高速数据流下无边界在线稀疏连续学习方法

doi:10.11772/j.issn.1001-9081.2025040452

《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (4): 1034-1041.DOI: 10.11772/j.issn.1001-9081.2025040452

高速数据流下无边界在线稀疏连续学习方法

韩雨晨¹^,²^,³, 徐峰磊¹, 吕凡⁴, 姚睿⁵, 胡伏原¹^,²^,³()

^1.苏州科技大学电子与信息工程学院，江苏苏州 215009
^2.江苏省工业智能低碳技术工程研究中心，江苏苏州 215009
^3.苏州市智能低碳技术应用重点实验室（苏州科技大学），江苏苏州 215009
^4.中国科学院自动化研究所，北京 100190
^5.中国矿业大学计算机科学与技术学院，江苏徐州 221116

收稿日期:2025-04-23 修回日期:2025-07-29 接受日期:2025-07-31 发布日期:2025-08-11 出版日期:2026-04-10
通讯作者: 胡伏原
作者简介:韩雨晨（2000—），女，江苏南京人，硕士研究生，CCF会员，主要研究方向：机器学习、连续学习
徐峰磊（1992—），男，江苏苏州人，副教授，博士，CCF会员，主要研究方向：模式识别、智能系统
吕凡（1993—），男，江苏宿迁人，博士，CCF会员，主要研究方向：连续学习
姚睿（1982—），男，河南南阳人，教授，博士，CCF会员，主要研究方向：计算机视觉、模式识别、深度学习、人工智能
基金资助:
国家自然科学基金资助项目(62476189)

Task-free online sparse continual learning method for high-speed data streams

Yuchen HAN¹^,²^,³, Fenglei XU¹, Fan LYU⁴, Rui YAO⁵, Fuyuan HU¹^,²^,³()

^1.School of Electronic and Information Engineering，Suzhou University of Science and Technology，Suzhou Jiangsu 215009，China
^2.Jiangsu Industrial Intelligent and Low-carbon Technology Engineering Research Center，Suzhou Jiangsu 215009，China
^3.Suzhou Key Laboratory of Intelligent and Low-carbon Technology Application （Suzhou University of Science and Technology），Suzhou Jiangsu 215009，China
^4.Institute of Automation，Chinese Academy of Sciences，Beijing 100190，China
^5.School of Computer Science and Technology，China University of Mining and Technology，Xuzhou Jiangsu 221116，China

Received:2025-04-23 Revised:2025-07-29 Accepted:2025-07-31 Online:2025-08-11 Published:2026-04-10
Contact: Fuyuan HU
About author:HAN Yuchen， born in 2000， M. S. candidate. Her research interests include machine learning， continual learning.
XU Fenglei， born in 1992， Ph. D.， associate professor. His research interests include pattern recognition， intelligent systems.
LYU Fan， born in 1993， Ph. D. His research interests include continual learning.
YAO Rui， born in 1982， Ph. D.， professor. His research interests include computer vision， pattern recognition， deep learning， artificial intelligence.
Supported by:
National Natural Science Foundation of China(62476189)

摘要/Abstract

摘要：

无边界在线连续学习是一种任务无关的自主高效机器学习方法，通过连续适应新数据和抑制遗忘实现模型的动态更新。现有在线连续学习方法通常追求模型准确率而牺牲计算效率，这使得在高速数据流场景中，模型因训练速度滞后，难以及时响应数据流的变化。针对以上问题，提出一种面向效率?性能协同优化的在线稀疏连续学习框架，通过构建双向稀疏自适应调控机制突破传统方法的瓶颈。首先，设计参数重要性度量的动态稀疏拓扑优化框架，融合参数敏感性分析，实现非结构化参数剪枝；其次，建立记忆?效率双目标优化模型，基于在线类别分布估计动态调节计算预算分配，实现计算资源的最优配置；最后，构建梯度解耦优化策略，采用梯度掩码方法实现新旧知识的双向优化，在加速模型更新的同时保持知识拓扑的完整性。基准测试结果表明，所提框架展现出显著优势，当内存缓冲区大小为100时，相较于基线ER（Experience Replay），在CIFAR-10数据集上，平均在线准确率（AOA）和测试准确率（TA）分别提升了4.86%和6.25%；在CIFAR-100数据集上，AOA和TA分别提升了13.77%和3.08%；在Mini-ImageNet数据集上，AOA和TA的提升幅度达17.83%和25.00%。可视化分析表明，所提框架在保持实时响应能力的同时，成功捕获了数据流中的潜在概念漂移模式。可见，所提框架突破了传统方法在计算效率与模型性能间的权衡困境，为开放环境下的在线连续学习系统建立了新范式。

