-
A Brief Survey on Autonomous Vehicle Possible Attacks, Exploits and Vulnerabilities
Authors:
Amara Dinesh Kumar,
Koti Naga Renu Chebrolu,
Vinayakumar R,
Soman KP
Abstract:
Advanced driver assistance systems are advancing at a rapid pace and all major companies started investing in develo** the autonomous vehicles. But the security and reliability is still uncertain and debatable. Imagine that a vehicle is compromised by the attackers and then what they can do. An attacker can control brake, accelerate and even steering which can lead to catastrophic consequences.…
▽ More
Advanced driver assistance systems are advancing at a rapid pace and all major companies started investing in develo** the autonomous vehicles. But the security and reliability is still uncertain and debatable. Imagine that a vehicle is compromised by the attackers and then what they can do. An attacker can control brake, accelerate and even steering which can lead to catastrophic consequences. This paper gives a very short and brief overview of most of the possible attacks on autonomous vehicle software and hardware and their potential implications.
△ Less
Submitted 3 October, 2018;
originally announced October 2018.
-
DeepImageSpam: Deep Learning based Image Spam Detection
Authors:
Amara Dinesh Kumar,
Vinayakumar R,
Soman KP
Abstract:
Hackers and spammers are employing innovative and novel techniques to deceive novice and even knowledgeable internet users. Image spam is one of such technique where the spammer varies and changes some portion of the image such that it is indistinguishable from the original image fooling the users. This paper proposes a deep learning based approach for image spam detection using the convolutional…
▽ More
Hackers and spammers are employing innovative and novel techniques to deceive novice and even knowledgeable internet users. Image spam is one of such technique where the spammer varies and changes some portion of the image such that it is indistinguishable from the original image fooling the users. This paper proposes a deep learning based approach for image spam detection using the convolutional neural networks which uses a dataset with 810 natural images and 928 spam images for classification achieving an accuracy of 91.7% outperforming the existing image processing and machine learning techniques
△ Less
Submitted 3 October, 2018;
originally announced October 2018.
-
Novel Deep Learning Model for Traffic Sign Detection Using Capsule Networks
Authors:
Amara Dinesh Kumar
Abstract:
Convolutional neural networks are the most widely used deep learning algorithms for traffic signal classification till date but they fail to capture pose, view, orientation of the images because of the intrinsic inability of max pooling layer.This paper proposes a novel method for Traffic sign detection using deep learning architecture called capsule networks that achieves outstanding performance…
▽ More
Convolutional neural networks are the most widely used deep learning algorithms for traffic signal classification till date but they fail to capture pose, view, orientation of the images because of the intrinsic inability of max pooling layer.This paper proposes a novel method for Traffic sign detection using deep learning architecture called capsule networks that achieves outstanding performance on the German traffic sign dataset.Capsule network consists of capsules which are a group of neurons representing the instantiating parameters of an object like the pose and orientation by using the dynamic routing and route by agreement algorithms.unlike the previous approaches of manual feature extraction,multiple deep neural networks with many parameters,our method eliminates the manual effort and provides resistance to the spatial variances.CNNs can be fooled easily using various adversary attacks and capsule networks can overcome such attacks from the intruders and can offer more reliability in traffic sign detection for autonomous vehicles.Capsule network have achieved the state-of-the-art accuracy of 97.6% on German Traffic Sign Recognition Benchmark dataset (GTSRB).
△ Less
Submitted 11 May, 2018;
originally announced May 2018.
-
A Two Stage Selective Averaging LDPC Decoding
Authors:
A. Dinesh Kumar,
Ambedkar Dukkipati
Abstract:
Low density parity-check (LDPC) codes are a class of linear block codes that are decoded by running belief propagation (BP) algorithm or log-likelihood ratio belief propagation (LLR-BP) over the factor graph of the code. One of the disadvantages of LDPC codes is the onset of an error floor at high values of signal to noise ratio caused by trap** sets. In this paper, we propose a two stage decode…
▽ More
Low density parity-check (LDPC) codes are a class of linear block codes that are decoded by running belief propagation (BP) algorithm or log-likelihood ratio belief propagation (LLR-BP) over the factor graph of the code. One of the disadvantages of LDPC codes is the onset of an error floor at high values of signal to noise ratio caused by trap** sets. In this paper, we propose a two stage decoder to deal with different types of trap** sets. Oscillating trap** sets are taken care by the first stage of the decoder and the elementary trap** sets are handled by the second stage of the decoder. Simulation results on regular PEG (504,252,3,6) code shows that the proposed two stage decoder performs significantly better than the standard decoder.
△ Less
Submitted 14 July, 2011;
originally announced July 2011.
