-
FedUKD: Federated UNet Model with Knowledge Distillation for Land Use Classification from Satellite and Street Views
Authors:
Renuga Kanagavelu,
Kinshuk Dua,
Pratik Garai,
Susan Elias,
Neha Thomas,
Simon Elias,
Qingsong Wei,
Goh Siow Mong Rick,
Liu Yong
Abstract:
Federated Deep Learning frameworks can be used strategically to monitor Land Use locally and infer environmental impacts globally. Distributed data from across the world would be needed to build a global model for Land Use classification. The need for a Federated approach in this application domain would be to avoid transfer of data from distributed locations and save network bandwidth to reduce c…
▽ More
Federated Deep Learning frameworks can be used strategically to monitor Land Use locally and infer environmental impacts globally. Distributed data from across the world would be needed to build a global model for Land Use classification. The need for a Federated approach in this application domain would be to avoid transfer of data from distributed locations and save network bandwidth to reduce communication cost. We use a Federated UNet model for Semantic Segmentation of satellite and street view images. The novelty of the proposed architecture is the integration of Knowledge Distillation to reduce communication cost and response time. The accuracy obtained was above 95% and we also brought in a significant model compression to over 17 times and 62 times for street View and satellite images respectively. Our proposed framework has the potential to be a game-changer in real-time tracking of climate change across the planet.
△ Less
Submitted 5 December, 2022;
originally announced December 2022.
-
HyBNN and FedHyBNN: (Federated) Hybrid Binary Neural Networks
Authors:
Kinshuk Dua
Abstract:
Binary Neural Networks (BNNs), neural networks with weights and activations constrained to -1(0) and +1, are an alternative to deep neural networks which offer faster training, lower memory consumption and lightweight models, ideal for use in resource constrained devices while being able to utilize the architecture of their deep neural network counterpart. However, the input binarization step used…
▽ More
Binary Neural Networks (BNNs), neural networks with weights and activations constrained to -1(0) and +1, are an alternative to deep neural networks which offer faster training, lower memory consumption and lightweight models, ideal for use in resource constrained devices while being able to utilize the architecture of their deep neural network counterpart. However, the input binarization step used in BNNs causes a severe accuracy loss. In this paper, we introduce a novel hybrid neural network architecture, Hybrid Binary Neural Network (HyBNN), consisting of a task-independent, general, full-precision variational autoencoder with a binary latent space and a task specific binary neural network that is able to greatly limit the accuracy loss due to input binarization by using the full precision variational autoencoder as a feature extractor. We use it to combine the state-of-the-art accuracy of deep neural networks with the much faster training time, quicker test-time inference and power efficiency of binary neural networks. We show that our proposed system is able to very significantly outperform a vanilla binary neural network with input binarization. We also introduce FedHyBNN, a highly communication efficient federated counterpart to HyBNN and demonstrate that it is able to reach the same accuracy as its non-federated equivalent. We make our source code, experimental parameters and models available at: https://anonymous.4open.science/r/HyBNN.
△ Less
Submitted 19 May, 2022;
originally announced May 2022.
-
Implementation of an efficient, portable and platform-agnostic cryptocurrency mining algorithm for Internet of Things devices
Authors:
Kinshuk Dua
Abstract:
Recently, there has been a remarkable amount of research being done in both, the fields of Blockchain and Internet of Things (IoT). Blockchain technology synergises well with IoT, solving key problems such as privacy, concerns with interoperability and security. However, the consensus mechanisms that allows trustless parties to maintain an agreement, the same algorithms that underpins cryptocurren…
▽ More
Recently, there has been a remarkable amount of research being done in both, the fields of Blockchain and Internet of Things (IoT). Blockchain technology synergises well with IoT, solving key problems such as privacy, concerns with interoperability and security. However, the consensus mechanisms that allows trustless parties to maintain an agreement, the same algorithms that underpins cryptocurrency mining, are usually extremely computationally expensive, making implementation on low-power IoT devices difficult. More importantly, mining requires downloading and synchronizing hundred of gigabytes worth of blocks which is far beyond the capabilities of most IoT devices. In this paper, we present an efficient, portable and platform-agnostic cryptocurrency mining algorithm using the Stratum protocol to avoid downloading the entire blockchain. We implement the algorithm in four different platforms- PC, ESP32, an emulator and an old PlayStation Portable (PSP) to demonstrate that it is indeed possible for any device to mine cryptocurrencies with no assumptions except the ability to connect to the internet. To make sure of ease of portability on any platform and for reproducibility of the reported results we make the implementation publicly available with detailed instructions at: https://anonymous.4open.science/r/cryptominer.
△ Less
Submitted 3 May, 2022;
originally announced May 2022.
-
Java File Security System (JFSS)
Authors:
Dr. Brijender Kahanwal,
Kanishak Dua,
Dr. Girish Pal Singh
Abstract:
Nowadays, storage systems are increasingly subject to attacks. So the security system is quickly becoming mendatory feature of the data storage systems. For the security purpose we are always dependent on the cryptography techniques. These techniques take the performance costs for the complete system. So we have proposed the Java File Security System(JFSS). It is based on the on-demand computing s…
▽ More
Nowadays, storage systems are increasingly subject to attacks. So the security system is quickly becoming mendatory feature of the data storage systems. For the security purpose we are always dependent on the cryptography techniques. These techniques take the performance costs for the complete system. So we have proposed the Java File Security System(JFSS). It is based on the on-demand computing system concept, because of the performance issues. It is a greate comback for the system performance. The concept is used because, we are not always in need the secure the files, but the selected one only. In this paper, we have designed a file security system on Windows XP. When we use the operating system, we have to secure some important data. The date is always stored in the files, so we secure the important files well. To check the proposed functionality, we experiment the above said system on the Windows operating system. With these experiments, we have found that the proposed system is working properly, according to the needs of the users.
△ Less
Submitted 13 November, 2013;
originally announced November 2013.
