides an overview of deep learning techniques for improving the spatial resolution of MRI, ranging from convolutional neural networks, generative adversarial networks, to more advanced models including transformers, diffusion models, and implicit neural representations. Our exploration extends beyond the methodologies to scrutinize the impact of super-resolved images on clinical and neuroscientific assessments. We also cover various practical topics such as network architectures, image evaluation metrics, network loss functions, and training data specifics, including downsampling methods for simulating low-resolution images and dataset selection. Finally, we discuss existing challenges and potential future directions regarding the feasibility and reliability of deep learning-based MRI super-resolution, with the aim to facilitate its wider adoption to benefit various clinical and neuroscientific applications.
Magnetic resonance imaging (MRI) has significantly benefited from the resurgence of artificial intelligence (AI). By leveraging AI's capabilities in large-scale optimization and pattern recognition, innovative methods are transforming the MRI acquisition workflow, including planning, sequence design, and correction of acquisition artifacts. These emerging algorithms demonstrate substantial potential in enhancing the efficiency and throughput of acquisition steps. This review discusses several pivotal AI-based methods in neuro MRI acquisition, focusing on their technological advances, impact on clinical practice, and potential risks.
Pre-trained deep learning (DL) models are increasingly accessible in public repositories, i.e., model zoos. Given a new prediction task, finding the best model to fine-tune can be computationally intensive and costly, especially when the number of pre-trained models is large. Selecting the right pre-trained models is crucial, yet complicated by the diversity of models from various model families (like ResNet, Vit, Swin) and the hidden relationships between models and datasets. Existing methods, which utilize basic information from models and datasets to compute scores indicating model performance on target datasets, overlook the intrinsic relationships, limiting their effectiveness in model selection. In this study, we introduce TransferGraph, a novel framework that reformulates model selection as a graph learning problem. TransferGraph constructs a graph using extensive metadata extracted from models and datasets, while capturing their inherent relationships. Through comprehensive experiments across 16 real datasets, both images and texts, we demonstrate TransferGraph's effectiveness in capturing essential model-dataset relationships, yielding up to a 32% improvement in correlation between predicted performance and the actual fine-tuning results compared to the state-of-the-art methods.
Large Language Models (LLMs) have shown remarkable capabilities in processing both natural and programming languages, which have enabled various applications in software engineering, such as requirement engineering, code generation, and software testing. However, existing code generation benchmarks do not necessarily assess the code understanding performance of LLMs, especially for the subtle inconsistencies that may arise between code and its semantics described in natural language. In this paper, we propose a novel method to systematically assess the code understanding performance of LLMs, particularly focusing on subtle differences between code and its descriptions, by introducing code mutations to existing code generation datasets. Code mutations are small changes that alter the semantics of the original code, creating a mismatch with the natural language description. We apply different types of code mutations, such as operator replacement and statement deletion, to generate inconsistent code-description pairs. We then use these pairs to test the ability of LLMs to correctly detect the inconsistencies. We propose a new LLM testing method, called Mutation-based Consistency Testing (MCT), and conduct a case study on the two popular LLMs, GPT-3.5 and GPT-4, using the state-of-the-art code generation benchmark, HumanEval-X, which consists of six programming languages (Python, C++, Java, Go, JavaScript, and Rust). We compare the performance of the LLMs across different types of code mutations and programming languages and analyze the results. We find that the LLMs show significant variation in their code understanding performance and that they have different strengths and weaknesses depending on the mutation type and language.
3D multi-slab acquisitions are an appealing approach for diffusion MRI because they are compatible with the imaging regime delivering optimal SNR efficiency. In conventional 3D multi-slab imaging, shot-to-shot phase variations caused by motion pose challenges due to the use of multi-shot k-space acquisition. Navigator acquisition after each imaging echo is typically employed to correct phase variations, which prolongs scan time and increases the specific absorption rate (SAR). The aim of this study is to develop a highly efficient, self-navigated method to correct for phase variations in 3D multi-slab diffusion MRI without explicitly acquiring navigators. The sampling of each shot is carefully designed to intersect with the central kz plane of each slab, and the multi-shot sampling is optimized for self-navigation performance while retaining decent reconstruction quality. The central kz intersections from all shots are jointly used to reconstruct a 2D phase map for each shot using a structured low-rank constrained reconstruction that leverages the redundancy in shot and coil dimensions. The phase maps are used to eliminate the shot-to-shot phase inconsistency in the final 3D multi-shot reconstruction. We demonstrate the method's efficacy using retrospective simulations and prospectively acquired in-vivo experiments at 1.22 mm and 1.09 mm isotropic resolutions. Compared to conventional navigated 3D multi-slab imaging, the proposed self-navigated method achieves comparable image quality while shortening the scan time by 31.7% and improving the SNR efficiency by 15.5%. The proposed method produces comparable quality of DTI and white matter tractography to conventional navigated 3D multi-slab acquisition with a much shorter scan time.