关键词: 连续学习, 稀疏化, 在线学习, 参数掩码, 计算效率

Abstract:

Task-free online continual learning is a task-agnostic and autonomous machine learning approach, where model dynamic update is achieved through continuous adaptation to new data and forgetting suppression. In the existing online continual learning methods, model accuracy is prioritized typically at the expense of computational efficiency, making it difficult for the model to respond promptly to changes in the data stream due to the lag in training speed in high-speed data stream scenarios. To address the above challenge, an efficiency-performance co-optimized online sparse continual learning framework was proposed to overcome the limitations of conventional approaches through constructing a bidirectional sparse adaptive regulation mechanism. Firstly, a dynamic sparse topology optimization framework for parameter importance measurement was designed, so that unstructured parameter pruning was achieved by incorporating parameter sensitivity analysis. Secondly, a memory-efficiency dual-objective optimization model was established, in which computational budget allocation was adjusted dynamically based on online class distribution estimation, so as to realize optimal computational resource configuration. Finally, a gradient decoupling optimization strategy was developed to employ gradient masking to enable bidirectional optimization of both old and new knowledge, thereby accelerating model updates and preserving the integrity of the knowledge topology at the same time. The benchmark tests results show that the proposed framework has significant advantages. Compared to the baseline ER (Experience Replay), with a memory buffer of 100, on CIFAR-10 dataset the framework achieves average improvements of 4.86% in Average Online Accuracy (AOA) and 6.25% in Test Accuracy (TA) ; on CIFAR-100 dataset, the framework obtains enhancements of 13.77% in AOA and 3.08% in TA; on Mini-ImageNet dataset, it shows performance gains of 17.83% and 25.00% in AOA and TA respectively. Visualization analysis shows that the proposed framework captures underlying concept drift patterns in data streams successfully while maintaining real-time response ability. It can be seen that the proposed framework breaks through the traditional methods’ dilemma of trade-off between computational efficiency and model performance, and establishes a new paradigm for online continual learning systems in open environments.

Key words: Continual Learning (CL), sparsity, online learning, parameter masking, computational efficiency

中图分类号:

TP391.4

韩雨晨, 徐峰磊, 吕凡, 姚睿, 胡伏原. 高速数据流下无边界在线稀疏连续学习方法[J]. 计算机应用, 2026, 46(4): 1034-1041.

Yuchen HAN, Fenglei XU, Fan LYU, Rui YAO, Fuyuan HU. Task-free online sparse continual learning method for high-speed data streams[J]. Journal of Computer Applications, 2026, 46(4): 1034-1041.

