XHXiaoran Hao

Xiaoran Hao

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Xiaoran Hao has a total of 25 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Memristor-Based Reservoir Computing for Limbic Epileptic Seizures Prediction in Low-Power Wearable IoT Devices

    by Hao Chen, Hao Sun, Xiaoran Hao

    Epilepsy is a prevalent chronic neurological disorder, and electroencephalogram (EEG) is a key tool for detecting it by measuring brain electrical activity. Machine learning methods show promise for predicting brain electrical activity. However, traditional approaches often have high power consumption, challenging their deployment on low-power wearable Internet of Things (IoT) devices crucial for real-time seizure prediction. This study presents a memristor-based reservoir computing (RC) model aimed at addressing these challenges by utilizing the nonlinear dynamics of memristors for effective EEG signal processing. The system integrates discrete wavelet transform (DWT) and controllable masks to enhance feature extraction, applies K-means synthetic minority oversampling technique (K-means SMOTE) for data balancing, and ultimately converts the input signals into voltage signals, which are then fed into the reservoir. This approach achieves a seizure prediction classification accuracy of 96.8% on a simple, easily integrable reservoir with tolerance characteristic. The proposed approach operates at a microwatt power consumption level, 46.53% lower than that of the latest RNN-based implantable neural stimulation methods. These findings underscore the potential of memristor-based RC systems for wearable applications, offering a significant reduction in power consumption while maintaining high prediction accuracy and providing a viable solution for actual edge side epileptic seizures prediction.

  • Open Access

    Sky-YOLO: Advanced Small Object Detection for UAV Applications

    by ZiYi CHEN, XiaoRan HAO, Ming Chen, Mao NI, Yi Heng Zhang

    Object detection from a UAV perspective faces challenges such as object occlusion, unclear boundaries, and small target sizes, resulting in reduced detection accuracy. Additionally, traditional object detection algorithms have large parameter counts, making them unsuitable for resource-constrained edge devices. To address this issue, we propose a lightweight small-object detection algorithm: Sky-YOLO. Specifically, we introduce the MSFConv multi-scale feature map fusion convolution module into the backbone network to enhance feature extraction capability. The Neck part is replaced with the L-BiFPN module to reduce parameter count and strengthen feature fusion between layers. Additionally, based on the characteristics of UAV imagery, we incorporate the WIOU loss function, enabling efficient detection of blurred and occluded targets. Experimental results show that the Sky-YOLO model, with 60% fewer parameters than the original model, still achieves 39.7% accuracy on the VisDrone2019 validation dataset, a 6.7% improvement in accuracy over the original model.


  • Open Access

    LMAI2C: Low memory access Im2col method for CNN inference

    by Mengda Li, Ziyi Chen, Jiangen Hong, Yiheng Zhang, Xiaoran Hao, Ming Chen, Mao Ni

    For neural network accelerators with General Matrix Multiplication (GEMM) as the computational core, the input feature maps of convolution must be converted into 2D matrices through the Im2col operation. Conventional approaches utilize CPUs to execute Im2col management and data transfer operations. Conventional methods suffer from memory expansion due to redundant data in overlapping convolutional windows, thus incurring non-negligible memory access energy consumption and transmission latency overheads. This severely limits the feasibility of efficient GEMM acceleration in resource-constrained edge devices. This paper proposes a novel Low Memory Access Im2col Method (LMAI2C) and present its dedicated hardware implementation. By restructuring data from overlapping convolutional windows, LMAI2C significantly reduces DRAM memory access volume while improving feature map transfer efficiency. Experimental results on convolutional layers of the YOLOv4-tiny network demonstrate that LMAI2C reduces DDR memory access by approximately 79.8% compared to traditional methods. Furthermore, LMAI2C demonstrates an average speedup of 69 times compared to CPU-based methodologies and 43 times over DMA-accelerated CPU implementations.

