Search

Artefacts that differentiate spoofed from bona-fide utterances can reside in spectral or temporal domains. Their reliable detection usually depends upon computationally demanding ensemble systems where each subsystem is tuned to some specific artefacts. We seek to develop an efficient, single system that can detect a broad range of different spoofing attacks without score-level ensembles. We propose a novel heterogeneous stacking graph attention layer which models artefacts spanning heterogeneous temporal and spectral domains with a heterogeneous attention mechanism and a stack node. With a new max graph operation that involves a competitive mechanism and an extended readout scheme, our approach, named AASIST, outperforms the current state-of-the-art by 20% relative. Even a lightweight variant, AASIST-L, with only 85K parameters, outperforms all competing systems., 5 pages, 1 figure, 3 tables, submitted to ICASSP2022

Many industries implement the blockchain for their application as it is a distributed ledger with a trust. The first application of blockchain is Bitcoin. Depending on the applicability of blockchain we have to consider different parameters and accordingly, we have to select the consensus algorithm. Consensus algorithm is the crucial part while designing any blockchain for the desired application. To confirm any transaction different consensus algorithms are used. In this paper, we will discuss different consensus algorithms. In addition to this, we will compare different algorithms concerning different parameters like time required to confirm the transaction, throughput, network used for the implementation. This article summarizes the factors to be considered while implementing the blockchain.

Artificial Intelligence (AI) is practically apply in all commerce and notable labors be able to witness in our routine life. E.g. chat bots, smart cars, IoT devices, healthcare, banking, and logistics and many more. Simple examples for the use of AI in day to day life are Alexa and Siri which are ready to assist to ease the work. They can be illustrated as consumer Artificial Intelligence powered smart assistants. They are of help that requires unskilled tasks of an individual's life, thus one be capable of centering significant stuff. But, have we ever thought that AI can infiltrate the limitations of the by hand encumbered legal processes as well? Can it help professionals and advocates with mechanization, by sinking their time on repetitive responsibilities? Taking about why and how AI and law connects, the prime reason for the same is to craft inventive applications to computerized routinely procedure or working related to legal matters. Artificial intelligence and the law cannot go hand in hand that too with synchronization. There could be many reasons, on among them could be immensely diverse attribute. In the current era, almost every legal practitioner is making use of Artificial intelligence in their day to day carry out procedures in which e-courts and use of software's are common which helps them to complete their responsibilities effectively and efficiently. This paper provides a general introduction to the field of AI and Law. This paper aims to introduce to the way in which the increasing use of artificially intelligent technologies affects the law.

It is very tedious for anyone to go through the whole document to get answers for their queries since there is a need of Question Answering system to make life easier. In this research, we used machine learning architectures in Question Answering field, based on the Stanford Question Answering dataset (SQuAD). In our work, build two models in which first model used unsupervised learning algorithm GloVe to get vector representation of word and trained using bidirectional-LSTM in which the training accuracy is 64.93% and testing accuracy is 60.33% scored with the model. In Second model used InferSent which is a sentence embedding method to get vector representation of data. This vector representation data is used to train model. The machine learning algorithms used are XG Boost and Multinomial Logistic Regression in which scored 70.02 percent in training and 66.03 percent in testing. The aim of this research is to build the best accuracy model using Glove and InferSent to use vector of various dimensionality to represent it numerically so that machine can interpret it.

In this research we have explained to set up an Infrared Array Sensor system that is IOT based in order to provide security at remote location. We have tried to Establishment of cloud environment to host IOT application & Development of IOT Application using Asp.net with C# programming platform. We have Integrated IOT with Infrared Array sensors in order to implement proposed work. In this research camera captures the external event and sent signal to Infrared grid array sensor. Internet of Things (IoT) would enable applications of utmost societal value including smart cities, smart grids & smart healthcare. For majority of such applications, strict dependability requirements are placed on IOT performance, & sensor data as well as actuator commands must be delivered reliably & timely.

This paper presents an approach to dynamically generating representative external driving cell and external wire parasitic assertions for the ports of sub-blocks of a hierarchical design. The assertions are based on a technology lookup table and use attributes of the port and the hierarchical wire connected to the port as keys. A concept of reverse timing calculation at the input of the driving cell is described that facilitates the approach to drive efficient timing optimization of boundary paths of design sub-blocks. Experimental results in an industrial timing environment demonstrate significantly improved timing optimization accuracy when compared to prior work.

