Jinbin Zhu

Dr. Jinbin Zhu


Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
3 October 2020 in Electronics
Verified
Journal Article
Open access
PDF file
FWFei Wu
Fei Wu
WYWenxue Yang
Wenxue Yang
LXLimin Xiao
Limin Xiao
JZJinbin Zhu
Jinbin Zhu
Deep neural network has been widely used in pattern recognition and speech processing, but its vulnerability to adversarial attacks also proverbially demonstrated. These attacks perform unstructured pixel-wise perturbation to fool the classifier, which does not affect the human visual system. The role of adversarial examples in the information security field has received increased attention across a number of disciplines in recent years. An alternative approach is “like cures like”. In this paper, we propose to utilize common noise and adaptive wiener filtering to mitigate the perturbation. Our method includes two operations: noise addition, which adds natural noise to input adversarial examples, and adaptive wiener filtering, which denoising the images in the previous step. Based on the study of the distribution of attacks, adding natural noise has an impact on adversarial examples to a certain extent and then they can be removed through adaptive wiener filter, which is an optimal estimator for the local variance of the image. The proposed improved adaptive wiener filter can automatically select the optimal window size between the given multiple alternative windows based on the features of different images. Based on lots of experiments, the result demonstrates that the proposed method is capable of defending against adversarial attacks, such as FGSM (Fast Gradient Sign Method), C&W, Deepfool, and JSMA (Jacobian-based Saliency Map Attack). By compared experiments, our method outperforms or is comparable to state-of-the-art methods.

9 December 2022 in Journal of Systems Architecture
Verified
Journal Article
Closed access
JZJinbin Zhu
Jinbin Zhu
LWLiang Wang
Liang Wang
LXLimin Xiao
Limin Xiao
LLLei Liu
Lei Liu
GQGuangjun Qin
Guangjun Qin

9 May 2024 in IEEE Transactions on Computers
Verified
Journal Article
Closed access
JWJing Wang
Jing Wang
JZJinbin Zhu
Jinbin Zhu
XFXin Fu
Xin Fu
DZDi Zang
Di Zang
KLKeyao Li
Keyao Li
WZWeigong Zhang
Weigong Zhang
Ensuring the reliability of deep neural networks (DNNs) is paramount in safety-critical applications. Although introducing supplementary fault-tolerant mechanisms can augment the reliability of DNNs, an efficiency tradeoff may be introduced. This study reveals the inherent fault tolerance of neural networks, where individual neurons exhibit varying degrees of fault tolerance, by thoroughly exploring the structural attributes of DNNs. We thereby develop a hardware/software collaborative method that guarantees the reliability of DNNs while minimizing performance degradation. We introduce the neuron vulnerability factor (NVF) to quantify the susceptibility to soft errors. We propose two efficient methods that leverage the NVF to minimize the negative effects of soft errors on neurons. First, we present a novel computational scheduling scheme. By prioritizing error-prone neurons, the expedited completion of their computations is facilitated to mitigate the risk of neural computing errors that arise from soft errors without sacrificing efficiency. Second, we propose the NVF-guided heterogeneous memory system. We employ variable-strength error-correcting codes and tailor their error-correction mechanisms to the vulnerability profile of specific neurons to ensure a highly targeted approach for error mitigation. Our experimental results demonstrate that the proposed scheme enhances the neural network accuracy by 18% on average, while significantly reducing the fault-tolerance overhead.

7 June 2023 in IEEE Transactions on Computers
Verified
Journal Article
Closed access
BYBaoyue Yan
Baoyue Yan
JZJinbin Zhu
Jinbin Zhu
BJBo Jiang
Bo Jiang
Modern fast NVMe devices with high throughput and ultra-low latency have brought new opportunities for persistent key-value (KV) engines. In this paper, we propose Limon, a persistent KV engine to exploit the performance potentials of fast NVMe devices. Limon targets three practical design aspects that existing KV engines fail to consider simultaneously: functionality, scalability, and stability. Limon carefully redesigns the index structure, on-disk KV record layout, and I/O processing of a persistent KV engine for these aspects. Specifically, Limon (i) proposes a semi-shared global index to improve scalability and range queries. (ii) employs a fast slab-based record layout with light-weight defragmentation to enable stable performance, and (iii) uses efficient asynchronous per-core I/O processing with two optimizations: DMA-backed buffer pool and page deduplication, to further improve scalability. Our evaluations with the YCSB benchmark and one production workload show that Limon outperforms state-of-the-art persistent key-value engines (i.e., SpanDB, KVell, and uDepot) by up to 1.2x to 3.8x and has the best scalability. Moreover, Limon has stable and predictable performance due to its novel record layout strategy.

18 July 2023 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Verified
Journal Article
Closed access
JZJinbin Zhu
Jinbin Zhu
LWLiang Wang
Liang Wang
LXLimin Xiao
Limin Xiao
LLLei Liu
Lei Liu
GQGuangjun Qin
Guangjun Qin
With the advent of high-performance nonvolatile memory express (NVMe) SSD, the overhead caused by the storage software stack becomes a significant bottleneck for exploiting the potential of NVMe SSD. Recent I/O isolation approaches eliminate CPU switching, I/O interference, and lock contention for I/O queues by pinning I/O threads in isolated and dedicated I/O paths. However, they degrade the overall performance of mixed workloads with heterogeneous I/O demands. The I/O-intensive workloads issue multiple I/O requests and quickly fill up their dedicated I/O queues, resulting in I/O wait. On the contrary, the non-I/O-intensive workloads cannot deliver enough I/O requests to saturate their associated I/O queues. Moreover, frequent allocations and deallocations of I/O request objects expose a significant overhead for I/O-intensive workloads. In this article, we propose EBIO, an efficient block I/O stack for NVMe SSD, to improve the overall performance of mixed workloads. EBIO contains an on-demand queue management strategy (ODQM) and a reusable I/O management strategy (RERM). Specifically, ODQM eliminates I/O wait by dynamically adding queues for I/O-intensive workloads and leverages I/O generation time to guarantee strong sequential consistency and fairness in scheduling I/O requests. RERM reduces the overhead caused by repeatedly allocating I/O request objects for I/O-intensive workloads by reusing the allocated objects. Experimental results show that, compared to the state-of-the-art approaches, EBIO improves input/output operations per second by up to 16.44% and reduces I/O latency by up to 32.76%.

10 January 2023 in Applied Sciences
Verified
Journal Article
Open access
PDF file
JZJinbin Zhu
Jinbin Zhu
LWLiang Wang
Liang Wang
LXLimin Xiao
Limin Xiao
GQGuangjun Qin
Guangjun Qin
The Non-Volatile Memory Express (NVMe) SSD provides high I/O performance for current computer systems, and direct memory access (DMA) is the critical enabling mechanism for direct I/O. However, the lengthy I/O stack becomes a new bottleneck that degrades the potential of NVMe SSD. This paper reveals that existing user-level DMA introduces additional overhead for pinning memory used by DMA from the user space. Moreover, it cannot adapt to I/O requests of different data sizes. This paper proposes an efficient and dynamically adaptive user-level DMA (uDMA) mechanism that can adapt to I/O requests for different data sizes and lighten the I/O software stack by amortizing per-request latency. The critical component of uDMA is the pinned memory pool, which avoids frequently pinning new memory blocks by reusing allocated and pinned memory blocks. In addition, it effectively connects the discrete pinned memory blocks by the scatter/gather lists, improving the utilization of the pinned memory pool. Compared with the latest user-level DMA method, uDMA has an improvement of at least 17% under various data sizes.