  • Hybrid Storage Class Memory-Enhanced Machine Learning Inference in Embedded Edge Devices

    by Hao Sun, Xiaoran Hao, Mande Xie

    With the rapid proliferation of machine learning inference (MLI) tasks on Internet of Things (IoT) devices, the demand for enhanced memory system performance in these devices has become increasingly critical. This letter proposes a hybrid storage class memory (HSCM) optimization method to improve the efficiency of MLI on IoT devices. We conducted an in-depth analysis of memory access characteristics in MLI scenarios and found that the diversity and frequency of data access patterns significantly affect storage performance. Based on these characteristics, we propose a hybrid memory migration management strategy using the Markov model, which can dynamically adjust data migration between different storage levels to adapt to changing access patterns. We designed an acceleration method based on hybrid storage class memory that improves inference speed and energy efficiency by optimizing data access paths and reducing latency. Experiments demonstrate that the proposed method outperforms existing memory systems, providing a new solution for efficient MLI on embedded IoT devices.

  • Open Access

    Object Detection Post Processing Accelerator Based on Co-Design of Hardware and Software

    by Dengtian Yang, Lan Chen, Xiaoran Hao, Yiheng Zhang

    Deep learning significantly advances object detection. Post processes, a critical component of this process, select valid bounding boxes to represent the true targets during inference and assign boxes and labels to these objects during training to optimize the loss function. However, post processes constitute a substantial portion of the total processing time for a single image. This inefficiency primarily arises from the extensive Intersection over Union (IoU) calculations required between numerous redundant bounding boxes in post processing algorithms. To reduce these redundant IoU calculations, we introduce a classification prioritization strategy during both training and inference post processes. Additionally, post processes involve sorting operations that contribute to their inefficiency. To minimize unnecessary comparisons in Top-K sorting, we have improved the bitonic sorter by developing a hybrid bitonic algorithm. These improvements have effectively accelerated the post processing. Given the similarities between the training and inference post processes, we unify four typical post processing algorithms and design a hardware accelerator based on this framework. Our accelerator achieves at least 7.55 times the speed in inference post processing compared to that of recent accelerators. When compared to the RTX 2080 Ti system, our proposed accelerator offers at least 21.93 times the speed for the training post process and 19.89 times for the inference post process, thereby significantly enhancing the efficiency of loss function minimization.


  • Open Access

    Object Detection Post-Processing Accelerator Based on Co-Design of Hardware and Software

    by Dengtian Yang, Lan Chen, Xiaoran Hao, Mao Ni, Ming Chen, Yiheng Zhang

    Deep learning significantly advances object detection. Post process, a critical component of this process, selects valid bounding boxes to represent true targets during inference and assigns boxes and labels to these objects during training to optimize the loss function. However, post process constitutes a substantial portion of the total processing time for a single image. This inefficiency primarily arises from the extensive Intersection over Union (IoU) calculations required between numerous redundant bounding boxes in post-processing algorithms. To reduce the redundant IoU calculations, we introduce a classification prioritization strategy during both training and inference post processes. Additionally, post process involves sorting operations that contribute to inefficiency. To minimize unnecessary comparisons in Top-K sorting, we have improved the bitonic sorter by developing a hybrid bitonic algorithm. These improvements have effectively accelerated post process. Given the similarities between training and inference post processes, we unify four typical post-processing algorithms and design a hardware accelerator based on this framework. Our accelerator achieves at least 7.55 times the speed in inference post process compared to recent accelerators. When compared to the RTX 2080 Ti system, our proposed accelerator offers at least 21.93 times the speed for training post process and 19.89 times for inference post process, thereby significantly enhancing the efficiency of loss function minimization.

  • Open Access

    Automated feature map padding and transfer circuit for CNN inference

    by Hongying Zhang, Ming Chen, Mao Ni, Lan Chen, Yiheng Zhang, Xiaoran Hao

    This paper introduces a novel hardware acceleration circuit designed to address the storage address offset issue in Convolutional Neural Networks (CNNs) during the feature map padding process. Traditional CPU-based padding and data transfer methods are computationally intensive and lead to high latency and power consumption, especially on edge devices. Our solution automates and integrates feature map padding and transfer. This significantly reduces DRAM access and improves the speed of transferring feature maps between DRAM and on-chip SRAM. The proposed circuit, tested on the ZCU102 development board using YOLOv4-tiny’s convolutional layers, demonstrates a speedup of over 20 times compared to CPU-based methods and more than 4 times compared to CPU with DMA.