Dropout is conventionally used during the training phase as regularization method and for quantifying uncertainty in deep learning. We propose to use dropout during training as well as inference steps, and average multiple predictions to improve the accuracy, while reducing and quantifying the uncertainty. The results are evaluated for fractional anisotropy (FA) and mean diffusivity (MD) maps which are obtained from only 3 direction scans. With our method, accuracy can be improved significantly compared to network outputs without dropout, especially when the training dataset is small. Moreover, confidence maps are generated which may aid in diagnosis of unseen pathology or artifacts.
We report the elastic properties of the Kondo semimetal CeBi by resonant ultrasound spectroscopy measurements at zero magnetic field. Clear elastic softening is found in bulk modulus $C_B$ below $\sim 60$ K. Such a softening in $C_B$, in addition to the anomalous temperature dependent Poisson's ratio, is hardly attributable to multipolar response for stable localized $4f$ orbital, but can be well described by a two-band model arising from the hybridization between conduction- and $4f$- electrons. These results probably are consequences of the valence fluctuations in this Kondo semimetal as originally suggested by a Fermi-surface expansion observed in a previous angle-resolved photoemission spectroscopy study [P. Li \textit{et al.}, Phys. Rev. B $\mathbf{100}$, 155110 (2019)].
LiDAR-based 3D detection has made great progress in recent years. However, the performance of 3D detectors is considerably limited when deployed in unseen environments, owing to the severe domain gap problem. Existing domain adaptive 3D detection methods do not adequately consider the problem of the distributional discrepancy in feature space, thereby hindering generalization of detectors across domains. In this work, we propose a novel unsupervised domain adaptive \textbf{3D} detection framework, namely \textbf{G}eometry-aware \textbf{P}rototype \textbf{A}lignment (\textbf{GPA-3D}), which explicitly leverages the intrinsic geometric relationship from point cloud objects to reduce the feature discrepancy, thus facilitating cross-domain transferring. Specifically, GPA-3D assigns a series of tailored and learnable prototypes to point cloud objects with distinct geometric structures. Each prototype aligns BEV (bird's-eye-view) features derived from corresponding point cloud objects on source and target domains, reducing the distributional discrepancy and achieving better adaptation. The evaluation results obtained on various benchmarks, including Waymo, nuScenes and KITTI, demonstrate the superiority of our GPA-3D over the state-of-the-art approaches for different adaptation scenarios. The MindSpore version code will be publicly available at \url{https://github.com/Liz66666/GPA3D}.
Purpose: This study aims to propose a model-based reconstruction algorithm for simultaneous multi-slab diffusion MRI acquired with blipped-CAIPI gradients (blipped-SMSlab), which can also incorporate distortion correction. Methods: We formulate blipped-SMSlab in a 4D k-space with kz gradients for the intra-slab slice encoding and km (blipped-CAIPI) gradients for the inter-slab encoding. Because kz and km gradients share the same physical axis, the blipped-CAIPI gradients introduce phase interference in the z-km domain while motion induces phase variations in the kz-m domain. Thus, our previous k-space-based reconstruction would need multiple steps to transform data back and forth between k-space and image space for phase correction. Here we propose a model-based hybrid-space reconstruction algorithm to correct the phase errors simultaneously. Moreover, the proposed algorithm is combined with distortion correction, and jointly reconstructs data acquired with the blip-up/down acquisition to reduce the g-factor penalty. Results: The blipped-CAIPI-induced phase interference is corrected by the hybrid-space reconstruction. Blipped-CAIPI can reduce the g-factor penalty compared to the non-blipped acquisition in the basic reconstruction. Additionally, the joint reconstruction simultaneously corrects the image distortions and improves the 1/g-factors by around 50%. Furthermore, through the joint reconstruction, SMSlab acquisitions without the blipped-CAIPI gradients also show comparable correction performance with blipped-SMSlab. Conclusion: The proposed model-based hybrid-space reconstruction can reconstruct blipped-SMSlab diffusion MRI successfully. Its extension to a joint reconstruction of the blip-up/down acquisition can correct EPI distortions and further reduce the g-factor penalty compared with the separate reconstruction.
The data needed for machine learning (ML) model training, can reside in different separate sites often termed data silos. For data-intensive ML applications, data silos pose a major challenge: the integration and transformation of data demand a lot of manual work and computational resources. With data privacy and security constraints, data often cannot leave the local sites, and a model has to be trained in a decentralized manner. In this work, we present a vision on how to bridge the traditional data integration (DI) techniques with the requirements of modern machine learning. We explore the possibilities of utilizing metadata obtained from data integration processes for improving the effectiveness and efficiency of ML models. We analyze two common use cases over data silos, feature augmentation and federated learning. Bringing data integration and machine learning together, we highlight the new research opportunities from the aspects of systems, representations, factorized learning and federated learning.
