{"title":"A Buyer-traceable DNN Model IP Protection Method Against Piracy and Misappropriation","authors":"Si Wang, Chaohui Xu, Yue Zheng, Chip-Hong Chang","doi":"10.1109/AICAS54282.2022.9869923","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9869923","url":null,"abstract":"Recently proposed model functionality and attribute extraction techniques have exacerbated unauthorized low-cost reproduction of deep neural network (DNN) models for similar applications. In particular, intellectual property (IP) theft and unauthorized distribution of DNN models by dishonest buyers are very difficult to trace by existing framework of digital rights management (DRM). This paper presents a new buyer-traceable DRM scheme against model piracy and misappropriation. Unlike existing methods that require white-box access to extract the latent information for verification, the proposed method utilizes data poisoning for distributorship embedding and black-box verification. Composite backdoors are installed into the target model during the training process. Each backdoor is created by applying a data augmentation method to some clean images of a selected class. The data-augmented images with a wrong label associated with a buyer are injected into the training dataset. The ownership and distributorship of a backdoor-trained user model can be validated by querying the suspect model with a set of composite triggers. A positive suspect will output the dirty labels that pinpoint the dishonest buyer while an innocent model will output the correct labels with high confidence. The tracking accuracy and robustness of the proposed IP protection method are evaluated on CIFAR-10, CIFAR-100 and GTSRB datasets for different applications. The results show an average of 100% piracy detection rate, 0% false positive rate and 96.81 % traitor tracking success rate with negligible model accuracy degradation.","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"94 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115632486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Hand Gesture Recognition Using IR-UWB Radar with Spiking Neural Networks","authors":"Wang Shule, Yulong Yan, Haoming Chu, Guangxi Hu, Zhi Zhang, Zhuo Zou, Lirong Zheng","doi":"10.1109/AICAS54282.2022.9870013","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9870013","url":null,"abstract":"Hand gesture recognition has emerged in recent years as a robust method in non-contact human-computer interfaces, especially in the application scenario of the Internet of Things. This paper proposes a high-accuracy and low-power algorithm for hand gesture recognition. The hand gesture dataset was collected by Integrated Systems Lab at ETH Zurich using a low-cost impulse radio ultra-wideband (IR-UWB) radar. The signals are transformed into spikes sequence by time-event coding and level-crossing sampling. These spike arrays are processed by spiking neural networks (SNNs), which have more biological interpretability and are inherently suitable for processing time-series signals. The algorithm has achieved 95.44% accuracy in 5 hand gestures and 96.60% accuracy in 6 hand gestures. As for power consumption, the classification network operates 350 kFLOPs per data sequence on 5 hand gesture datasets, which is 90× smaller than the previous approach.","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121113595","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Lightweight and Efficient Neural Network Using Progressively Greedy Search","authors":"Jheng-Yi Chang, Ching-Te Chiu, Pin-Hsuan Chen","doi":"10.1109/AICAS54282.2022.9869968","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9869968","url":null,"abstract":"In this paper, we propose a progressively greedy search (PGS) method for lightweight and efficient neural networks. We search neural network architectures in a restricted design space to reduce the search cost without sampling models. The searched models have regular weights which is similar to structural pruning. Our search strategy can automatically find a suitable channel number in each layer without manual setting. We pay more attention to the architecture, not the model weights, which is different from the pruning methods. Compared to RegNetX-200MF [1], our searched model, PGS 178MF, reduces 21.65MFLOPs and increases 0.93% accuracy in the same training cost. Compared to conventional width scaling, the architecture we have searched on MobileNet-V3-large improves accuracy by 0.72% under a similar model size.","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123560819","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jiawei Liao, L. Widmer, Xiaying Wang, Alfio Di Mauro, Samuel R. Nason-Tomaszewski, C. Chestek, L. Benini, Taekwang Jang
