Filters








25,354 Hits in 5.2 sec

Deploying Deep Neural Networks in the Embedded Space [article]

Stylianos I. Venieris, Alexandros Kouris, Christos-Savvas Bouganis
2018 arXiv   pre-print
Recently, Deep Neural Networks (DNNs) have emerged as the dominant model across various AI applications.  ...  In the era of IoT and mobile systems, the efficient deployment of DNNs on embedded platforms is vital to enable the development of intelligent applications.  ...  CONCLUSION The presented set of works focuses on bridging the gap between the deep learning community and the deployment of models in the embedded space.  ... 
arXiv:1806.08616v1 fatcat:52xugpvnkzeuloufnsfp673oo4

TinyML for Ubiquitous Edge AI [article]

Stanislava Soro
2021 arXiv   pre-print
TinyML addresses the challenges in designing power-efficient, compact deep neural network models, supporting software framework, and embedded hardware that will enable a wide range of customized, ubiquitous  ...  TinyML is a fast-growing multidisciplinary field at the intersection of machine learning, hardware, and software, that focuses on enabling deep learning algorithms on embedded (microcontroller powered)  ...  The two key challenges in deploying neural networks on microcontrollers are the small memory size and the short battery life [5] .  ... 
arXiv:2102.01255v1 fatcat:if5ny6kcirdkhnj56mswfaptlm

Imageem: Pre-Trained Encoded Vector Embeddings for Image Modelling

2020 VOLUME-8 ISSUE-10, AUGUST 2019, REGULAR ISSUE  
Our aim is to extract the features of the images by training on these vector embeddings locally and then deploying it in the server for easy access to the researchers.  ...  These Image modelling problems are solved by training a Convolutional Neural Network (CNN) which is computationally expensive process.  ...  of deep learning and Artificial Intelligence in general.  ... 
doi:10.35940/ijitee.e1999.059720 fatcat:jctpzudz2rebzirn6pqdcafhku

Deep Koopman Operator with Control for Nonlinear Systems [article]

Haojie Shi, Max Q.-H. Meng
2022 arXiv   pre-print
We first parameterize the embedding function and Koopman Operator with the neural network and train them end-to-end with the K-steps loss function.  ...  This encoded term is considered the new control variable instead to ensure linearity of the modeled system in the embedding system.We next deploy Linear Quadratic Regulator (LQR) on the linear embedding  ...  In contrast, deep learning approaches promote the learning of embedding functions with deep neural networks [2] , [14] , improving the prediction quality for a long time horizon.  ... 
arXiv:2202.08004v2 fatcat:eapumwddnzgmzk2iw7s6ne63xe

Memory Optimized Deep Learning based Face Recognization

Amit Kumar Mr., Minakshi Memoria Dr., Vinod Kumar Dr.
2021 Indian Journal of Computer Science and Engineering  
In our tests, we ran our proposed Network Architecture on the network at some very memory constrained systems (as low as 2 MB).  ...  We present a novel and "very efficient" Network Architecture in this paper that includes MobileNetV2 with center loss and training tricks for "deep face recognition" for an embedded system that is memory  ...  The CNN (Convolutional Neural Networks) is at the core of almost all the devices. The use of CNN and AI / ML is not the new thing in Face recognition.  ... 
doi:10.21817/indjcse/2021/v12i1/211201066 fatcat:v3tumrl3sfcdxjbil7swohsiny

A Low-cost Artificial Neural Network Model for Raspberry Pi

S. N. Truong
2020 Zenodo  
The proposed ternary neural network is deployed on a low-cost Raspberry Pi board embedded system for the application of speech and image recognition.  ...  In conventional neural networks, the signed synaptic weights of –1, 0, and 1 are represented by 8-bit integers.  ...  ACKNOWLEDGMENT This work is financially supported by the Ministry of Education and Training (Grant number B2019-SPK-05, 2019).  ... 
doi:10.5281/zenodo.3748348 fatcat:yn7nf2qv4jfy3nged4kdrszklu

Real-Time Driver Drowsiness Detection for Embedded System Using Model Compression of Deep Neural Networks

Bhargava Reddy, Ye-Hoon Kim, Sojung Yun, Chanwon Seo, Junik Jang
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
In this paper, a novel approach towards real-time drowsiness detection based on deep learning which can be implemented on a low cost embedded board and performs with a high accuracy is proposed.  ...  Moreover, minimized network structure was designed based on facial landmark input to recognize whether driver is drowsy or not.  ...  Conclusion and Future Works In this paper, highly optimized deep neural network model for driver's drowsiness detection is designed and compressed for embedded system.  ... 
doi:10.1109/cvprw.2017.59 dblp:conf/cvpr/ReddyKYSJ17 fatcat:7gh3ocfcpnekdcdng57wiqzini

Self-Governing Neural Networks for On-Device Short Text Classification

Sujith Ravi, Zornitsa Kozareva
2018 Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing  
Deep neural networks reach state-of-the-art performance for wide range of natural language processing, computer vision and speech applications.  ...  The key advantage of SGNNs over existing work is that they surmount the need for pre-trained word embeddings and complex networks with huge parameters.  ...  Acknowledgments We would like to thank the anonymous reviewers for their valuable feedback and suggestions.  ... 
doi:10.18653/v1/d18-1105 dblp:conf/emnlp/RaviK18a fatcat:4efmvk5skbd23mt4lwbgplidue