图/表 10

参考文献 32

[1]	KIRKPATRICK J， PASCANU R， RABINOWITZ N， et al. Overcoming catastrophic forgetting in neural networks［J］. National Academy of Sciences of the United States of America， 2017， 114（13）： 3521-3526.
[2]	ALFARRA M， CAI Z， BIBI A， et al. SimCS： simulation for domain incremental online continual segmentation［C］// Proceedings of the 38th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2024： 10795-10803.
[3]	YE F， BORS A G. Task-free continual learning via online discrepancy distance learning［C］// Proceedings of the 36th Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2022： 23675-23688.
[4]	李文斌，熊亚锟，范祉辰，等. 持续学习的研究进展与趋势［J］. 计算机研究与发展， 2024， 61（6）： 1476-1496.
	LI W B， XIONG Y K， FAN Z C， et al. Advances and trends of continual learning［J］. Journal of Computer Research and Development， 2024， 61（6）： 1476-1496.
[5]	王卓，瞿绍军. 深度学习实时语义分割研究进展和挑战［J］. 中国图象图形学报， 2024， 29（5）： 1188-1220.
	WANG Z， QU S J. Research progress and challenges in real-time semantic segmentation for deep learning［J］. Journal of Image and Graphics， 2024， 29（5）： 1188-1220.
[6]	CAI Z， SENER O， KOLTUN V. Online continual learning with natural distribution shifts： an empirical study with visual data［C］// Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision. Piscataway： IEEE， 2021： 8261-8270.
[7]	LI Z， HOIEM D. Learning without forgetting［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2018， 40（12）： 2935-2947.
[8]	MOCANU D C， MOCANU E， STONE P， et al. Scalable training of artificial neural networks with adaptive sparse connectivity inspired by network science［J］. Nature Communications， 2018， 9： No.2383.
[9]	ALJUNDI R， KELCHTERMANS K， TUYTELAARS T. Task-free continual learning［C］// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2019： 11246-11255.
[10]	ALJUNDI R， CACCIA L， BELILOVSKY E， et al. Online continual learning with maximal interfered retrieval［C］// Proceedings of the 33rd Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2019： 11872-11883.
[11]	ALJUNDI R， LIN M， GOUJAUD B， et al. Gradient based sample selection for online continual learning［C］// Proceedings of the 33rd Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2019： 11817-11826.
[12]	LANGE M D， TUYTELAARS T. Continual prototype evolution： learning online from nonstationary data streams［C］// Proceedings of the 2021 International Conference on Computer Vision. Piscataway： IEEE， 2021： 8250-8259.
[13]	JIN X， SADHU A， DU J， et al. Gradient-based editing of memory examples for online task-free continual learning［C］// Proceedings of the 35th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2021： 29193-29205.
[14]	CHAUDHRY A， DOKANIA P K， AJANTHAN T， et al. Riemannian walk for incremental learning： understanding forgetting and intransigence［C］// Proceedings of the 2018 European Conference on Computer Vision， LNCS 11215. Cham： Springer， 2018： 556-572.
[15]	WANG Z， JIAN T， CHOWDHURY K， et al. Learn-prune-share for lifelong learning［C］// Proceedings of the 2020 IEEE International Conference on Data Mining. Piscataway： IEEE， 2020： 641-650.
[16]	DETTMERS T， ZETTLEMOYER L. Sparse networks from scratch： faster training without losing performance［EB/OL］. ［2019-08-23］..
[17]	YAN S， XIE J， HE X. DER： dynamically expandable representation for class incremental learning［C］// Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2021： 3013-3022.
[18]	ADEL T. Similarity-based adaptation for task-aware and task-free continual learning［J］. Journal of Artificial Intelligence Research， 2024， 80： 377-417.
[19]	YE F， BORS A G. Task-free dynamic sparse vision transformer for continual learning［C］// Proceedings of the 38th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2024： 16442-16450.
[20]	YE F， BORS A G. Continual variational autoencoder learning via online cooperative memorization［C］// Proceedings of the 2022 European Conference on Computer Vision， LNCS 13683. Cham： Springer， 2022： 531-549.
[21]	KRIZHEVSKY A. Learning multiple layers of features from tiny images［R/OL］. ［2025-04-08］..
[22]	VINYALS O， BLUNDELL C， LILLICRAP T， et al. Matching networks for one shot learning［C］// Proceedings of the 30th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2016： 3637-3645.
[23]	周大蔚，汪福运，叶翰嘉，等. 基于深度学习的类别增量学习算法综述［J］. 计算机学报， 2023， 46（8）： 1577-1605.
	ZHOU D W， WANG F Y， YE H J， et al. Deep learning for class-incremental learning： a survey［J］. Chinese Journal of Computers， 2023， 46（8）： 1577-1605.
[24]	ZHU X， YI J， ZHANG L. Continual learning with unknown task boundary［J］. IEEE Transactions on Neural Networks and Learning Systems， 2025， 36（5）： 8140-8152.
[25]	CHEN X， LI H， LI M， et al. Learning a sparse Transformer network for effective image deraining［C］// Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2023： 5896-5905.
[26]	WANG Z， ZHAN Z， GONG Y， et al. SparCL： sparse continual learning on the edge［C］// Proceedings of the 36th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2022： 20366-20380.
[27]	GHUNAIM Y， BIBI A， ALHAMOUD K， et al. Real-time evaluation in online continual learning： a new hope［C］// Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2023： 11888-11897.
[28]	CHAUDHRY A， ROHRBACH M， ELHOSEINY M， et al. Continual learning with tiny episodic memories［EB/OL］. ［2019-02-27］..
[29]	CACCIA L， ALJUNDI R， ASADI N， et al. New insights on reducing abrupt representation change in online continual learning［EB/OL］. ［2019-02-27］..
[30]	BOSCHINI M， BONICELLI L， BUZZEGA P， et al. Class-incremental continual learning into the eXtended DER-verse［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2023， 45（5）： 5497-5512.
[31]	SEO M， KOH H， JEUNG W， et al. Learning equi-angular representations for online continual learning［C］// Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2024： 23933-23942.
[32]	LEE B H， OH M H， CHUN S Y. Doubly perturbed task free continual learning［C］// Proceedings of the 38th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2024： 13346-13354.