{"title":"An Energy-Efficient Spiking Neural Network for Finger Velocity Decoding for Implantable Brain-Machine Interface","authors":"Jiawei Liao, L. Widmer, Xiaying Wang, Alfio Di Mauro, Samuel R. Nason-Tomaszewski, C. Chestek, L. Benini, Taekwang Jang","doi":"10.1109/AICAS54282.2022.9869846","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9869846","url":null,"abstract":"Brain-machine interfaces (BMIs) are promising for motor rehabilitation and mobility augmentation. High-accuracy and low-power algorithms are required to achieve implantable BMI systems. In this paper, we propose a novel spiking neural network (SNN) decoder for implantable BMI regression tasks. The SNN is trained with enhanced spatio-temporal backpropagation to fully leverage its ability in handling temporal problems. The proposed SNN decoder achieves the same level of correlation coefficient as the state-of-the-art ANN decoder in offline finger velocity decoding tasks, while it requires only 6.8% of the computation operations and 9.4% of the memory access.","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"89 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127024148","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"TAC-RAM: A 65nm 4Kb SRAM Computing-in-Memory Design with 57.55 TOPS/W supporting Multibit Matrix-Vector Multiplication for Binarized Neural Network","authors":"Xiaomeng Wang, Xuejiao Liu, Xianghong Hu, Xiaopeng Zhong, Xizi Chen, Yu Liu, Patrick Kong, Fengshi Tian, Chiying Tsui","doi":"10.1109/AICAS54282.2022.9869970","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9869970","url":null,"abstract":"In this paper, an energy-efficient computing-in-memory architecture using 10T-SRAM cells is proposed to resolve the design challenges of mixed-signal Computing-in-memory (CIM) schemes. A time-assisted current-based (TAC-based) MAC and ADC mechanism is developed to maintain good linearity throughout the multiplication and AD conversion process. A binary-search discharge-based ADC (DB-ADC) ensures high energy efficiency and throughput in high precision mixed-signal computations. A 4Kb SRAM array is fabricated in TSMC 65nm CMOS process and measurement results show that a throughput of 24.1 GOPS and a peak energy efficiency of 57.55 TOPS/W are achieved with multi-bit MAC operation for 4-bit input, 1-bit binarized weight, and 11-bit signed output.","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"52 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116561201","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Y. Chung, Hao Lian, Yong-Lun Xiao, Chih-Tsun Huang, J. Liou
{"title":"Fast DNN-based Mechatronics Prototyping Platform on Robotic Arm Control","authors":"Y. Chung, Hao Lian, Yong-Lun Xiao, Chih-Tsun Huang, J. Liou","doi":"10.1109/AICAS54282.2022.9869932","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9869932","url":null,"abstract":"In industrial applications, a robotic controller re-quires a low-latency computation process for real-time con-straints. In the meantime, more controllers are designed with DNN-based reinforcement learning, which needs increasing computation power. In this demo, we developed a fast prototyping infrastructure in AI -based mechatronics. Our software/hardware co-optimization incorporates a cyber-physical system (CPS), a host computer, and a DNN-based accelerator on an FPGA. The holistic accelerator is built upon the ESP SoC (System-on-Chip) platform with the high-level synthesis (HLS) technique and an improved interface. Our demonstration on an intelligent robotic arm showcases 101 times speedup over a CPU-based software implementation.","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"298 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122710491","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Burrello, Giacomo Sintoni, D. Brunelli, L. Benini
{"title":"Adversarially-Trained Tiny Autoencoders for Near-Sensor Continuous Structural Health Monitoring","authors":"A. Burrello, Giacomo Sintoni, D. Brunelli, L. Benini","doi":"10.1109/AICAS54282.2022.9869952","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9869952","url":null,"abstract":"Structural Health Monitoring (SHM) systems are increasingly employed in many civil structures such as buildings, tunnels and viaducts. Typical installations consist of sensors that gather information and send it to a central computing unit, which then periodically analyzes the incoming data and produces an assessment of the structure conditions. To avoid the transmission of a huge amount of raw data and reduce latency in the detection of structural anomalies, recent works focus on moving computation on the sensor nodes. This work shows that a small autoencoder, which fits the tiny 2 MB memory of a typical microcontroller used for SHM sensor nodes can achieve very competitive accuracy in detecting structural anomalies as well as vehicle passage on bridges by leveraging adversarial training based on generative adversarial networks (GANs). We improve accuracy over state-of-the-art algorithms in two use-cases on real-standing buildings: i) predicting anomalies on a bridge (+7.4%) and ii) detecting vehicles on a viaduct (2.30 ×).","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122065597","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A General-Purpose and Configurable Planar Data Processor for Energy-Efficient Pooling Computation","authors":"Lunshuai Pan, Peng