Machine learning (ML) practitioners and organizations are building model zoos of pre-trained models, containing metadata describing properties of the ML models and datasets that are useful for reporting, auditing, reproducibility, and interpretability purposes. The metatada is currently not standardised; its expressivity is limited; and there is no interoperable way to store and query it. Consequently, model search, reuse, comparison, and composition are hindered. In this paper, we advocate for standardized ML model meta-data representation and management, proposing a toolkit supported to help practitioners manage and query that metadata.
Messenger RNA (mRNA) vaccines are being used for COVID-19, but still suffer from the critical issue of mRNA instability and degradation, which is a major obstacle in the storage, distribution, and efficacy of the vaccine. Previous work showed that optimizing secondary structure stability lengthens mRNA half-life, which, together with optimal codons, increases protein expression. Therefore, a principled mRNA design algorithm must optimize both structural stability and codon usage to improve mRNA efficiency. However, due to synonymous codons, the mRNA design space is prohibitively large, e.g., there are $\sim\!10^{632}$ mRNAs for the SARS-CoV-2 Spike protein, which poses insurmountable challenges to previous methods. Here we provide a surprisingly simple solution to this hard problem by reducing it to a classical problem in computational linguistics, where finding the optimal mRNA is akin to finding the most likely sentence among similar sounding alternatives. Our algorithm, named LinearDesign, takes only 11 minutes for the Spike protein, and can jointly optimize stability and codon usage. Experimentally, without chemical modification, our designs substantially improve mRNA half-life and protein expression in vitro, and dramatically increase antibody response by up to 23$\times$ in vivo, compared to the codon-optimized benchmark. Our work enables the exploration of highly stable and efficient designs that are previously unreachable and is a timely tool not only for vaccines but also for mRNA medicine encoding all therapeutic proteins (e.g., monoclonal antibodies and anti-cancer drugs).
To handle the different types of surface reconstruction tasks, we have replicated as well as modified a few of reconstruction methods and have made comparisons between the traditional method and data-driven method for reconstruction the surface of an object with dense point cloud as input. On top of that, we proposed a system using tightly-coupled SLAM as an input to generate deskewed point cloud and odometry and a Truncated Signed Distance Function based Surface Reconstruction Library. To get higher accuracy, IMU(Inertial Measurement Unit) pre-integration and pose graph optimization are conduct in the SLAM part. With the help of the Robot Operating System, we could build a system containing those two parts, which can conduct a real-time outdoor surface reconstruction.
The noise in diffusion-weighted images (DWIs) decreases the accuracy and precision of diffusion tensor magnetic resonance imaging (DTI) derived microstructural parameters and leads to prolonged acquisition time for achieving improved signal-to-noise ratio (SNR). Deep learning-based image denoising using convolutional neural networks (CNNs) has superior performance but often requires additional high-SNR data for supervising the training of CNNs, which reduces the practical feasibility. We develop a self-supervised deep learning-based method entitled "SDnDTI" for denoising DTI data, which does not require additional high-SNR data for training. Specifically, SDnDTI divides multi-directional DTI data into many subsets, each consisting of six DWI volumes along optimally chosen diffusion-encoding directions that are robust to noise for the tensor fitting, and then synthesizes DWI volumes along all acquired directions from the diffusion tensors fitted using each subset of the data as the input data of CNNs. On the other hand, SDnDTI synthesizes DWI volumes along acquired diffusion-encoding directions with higher SNR from the diffusion tensors fitted using all acquired data as the training target. SDnDTI removes noise from each subset of synthesized DWI volumes using a deep 3-dimensional CNN to match the quality of the cleaner target DWI volumes and achieves even higher SNR by averaging all subsets of denoised data. The denoising efficacy of SDnDTI is demonstrated on two datasets provided by the Human Connectome Project (HCP) and the Lifespan HCP in Aging. The SDnDTI results preserve image sharpness and textural details and substantially improve upon those from the raw data. The results of SDnDTI are comparable to those from supervised learning-based denoising and outperform those from state-of-the-art conventional denoising algorithms including BM4D, AONLM and MPPCA.
Surgical skills have a great influence on surgical safety and patients' well-being. Traditional assessment of surgical skills involves strenuous manual efforts, which lacks efficiency and repeatability. Therefore, we attempt to automatically predict how well the surgery is performed using the surgical video. In this paper, a unified multi-path framework for automatic surgical skill assessment is proposed, which takes care of multiple composing aspects of surgical skills, including surgical tool usage, intraoperative event pattern, and other skill proxies. The dependency relationships among these different aspects are specially modeled by a path dependency module in the framework. We conduct extensive experiments on the JIGSAWS dataset of simulated surgical tasks, and a new clinical dataset of real laparoscopic surgeries. The proposed framework achieves promising results on both datasets, with the state-of-the-art on the simulated dataset advanced from 0.71 Spearman's correlation to 0.80. It is also shown that combining multiple skill aspects yields better performance than relying on a single aspect.