内存缓冲区大小	方法	CIFAR-10			CIFAR-100			Mini-ImageNet
内存缓冲区大小	方法	AOA/%	TA/%	运算浮点数/GFLOPs	AOA/%	TA/%	运算浮点数/GFLOPs	AOA/%	TA/%	运算浮点数/GFLOPs
100	ACE	53.05	67.01	657.93	12.15	17.92	666.64	16.07	0.99	5 205.41
	GSS	33.67	43.69	3 947.58	5.45	7.86	3 999.81	8.43	1.16	31 232.48
	LwF	49.08	59.07	877.24	10.99	17.98	888.85	14.85	0.75	5 783.41
	MIR	43.55	55.24	1 644.83	9.31	14.57	1 666.60	12.62	0.75	13 013.54
	RWalk	42.22	54.72	1 315.86	7.46	12.23	1 333.28	11.35	1.25	10 410.83
	ER	51.60	63.97	657.93	11.55	17.86	666.64	14.13	0.92	5 205.41
	X-DER	52.39	64.39	822.42	11.95	17.38	833.30	14.85	0.93	6 506.77
	EARL	52.82	64.93	1 291.83	12.01	17.72	1 308.93	15.41	1.04	10 220.71
	DPCL	52.73	64.29	756.76	11.98	17.93	766.77	15.38	0.93	5 987.27
	本文方法	54.11	67.97	162.51	13.14	18.41	181.33	16.65	1.15	1 395.05
300	ACE	56.77	68.69	657.93	13.87	20.89	666.64	16.76	1.23	5 205.41
	GSS	35.58	44.03	3 947.58	6.07	9.63	3 999.81	8.57	1.32	31 232.48
	LwF	50.74	61.95	877.24	13.83	22.82	888.85	17.64	1.01	5 783.41
	MIR	46.63	59.62	1 644.83	9.64	15.33	1 666.60	12.91	0.87	13 013.54
	RWalk	45.61	57.88	1 315.86	9.75	14.86	1 333.28	13.21	1.13	10 410.83
	ER	54.15	65.82	657.93	12.82	20.09	666.64	15.70	1.13	5 205.41
	X-DER	54.35	66.16	822.42	13.28	20.75	833.30	15.59	1.14	6 506.77
	EARL	55.03	66.71	1 291.83	14.62	21.83	1 308.93	16.28	1.16	10 220.71
	DPCL	55.29	65.93	756.76	14.28	21.39	766.77	16.03	1.14	5 987.27
	本文方法	57.17	68.81	162.51	15.02	23.32	181.33	17.66	1.31	1 395.05
500	ACE	57.32	69.62	657.93	14.39	21.21	666.64	17.82	0.90	5 205.41
	GSS	35.86	44.13	3 947.58	6.41	9.66	3 999.81	8.28	1.70	31 232.48
	LwF	51.22	62.53	877.24	14.47	23.26	888.85	18.29	1.10	5 783.41
	MIR	45.96	57.56	1 644.83	10.38	15.08	1 666.60	13.52	1.23	13 013.54
	RWalk	45.59	56.68	1 315.86	10.33	15.61	1 333.28	13.89	1.06	10 410.83
	ER	55.53	65.09	657.93	13.13	20.95	666.64	16.62	0.94	5 205.41
	X-DER	55.94	66.30	822.42	13.94	21.04	833.30	16.40	0.94	6 506.77
	EARL	56.28	67.83	1 291.83	14.02	21.38	1 308.93	17.03	1.01	10 220.71
	DPCL	55.38	66.94	756.76	13.84	20.93	766.77	16.93	0.96	5 987.27
	本文方法	57.79	70.33	162.51	15.58	22.75	181.33	18.41	1.76	1 395.05