Xue, Hongxing Li, Litao Sun, Mingqiang Huang","doi":"10.1109/AICAS54282.2022.9869992","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9869992","url":null,"abstract":"Convolutional Neural Networks (CNN) have been widely used in artificial intelligence applications. A typical CNN contains both convolution and pooling layer, in which the convolution is to detect local conjunctions of features and the pooling is to merge similar patterns into one. It is necessary to make pooling operation, which plays a great role in CNN. Up to now, there have been numerous researches on CNN accelerators, however, most of the previous works are only focused on the acceleration of convolution layers, and the specific studies on pooling units are still lacking. Besides, the existing pooling designs are usually constrained by either the poor flexibility or the low energy/area efficiency. In this work, we propose a general purpose and energy-efficient planar data processor to support the pooling operation from different CNN structure. By using the configurable data path control method, the processor is able to support universal pooling operation with arbitrary input feature shape and arbitrary pooling kernel/stride/padding size. Besides, the processor exhibits high efficiency with hardware utilization ratio near 100% during operation, indicating good performance of the design. Most importantly, it is energy-efficient that exhibits 86%-off on power consumption and 62%-off on area utilization when compared with the separate pooling module of NVDLA (NVIDIA Deep Learning Accelerator), thus is particularly suitable for the resource-limited edge intelligent devices.","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"420 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123400201","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ying-i Wu, Mingfu Xue, Dujuan Gu, Yushu Zhang, Weiqiang Liu
{"title":"Sample-Specific Backdoor based Active Intellectual Property Protection for Deep Neural Networks","authors":"Ying-i Wu, Mingfu Xue, Dujuan Gu, Yushu Zhang, Weiqiang Liu","doi":"10.1109/AICAS54282.2022.9869927","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9869927","url":null,"abstract":"Recently, a number of researches have been proposed to protect the intellectual property (IP) of Deep Neural Network (DNN) models. However, most existing works are passive protection methods as they attempt to extract watermark from the pirated model after piracy occurs. In this paper, we propose an active IP protection method for DNN in which we utilize a variant of sample-specific backdoor attack to implement active authorization control for DNN models. During training, we mislabel all the clean images and keep the labels of backdoor instances as their ground-truth labels. Different from general backdoor trigger, we train a U-Net model to generate sample-specific trigger. This kind of trigger is sample-specific and invisible, which works as the secret key for each image and is hard to be noticed. Moreover, compared with existing active DNN IP protection methods, the proposed method can be applied in the black-box scenario. Experimental results on ImageNet and YouTube Aligned Face datasets demonstrate the effectiveness and robustness of the proposed method.","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134499822","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Improving Embedded Target Tracking Systems Based on Siamese Networks with Infrared Images","authors":"S. Horng, Yun Zeng","doi":"10.1109/AICAS54282.2022.9869937","DOIUrl":"https://doi.org/10.1109/AICAS54282.2022.9869937","url":null,"abstract":"Whether it is for hunting prey, or modern racing, games, or war, how to accurately track the target is the key to the victory. When we can have a machine to help us track the target, we can naturally solve more problems. Not only it can be used in military applications, such as warplane radar and missiles but it can be applied in civilian purposes including home security, event broadcasting, smart home. The target tracking is becoming increasingly mature. In this paper, we aim to develop a practical three-axis target tracking system by simulating the hunting ability of snakes in nature, and we add thermal infrared images to increase the model modality and optimize the ability of existing Siamese tracking network. Multi-Modal Machine Learning (MMML) is modified with the Siamese tracking network into a multi-modal fusion model, so as to improve the model's perception. At the same time, the Siamese networks models for training normal images and thermal infrared images are separated and a pseudo-Siamese network is used as the main framework to process the different the databases. This paper successfully improves the stability and accuracy of the model. With the embedded development version and three-axis servo motor, our system can be more flexible in tracking applications.","PeriodicalId":169690,"journal":{"name":"2022 IEEE 4th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115649358","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}