Abhishek

Ehsan Nowroozi, Abhishek, Mohammadreza Mohammadi et al.

Malicious advertisement URLs pose a security risk since they are the source of cyber-attacks, and the need to address this issue is growing in both industry and academia. Generally, the attacker delivers an attack vector to the user by means of an email, an advertisement link or any other means of communication and directs them to a malicious website to steal sensitive information and to defraud them. Existing malicious URL detection techniques are limited and to handle unseen features as well as generalize to test data. In this study, we extract a novel set of lexical and web-scrapped features and employ machine learning technique to set up system for fraudulent advertisement URLs detection. The combination set of six different kinds of features precisely overcome the obfuscation in fraudulent URL classification. Based on different statistical properties, we use twelve different formatted datasets for detection, prediction and classification task. We extend our prediction analysis for mismatched and unlabelled datasets. For this framework, we analyze the performance of four machine learning techniques: Random Forest, Gradient Boost, XGBoost and AdaBoost in the detection part. With our proposed method, we can achieve a false negative rate as low as 0.0037 while maintaining high accuracy of 99.63%. Moreover, we devise a novel unsupervised technique for data clustering using K- Means algorithm for the visual analysis. This paper analyses the vulnerability of decision tree-based models using the limited knowledge attack scenario. We considered the exploratory attack and implemented Zeroth Order Optimization adversarial attack on the detection models.

Vishnu S. Chipade, Abhishek, Mangal Kothari et al.

This paper discusses the conceptual design and proof-of-concept flight demonstration of a novel variable pitch quadrotor biplane Unmanned Aerial Vehicle concept for payload delivery. The proposed design combines vertical takeoff and landing (VTOL), precise hover capabilities of a quadrotor helicopter and high range, endurance and high forward cruise speed characteristics of a fixed wing aircraft. The proposed UAV is designed for a mission requirement of carrying and delivering 6 kg payload to a destination at 16 km from the point of origin. First, the design of proprotors is carried out using a physics based modified Blade Element Momentum Theory (BEMT) analysis, which is validated using experimental data generated for the purpose. Proprotors have conflicting requirement for optimal hover and forward flight performance. Next, the biplane wings are designed using simple lifting line theory. The airframe design is followed by power plant selection and transmission design. Finally, weight estimation is carried out to complete the design process. The proprotor design with 24 deg preset angle and -24 deg twist is designed based on 70% weightage to forward flight and 30% weightage to hovering flight conditions. The operating RPM of the proprotors is reduced from 3200 during hover to 2000 during forward flight to ensure optimal performance during cruise flight. The estimated power consumption during forward flight mode is 64% less than that required for hover, establishing the benefit of this hybrid concept. A proof-of-concept scaled prototype is fabricated using commercial-off-the-shelf parts. A PID controller is developed and implemented on the PixHawk board to enable stable hovering flight and attitude tracking.

Namrata Gupta, Mangal Kothari, Abhishek

This paper carries out the mathematical modeling, simulation, and control law design for a quadrotor with variable-pitch propellers. The use of variable-pitch propeller for thrust variation instead of RPM regulation facilitates generation of negative thrust, thereby augmenting the rate of change of thrust generation amenable for aggressive maneuvering. Blade element theory along with momentum theory is used to estimate propeller thrust and torque essential for formulating equation of motion of the vehicle. The proposed flight dynamics model is used for non-linear control design using dynamic inversion technique, which is then used to stabilize, track reference trajectory, and simulate flip maneuver. The rotor torque is an irrational function of the control input which makes the control design challenging. To address this problem, the control design employs three loops. The outer loop solves the translational dynamics to generate the thrust, pitch angle, and roll angle commands required to track the prescribed trajectory. Using the command generated in the outer loop, the inner loop simplifies the rotational dynamics to provide the desired rate of angular velocities. A control allocation loop is added to address the problem of nonlinearity associated with rotor torque. This is done by introducing the derivative of thrust coefficient as a virtual control input. These virtual inputs determine the derivatives of thrust and body moments, which in turn is used to generate the required thrust and body moments. The concept is validated by showing attitude stabilization in real flight for a variable pitch quadrotor. The performance of the proposed design is shown through simulated results for attitude stabilization and trajectory following. Reverse thrust capability of variable-pitch quadrotor is also shown by performing flip maneuver in which quadrotor roll angle changes from 0 to 180 degrees.

Nidhish Raj, Ravi N. Banavar, Abhishek et al.

We consider the problem of attitude tracking for small-scale aerobatic helicopters. A small scale helicopter has two subsystems: the fuselage, modeled as a rigid body; and the rotor, modeled as a first order system. Due to the coupling between rotor and fuselage, the complete system does not inherit the structure of a simple mechanical system. The coupled rotor fuselage dynamics is first transformed to rigid body attitude tracking problem with a first order actuator dynamics. The proposed controller is developed using geometric and backstepping control technique. The controller is globally defined on $SO(3)$ and is shown to be locally exponentially stable. The controller is validated in simulation and experiment for a 10 kg class small scale flybarless helicopter by demonstrating aggressive roll attitude tracking.

Nidhish Raj, Ravi N Banavar, Abhishek et al.