内存缓冲区大小	方法	CIFAR-10			CIFAR-100			Mini-ImageNet
内存缓冲区大小	方法	AOA/%	TA/%	运算浮点数/GFLOPs	AOA/%	TA/%	运算浮点数/GFLOPs	AOA/%	TA/%	运算浮点数/GFLOPs
100	ACE	53.05	67.01	657.93	12.15	17.92	666.64	16.07	0.99	5 205.41
	GSS	33.67	43.69	3 947.58	5.45	7.86	3 999.81	8.43	1.16	31 232.48
	LwF	49.08	59.07	877.24	10.99	17.98	888.85	14.85	0.75	5 783.41
	MIR	43.55	55.24	1 644.83	9.31	14.57	1 666.60	12.62	0.75	13 013.54
	RWalk	42.22	54.72	1 315.86	7.46	12.23	1 333.28	11.35	1.25	10 410.83
	ER	51.60	63.97	657.93	11.55	17.86	666.64	14.13	0.92	5 205.41
	X-DER	52.39	64.39	822.42	11.95	17.38	833.30	14.85	0.93	6 506.77
	EARL	52.82	64.93	1 291.83	12.01	17.72	1 308.93	15.41	1.04	10 220.71
	DPCL	52.73	64.29	756.76	11.98	17.93	766.77	15.38	0.93	5 987.27
	本文方法	54.11	67.97	162.51	13.14	18.41	181.33	16.65	1.15	1 395.05
300	ACE	56.77	68.69	657.93	13.87	20.89	666.64	16.76	1.23	5 205.41
	GSS	35.58	44.03	3 947.58	6.07	9.63	3 999.81	8.57	1.32	31 232.48
	LwF	50.74	61.95	877.24	13.83	22.82	888.85	17.64	1.01	5 783.41
	MIR	46.63	59.62	1 644.83	9.64	15.33	1 666.60	12.91	0.87	13 013.54
	RWalk	45.61	57.88	1 315.86	9.75	14.86	1 333.28	13.21	1.13	10 410.83
	ER	54.15	65.82	657.93	12.82	20.09	666.64	15.70	1.13	5 205.41
	X-DER	54.35	66.16	822.42	13.28	20.75	833.30	15.59	1.14	6 506.77
	EARL	55.03	66.71	1 291.83	14.62	21.83	1 308.93	16.28	1.16	10 220.71
	DPCL	55.29	65.93	756.76	14.28	21.39	766.77	16.03	1.14	5 987.27
	本文方法	57.17	68.81	162.51	15.02	23.32	181.33	17.66	1.31	1 395.05
500	ACE	57.32	69.62	657.93	14.39	21.21	666.64	17.82	0.90	5 205.41
	GSS	35.86	44.13	3 947.58	6.41	9.66	3 999.81	8.28	1.70	31 232.48
	LwF	51.22	62.53	877.24	14.47	23.26	888.85	18.29	1.10	5 783.41
	MIR	45.96	57.56	1 644.83	10.38	15.08	1 666.60	13.52	1.23	13 013.54
	RWalk	45.59	56.68	1 315.86	10.33	15.61	1 333.28	13.89	1.06	10 410.83
	ER	55.53	65.09	657.93	13.13	20.95	666.64	16.62	0.94	5 205.41
	X-DER	55.94	66.30	822.42	13.94	21.04	833.30	16.40	0.94	6 506.77
	EARL	56.28	67.83	1 291.83	14.02	21.38	1 308.93	17.03	1.01	10 220.71
	DPCL	55.38	66.94	756.76	13.84	20.93	766.77	16.93	0.96	5 987.27
	本文方法	57.79	70.33	162.51	15.58	22.75	181.33	18.41	1.76	1 395.05

稀疏度	AOA/%	TA/%	运算浮点数/GFLOPs
Adaptive	13.14	18.41	181.33
0.1	11.82	18.93	599.97
0.2	10.69	15.26	533.31
0.3	10.41	15.60	466.65
0.4	10.21	14.89	399.98
0.5	10.08	15.33	333.32
0.6	9.74	15.03	266.66
0.7	9.62	14.73	199.99
0.8	9.25	14.44	133.33
0.9	8.47	13.42	66.66