Self-Governing Neural Networks for On-Device Short Text Classification

Sujith Ravi, Zornitsa Kozareva
2018 Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing  
Deep neural networks reach state-of-the-art performance for wide range of natural language processing, computer vision and speech applications.  ...  The key advantage of SGNNs over existing work is that they surmount the need for pre-trained word embeddings and complex networks with huge parameters.  ...  Acknowledgments We would like to thank the anonymous reviewers for their valuable feedback and suggestions.  ... 
doi:10.18653/v1/d18-1092 dblp:conf/emnlp/RaviK18 fatcat:hieuxynlxbacfahkh3bfs5b2a4

Multimodal and Crossmodal Representation Learning from Textual and Visual Features with Bidirectional Deep Neural Networks for Video Hyperlinking

Vedran Vukotić, Christian Raymond, Guillaume Gravier
2016 Proceedings of the 2016 ACM workshop on Vision and Language Integration Meets Multimedia Fusion - iV&L-MM '16  
More recently, deep neural networks have been successfully deployed in tasks requiring consideration of multiple modalities.  ...  Bidirectional Deep Neural Networks In bidirectional deep neural networks, learning is performed in both directions: one modality is presented as an input and the other as the expected output while at the  ... 
doi:10.1145/2983563.2983567 dblp:conf/mm/VukoticRG16 fatcat:7esjh3ghczd5dnk55tn4hw4zxe

Bidirectional Joint Representation Learning with Symmetrical Deep Neural Networks for Multimodal and Crossmodal Applications

Vedran Vukotić, Christian Raymond, Guillaume Gravier
2016 Proceedings of the 2016 ACM on International Conference on Multimedia Retrieval - ICMR '16  
By tying the weights of two deep neural networks, symmetry is enforced in central hidden layers thus yielding a multimodal representation space common to the two original representation spaces.  ...  Recently, deep neural networks, especially deep autoencoders, have proven promising both for crossmodal translation and for early fusion via multimodal embedding.  ...  More recently, deep neural networks have been successfully deployed in tasks requiring consideration of multiple modalities.  ... 
doi:10.1145/2911996.2912064 dblp:conf/mir/VukoticRG16 fatcat:cqepgneporfrfhyosjiwkghqg4

Embed Everything: A Method for Efficiently Co-Embedding Multi-Modal Spaces [article]

Sarah Di, Robin Yu, Amol Kapoor
2021 arXiv   pre-print
In the last decade, deep neural networks have seen remarkable success in unimodal data distributions, while transfer learning techniques have seen a massive expansion of model reuse across related domains  ...  We prove the use of this system in a joint image-audio embedding task.  ...  We deploy contrastive losses to train a small deep neural network that projects between the preexisting model spaces.  ... 
arXiv:2110.04599v1 fatcat:xtkozzcxmrdnbgrx3qompyjmai

Visual Analytics in Deep Learning: An Interrogative Survey for the Next Frontiers [article]

Fred Hohman, Minsuk Kahng, Robert Pienta, Duen Horng Chau
2018 arXiv   pre-print
However, because of the internal complexity and nonlinear structure of deep neural networks, the underlying decision making processes for why these models are achieving such performance are challenging  ...  We present a survey of the role of visual analytics in deep learning research, which highlights its short yet impactful history and thoroughly summarizes the state-of-the-art using a human-centered interrogative  ...  Once a neural network model has been trained, one can compute the activations for a given test dataset and visualize the activations in the Embedding Projector to visualize and explore the space that the  ... 
arXiv:1801.06889v3 fatcat:c5x3ftcf5fbapc5tsyhm5w2dhq

HG-Caffe: Mobile and Embedded Neural Network GPU (OpenCL) Inference Engine with FP16 Supporting [article]

Zhuoran Ji
2019 arXiv   pre-print
In this paper, we present a deep neural network inference engine named HG-Caffe, which supports GPUs with half precision.  ...  However, deep neural networks inference is still a challenging task for edge AI devices due to the computational overhead on mobile CPUs and a severe drain on the batteries.  ...  The reason is that, in Kirin 970, the memory space allocated to GPU computing kernel is limited to around 100 MB, which is too small for deep neural networks.  ... 
arXiv:1901.00858v1 fatcat:h66j7hes2vgmhapnb5uotazv5y

Machine Learning at the Network Edge: A Survey [article]

M.G. Sarwar Murshed, Christopher Murphy, Daqing Hou, Nazar Khan, Ganesh Ananthanarayanan, Faraz Hussain
2021 arXiv   pre-print
This survey describes major research efforts where machine learning systems have been deployed at the edge of computer networks, focusing on the operational aspects including compression techniques, tools  ...  To address this issue, efforts have been made to place additional computing devices at the edge of the network, i.e close to the IoT devices where the data is generated.  ...  Song et al. developed DRQ, a dynamic region-based Distributed deep neural network architectures allow the distribution of deep neural networks (DNNs) on the edgecloud infrastructure in order to facilitate  ... 
arXiv:1908.00080v4 fatcat:mw4lwwvzf5gupjr6pgdgnabeuu
« Previous Showing results 1 — 15 out of 25,354 results