Kuan Li

• PhD
• Researcher at National University of Defense Technology

Out-of-distribution (OOD) detection is a crucial problem in practice, especially, for the safe deployment of machine learning models in industrial settings. Previous work has used free energy as a score function and proposed a fine-tuning method that utilized OOD data in the training phase of the classification model, which achieves a higher perfor...

In the realm of human–robot interaction, the integration of visual and verbal cues has become increasingly significant. This paper focuses on the challenges and advancements in referring image segmentation (RIS), a task that involves segmenting images based on textual descriptions. Traditional approaches to RIS have primarily focused on pixel-level...

We propose a novel U-Net model named MS-UNet for the medical image segmentation task in this study. Instead of the single-layer U-Net decoder structure used in Swin-UNet and TransUnet, we specifically design a multi-scale nested decoder based on the Swin Transformer for U-Net. The new framework is proposed based on the observation that the single-l...

Positron Emission Tomography (PET), known for its sensitivity and non-invasiveness in visualizing metabolic processes in the human body, has been widely utilized for clinical diagnosis. However, the procedure of PET imaging requires the administration of a radioactive tracer, which poses potential risks to human health. Reducing the usage of radioa...

The progressive loss of motor function in the brain is a hallmark of Parkinson's disease (PD). Electroencephalogram (EEG) signals are commonly used for early diagnosis since they are associated with a brain disorder. This work aims to find a better way to represent electroencephalography (EEG) signals and enhance the classification accuracy of indi...

Video-based gesture recognition plays an important role in the field of human-computer interaction (HCI), and most of the existing video-based gesture recognition works are based on traditional RGB gesture videos. Compared with RGB gesture videos, RGB-D gesture videos contain additional depth information along with each data frame. Such depth infor...

In online learning, the dynamic regret metric chooses the reference oracle that may change over time, while the typical (static) regret metric assumes the reference solution to be constant over the whole time horizon. The dynamic regret metric is particularly interesting for applications such as online recommendation (since the customers' preferenc...

Stochastic Gradient Descent (SGD) with variance reduction techniques has been proved powerful to train the parameters of various machine learning models. However, it cannot support the distributed systems trivially due to the intrinsic design. Although conventional studies such as PetuumSGD perform well for distributed machine learning tasks, they...

Overlapping cell segmentation is a prerequisite for the analysis of cervical smear images. Because of the complexity of overlapping situations and the poor contrast of overlapping edges, this problem is one of the most challenges in this field. In this paper, a novel unsupervised segmentation method without needing the training data for overlapping...

Automatic detection of abnormal cells from cervical smear images is extremely demanded in annual diagnosis of women’s cervical cancer. For this medical cell recognition problem, there are three different feature sections, namely cytology morphology, nuclear chromatin pathology and region intensity. The challenges of this problem come from feature c...

Motion coherence index is a metric to evaluate the coherent motion among vectors. In this article, we propose a coherent motion indicator to describe coherent motion among four adjacent vectors. For the preliminary investigation on stationarity of dynamic texture, the coherent motion indicator is proposed as an attribute of vector. The coherent mot...

Automatic identification of abnormal cervical cells, including feature representation, feature combination and classification strategy, is highly demanded in women's annual cervical cancer screenings. However, previous methods only deal with one or two of these three phases, and currently there is few complete framework for this problem. A novel th...

Accurate and effective cervical smear image segmentation is required for automated cervical cell analysis systems. Thus, we proposed a novel superpixel-based Markov random field (MRF) segmentation framework to acquire the nucleus, cytoplasm and image background of cell images. We seek to classify color non-overlapping superpixel-patches on one imag...

The ARMv8 64-bit platform has been considered as an alternative for high performance computing (HPC). Stencil computations are a class of iterative kernels which update array elements according to a stencil. In this paper, we evaluate the performance and scalability of one ARMv8 64-bit Multi-Core Processor with 7-point 3D stencil code, and a series...

The communication and program performance in MIC-CPU symmetric mode was evaluated. First, we tested the bandwidths and latencies of communications between different computing units. Then, the performance of different combination of computing units using the Linpack and NPB was evaluated. The result shows that: The bandwidths and latencies between d...

The double-precision matrix-matrix multiplication (DGEMM) on ARMv8 64-bit multi-core processor architecture was realized and optimized, and the optimal model for the purpose of maximizing the compute-to-memory access ratio was built to design DGEMM kernel. The ARM 64-bit memory accessing instruction, Cache pre-fetching instruction and NEON vector F...

This paper proposes a new non-reference image quality metric that can be adopted by the state-of-the-art image/ video denoising algorithms for auto-denoising. The proposed metric is extremely simple and can be implemented in four lines of Matlab code. The basic assumption employed by the proposed metric is that the noise should be independent of th...

Export Date: 28 July 2014

A secure and practical outsourcing mechanism called Partitioned ELM is described to address the challenge of performing the ELM over large-scale data. The Partitioned ELM algorithm can significantly improve the training time of the original ELM algorithm by outsourcing the heaviest computation. The ELM is instinctively suitable for outsourcing in c...

Extreme learning machine (ELM) for single-hidden-layer feedforward neural networks (SLFN) is a powerful machine learning technique, and has been attracting attentions for its fast learning speed and good generalization performance. Recently, a weighted ELM is proposed to deal with data with imbalanced class distribution. The key essence of weighted...

In this paper, we present four image descriptors for HEp-2 cell staining patterns classification, including LBP, Gabor, DCT, and a global appearance statistical descriptor. A multiclass boosting SVM algorithm is proposed to integrate these descriptors together: (1) within each boosting round, four multiclass posterior probability SVMs are trained c...

Touchless fingerprint capturing can provide richer features and overcome the deficiency of 2D fingerprint recognition. In this paper, a clustering-based dynamic score selection (CDSS) algorithm is proposed for the combination of scores which are generated by different vision touchless fingerprint recognition systems. First, the scores are divided i...

A Radiating Gradient Vector Flow (RGVF) Snake aiming at accurate extraction of both the nucleus and cytoplasm from a single-cell cervical smear image is proposed. After preprocessing, the areas in the image are roughly clustered into nucleus, cytoplasm and the background by a spatial K-means clustering algorithm. After initial contours are extracte...

In context of incomplete image set, a novel EM-based Principal Component Analysis (PCA) algorithm (EM-PCAM) is proposed to exploit the principal components. Then the EM-PCAM is integrated seamlessly into the Local Tangent Space Alignment (LTSA) algorithm instead of standard SVD such that the modified LTSA algorithm can learn the manifold of incompl...

An improved recognition method of PDF417 barcode is proposed in this paper. It is efficient and accurate in complex image conditions, including high noises, distortions, low contrast and uneven environment illumination. The method consists of the following four steps: ROI detection-based on edge detection and morphological operation, barcode rectif...