This paper highlights the significance of the rotor dynamics in control design for small-scale aerobatic helicopters, and proposes two singularity free robust attitude tracking controllers based on the available states for feedback. 1. The first, employs the angular velocity and the flap angle states (a variable that is not easy to measure) and uses a backstepping technique to design a robust compensator (BRC) to \textbf{\textit{actively}} suppress the disturbance induced tracking error. 2. The second exploits the inherent damping present in the helicopter dynamics leading to a structure preserving, \textbf{\textit{passively}} robust controller (SPR), which is free of angular velocity and flap angle feedback. The BRC controller is designed to be robust in the presence of two types of uncertainties: structured and unstructured. The structured disturbance is due to uncertainty in the rotor parameters, and the unstructured perturbation is modeled as an exogenous torque acting on the fuselage. The performance of the controller is demonstrated in the presence of both types of disturbances through numerical simulations. In contrast, the SPR tracking controller is derived such that the tracking error dynamics inherits the natural damping characteristic of the helicopter. The SPR controller is shown to be almost globally asymptotically stable and its performance is evaluated experimentally by performing aggressive flip maneuvers. Throughout the study, a nonlinear coupled rotor-fuselage helicopter model with first order flap dynamics is used.

Enrico Lupi, Abhishek, Max Aehle et al.

We simulate hadrons impinging on a homogeneous lead-tungstate (PbWO4) calorimeter to investigate how the resulting light yield and its temporal structure, as detected by an array of light-sensitive sensors, can be processed by a neuromorphic computing system. Our model encodes temporal photon distributions as spike trains and employs a fully connected spiking neural network to estimate the total deposited energy, as well as the position and spatial distribution of the light emissions within the sensitive material. The extracted primitives offer valuable topological information about the shower development in the material, achieved without requiring a segmentation of the active medium. A potential nanophotonic implementation using III-V semiconductor nanowires is discussed. It can be both fast and energy efficient.

Jatin Lamba, Abhishek, Jayaprakash Akula et al.

In this paper, we present a novel approach to the audio-visual video parsing (AVVP) task that demarcates events from a video separately for audio and visual modalities. The proposed parsing approach simultaneously detects the temporal boundaries in terms of start and end times of such events. We show how AVVP can benefit from the following techniques geared towards effective cross-modal learning: (i) adversarial training and skip connections (ii) global context aware attention and, (iii) self-supervised pretraining using an audio-video grounding objective to obtain cross-modal audio-video representations. We present extensive experimental evaluations on the Look, Listen, and Parse (LLP) dataset and show that we outperform the state-of-the-art Hybrid Attention Network (HAN) on all five metrics proposed for AVVP. We also present several ablations to validate the effect of pretraining, global attention and adversarial training.

Jayaprakash A, Abhishek, Rishabh Dabral et al.

Video retrieval using natural language queries requires learning semantically meaningful joint embeddings between the text and the audio-visual input. Often, such joint embeddings are learnt using pairwise (or triplet) contrastive loss objectives which cannot give enough attention to 'difficult-to-retrieve' samples during training. This problem is especially pronounced in data-scarce settings where the data is relatively small (10% of the large scale MSR-VTT) to cover the rather complex audio-visual embedding space. In this context, we introduce Rudder - a multilingual video-text retrieval dataset that includes audio and textual captions in Marathi, Hindi, Tamil, Kannada, Malayalam and Telugu. Furthermore, we propose to compensate for data scarcity by using domain knowledge to augment supervision. To this end, in addition to the conventional three samples of a triplet (anchor, positive, and negative), we introduce a fourth term - a partial - to define a differential margin based partialorder loss. The partials are heuristically sampled such that they semantically lie in the overlap zone between the positives and the negatives, thereby resulting in broader embedding coverage. Our proposals consistently outperform the conventional max-margin and triplet losses and improve the state-of-the-art on MSR-VTT and DiDeMO datasets. We report benchmark results on Rudder while also observing significant gains using the proposed partial order loss, especially when the language specific retrieval models are jointly trained by availing the cross-lingual alignment across the language-specific datasets.

Abhishek, Shekhar Verma

In graph motivated learning, label propagation largely depends on data affinity represented as edges between connected data points. The affinity assignment implicitly assumes even distribution of data on the manifold. This assumption may not hold and may lead to inaccurate metric assignment due to drift towards high-density regions. The drift affected heat kernel based affinity with a globally fixed Parzen window either discards genuine neighbors or forces distant data points to become a member of the neighborhood. This yields a biased affinity matrix. In this paper, the bias due to uneven data sampling on the Riemannian manifold is catered to by a variable Parzen window determined as a function of neighborhood size, ambient dimension, flatness range, etc. Additionally, affinity adjustment is used which offsets the effect of uneven sampling responsible for the bias. An affinity metric which takes into consideration the irregular sampling effect to yield accurate label propagation is proposed. Extensive experiments on synthetic and real-world data sets confirm that the proposed method increases the classification accuracy significantly and outperforms existing Parzen window estimators in graph Laplacian manifold regularization methods.

Abhishek, Ashish Anand, Amit Awekar

Fine-grained entity type classification (FETC) is the task of classifying an entity mention to a broad set of types. Distant supervision paradigm is extensively used to generate training data for this task. However, generated training data assigns same set of labels to every mention of an entity without considering its local context. Existing FETC systems have two major drawbacks: assuming training data to be noise free and use of hand crafted features. Our work overcomes both drawbacks. We propose a neural network model that jointly learns entity mentions and their context representation to eliminate use of hand crafted features. Our model treats training data as noisy and uses non-parametric variant of hinge loss function. Experiments show that the proposed model outperforms previous state-of-the-art methods on two publicly available datasets, namely FIGER (GOLD) and BBN with an average relative improvement of 2.69% in micro-F1 score. Knowledge learnt by our model on one dataset can be transferred to other datasets while using same model or other FETC systems. These approaches of transferring knowledge further improve the performance of respective models.