  • Open Access

    Design of a Convolutional Neural Network Accelerator Based on On-Chip Data Reordering

    by Yang Liu, Yiheng Zhang, Xiaoran Hao, Lan Chen, Mao Ni, Ming Chen, Rong Chen

    Convolutional neural networks have been widely applied in the field of computer vision. In convolutional neural networks, convolution operations account for more than 90% of the total computational workload. The current mainstream approach to achieving high energy-efficient convolution operations is through dedicated hardware accelerators. Convolution operations involve a significant amount of weights and input feature data. Due to limited on-chip cache space in accelerators, there is a significant amount of off-chip DRAM memory access involved in the computation process. The latency of DRAM access is 20 times higher than that of SRAM, and the energy consumption of DRAM access is 100 times higher than that of multiply–accumulate (MAC) units. It is evident that the “memory wall” and “power wall” issues in neural network computation remain challenging. This paper presents the design of a hardware accelerator for convolutional neural networks. It employs a dataflow optimization strategy based on on-chip data reordering. This strategy improves on-chip data utilization and reduces the frequency of data exchanges between on-chip cache and off-chip DRAM. The experimental results indicate that compared to the accelerator without this strategy, it can reduce data exchange frequency by up to 82.9%.


  • Arithmetic Logic Unit Circuit Designed for Near-Sensor Computing Architecture With Complementary Carbon Nanotube Field-Effect Transistor Technology

    by Hao Ding, Lan Chen, Xiaoran Hao, Yu Ji

    Novel circuits designed using carbon nanotube field-effect transistors (CNFETs) are gaining increasing attention because of advantages such as low-power operation, high operating speed, and support of advanced integration technology. In this work, we present a novel arithmetic logic unit (ALU) circuit that is designed entirely using complementary CNFETs. The ALU is intended to process a custom-made algorithm for a near-sensor computing architecture that requires low power and large bit width operation. We completed the ALU design using a large-scale complementary CNFET circuit (number of CNFETs in this circuit is 8160). The total power consumption of the ALU is 218.64 μW, which will satisfy the low-power requirements of the wireless sensor terminal. To verify the behavioral functionality of the circuit, we designed a lightweight binarized neural network and accomplished Modified National Institute of Standards and Technology (MNIST) standard handwritten digit recognition successfully through a register-transfer logic simulation using the proposed circuit and external logic. The accuracy of the algorithm can reach 90.20%. Our research demonstrates the feasibility of employing CNFET circuits in fields that involve the use of complex algorithms, e.g., image classification. The circuits based on the aligned-CNFET model is simulated and compared with MOSFET circuits based on a commercial silicon-based process design kit (PDK). Compared with MOSFET circuits, the aligned-CNFET circuits have an average increase in speed by more than 93% and an order-of-magnitude improvement in the power-delay product (PDP). The ALU and algorithm proposed here are suitable for near-sensor computing architectures for future smart sensors.


  • Open Access

    Indoor Fire Detection Algorithm Based on Second-Order Exponential Smoothing and Information Fusion

    by Liuqi An, Lan Chen, Xiaoran Hao

    With the increasing complexity of building structures and interior materials, the danger of indoor fires has become more severe. It is effective to improve the accuracy and timeliness of fire-sensing devices in order to reduce the harm caused by fires. This paper focuses on the temporal characteristics of sensor information, creatively introducing second-order exponential smoothing into the information fusion algorithm. The RNN structure is used to fit the formula and adaptively trained with various types of fire data. Experimental results show that the proposed algorithm achieves an accuracy of 98% in fire recognition, significantly improving the accuracy of fire recognition. To avoid the issue of imbalanced positive and negative samples, this paper comprehensively evaluates parameters such as F1-score and Matthews correlation coefficient (MCC). The achieved scores are 0.97 and 0.95, respectively, indicating the algorithm’s good performance in detecting the presence or absence of fire. Furthermore, the proposed algorithm is tested for its alarm time. The experimental results show that the proposed algorithm can timely identify various types of fires and can give an alarm earlier than traditional fire alarms.