稀疏度	AOA/%	TA/%	运算浮点数/GFLOPs
Adaptive	13.14	18.41	181.33
0.1	11.82	18.93	599.97
0.2	10.69	15.26	533.31
0.3	10.41	15.60	466.65
0.4	10.21	14.89	399.98
0.5	10.08	15.33	333.32
0.6	9.74	15.03	266.66
0.7	9.62	14.73	199.99
0.8	9.25	14.44	133.33
0.9	8.47	13.42	66.66

方法	CIFAR-10	CIFAR-100	Mini-ImageNet
基线方法	1 094 750	1 109 240	11 227 812
本文方法	270 403	301 713	3 009 054

高速数据流下无边界在线稀疏连续学习方法

Task-free online sparse continual learning method for high-speed data streams

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 32

相关文章 13

编辑推荐

Metrics

[1]	郭书君, 任卫军, 陈倩倩, 游广飞. 基于聚类多变量时间序列模型的交通状态实时预测[J]. 《计算机应用》唯一官方网站, 2025, 45(7): 2253-2261.
[2]	董永峰, 王雅琮, 董瑶, 邓亚晗. 在线学习资源推荐综述[J]. 《计算机应用》唯一官方网站, 2023, 43(6): 1655-1663.
[3]	刘清华, 廖士中. 基于随机素描方法的在线核回归[J]. 《计算机应用》唯一官方网站, 2022, 42(3): 676-682.
[4]	樊海玮, 张锐驰, 安毅生, 秦佳杰. 融合知识图谱邻居双端的在线学习资源推荐算法[J]. 《计算机应用》唯一官方网站, 2022, 42(10): 3054-3059.
[5]	郭一村, 陈华辉. 在线哈希算法研究综述[J]. 《计算机应用》唯一官方网站, 2021, 41(4): 1106-1112.
[6]	崔艺馨, 陈晓东. Spark框架优化的大规模谱聚类并行算法[J]. 计算机应用, 2020, 40(1): 168-172.
[7]	吴婉婷, 朱燕, 黄定江. 在线投资组合选择的半指数梯度策略及实证分析[J]. 计算机应用, 2019, 39(8): 2462-2467.
[8]	郁顺昌, 黄定江. 基于自回归移动平均反转的在线投资组合选择[J]. 计算机应用, 2018, 38(5): 1505-1511.
[9]	高庆吉, 霍璐, 牛国臣. 基于改进霍夫森林框架的多目标跟踪算法[J]. 计算机应用, 2016, 36(8): 2311-2315.
[10]	刘建华, 毛可飞, 胡俊伟. 基于双线性对的无证书聚合签密方案[J]. 计算机应用, 2016, 36(6): 1558-1562.
[11]	许浩锋, 凌青. 分布式在线交替方向乘子法[J]. 计算机应用, 2015, 35(6): 1595-1599.
[12]	李唯实毛晓光谢建文. 学习者知识模型的在线学习算法[J]. 计算机应用, 2012, 32(02): 436-439.
[13]	吴宗亮窦衡. 一种新的最小二乘支持向量机稀疏化算法[J]. 计算机应用, 2009, 29(06): 1559-1581.

组号	MWI	AS	CGI	AOA/%	TA/%	运算浮点数/GFLOPs
1	×	×	×	11.55	17.86	666.64
2	×	×	√	12.82	20.28	577.75
3	√	×	×	10.74	16.78	622.20
4	×	√	√	13.46	18.77	343.10
5	√	×	√	12.20	18.05	533.31
6	√	√	×	12.29	18.92	504.87
7	√	√	√	13.14	18.41	181.33

组号	MWI	AS	CGI	AOA/%	TA/%	运算浮点数/GFLOPs
1	×	×	×	11.55	17.86	666.64
2	×	×	√	12.82	20.28	577.75
3	√	×	×	10.74	16.78	622.20
4	×	√	√	13.46	18.77	343.10
5	√	×	√	12.20	18.05	533.31
6	√	√	×	12.29	18.92	504.87
7	√	√	√	13.14	18.41	181.33