"The mind is the attribute of man. When man is born, he comes into existence with only one weapon with him—the reasoning mind". Now considering the current scenario and seriousness of COVID 19, human is ready to get challenged in this crucial situation. The paper presents the proposal is to do analysis of COVID affected patients. Machine learning models help in identifying the severity and decision making. Understanding pandemic will help in implementing proper measures at right time. Implementing these measures prior to the outbreak will help maintaining country’s health and economy. Here we observe and analyze the behavior of data, which helps in providing better results for decision making. The model proposed in this paper is in adjunct with an interactive application that gives updates regarding the pandemic; indicates the different zones of the local area prescribed by the government and also serves a platform for needy.

Daclatasvir (DCV) used for the treatment of chronic hepatitis C virus infection (HCV). The present piece of work contains development, validation and force degradation study for DCV has been carried out by using LC-MS/MS technique. The LC-MS system contains 4000 QTrap along with the Shimadzu LC 20AD LC System, controlled by Analyst 1.4.2 software for present study. For the preparation of standard drug DCV and its marketed formulations the diluent used was acetonitrile: water (50:50v/v). LC system runs with the conditions like use of C18 column (Thermo Scientific, 5µm, 4.6 X 250 mm size), column temperature 25°C (Ambient), Injection volume 2 µL, total run time 10 min, flow rate 0.5 mL/min and an isocratic mobile phase consist of 1mM Ammonium acetate Buffer pH=4 adjusted with 50% acetic acid in water as ‘A’ and acetonitrile as a solvent ‘B’ (20:80 v/v). Validation was carried out by testing parameters like LOD (limit of detection), LOQ (Limit of quantification) linearity, ruggedness, accuracy and precision in terms of reproducibility, repeatability according to ICH guidelines. The calibration range for DCV was 500 to 5000 ng/mL, where 800, 1600 & 3200 ng/mL concentrations of DCV set as an Internal QC with proper labeling. All Three QC samples applied n=6 times. For Mass spectrometry (MS) quantification of the DCV, the interest of mass m/z 740.50/514.10 was considered. The force degradation study has been carried out on LC-MS for DCV at different stress conditions. The drug was found to be stable in acidic, neutral, oxidative and photolytic but degrades in basic conditions. The proposed method is fast, economical, avoiding chemical steps and wet chemistry and thus provides a comparatively simpler, rapid green method for routine analysis in pharmaceutical industries.

A good saying “Precaution is better than cure”. For emergency situations, the victims can be cautioned prior to the disaster occurs. And the emergency never comes with a prior notice. In this paper, ICT as a new technology in disaster management &Information Technology in Emergency Preparedness and Response is proposed. Disaster preparedness and awareness is the only effective way of mitigating the impact of future disasters. The development in the field of information and communication technology can help a great deal in the planning and implementation of disaster mitigation schemes.

In this study, we explore the use of Convolutional Neural Networks (CNN) for replay spoof detection in Automatic Speaker Verification (ASV) system. The Amplitude and Frequency Modulation (AM-FM) feature sets obtained from the Hilbert transform (HT) and Energy Separation Algorithm (ESA) are used as the front end. We have observed the effect of max-pooling and fully connected (FC) layers, when replaced with the convolutional layers in CNN. The results are compared with Gaussian Mixture Model (GMM) classifier, furthermore to obtain the possible complementary information of both the GMM and CNN classifiers, we have explored classifier-level fusion. In addition, we compared our results with Constant-Q Cepstral Coefficients (CQCC) and Mel Frequency Cepstral Coefficients (MFCC) feature sets. The architecture with max-pooling when replaced with convolutional layer along with FC layers had performed relatively better on most of the AM-FM feature sets compared to other CNNs. The ESA-based AM features (i.e., Instantaneous Amplitude Cosine Coefficients (ESA-IACC)) performed better as AM do not have more fluctuation as FM have during models training. The lower EER is obtained with classifier-level fusion of ESA-IACC feature set resulting in 2.54 % EER on development set and 6.04 % on evaluation set of ASVspoof 2017 Challenge database.

The purpose of using road divider is to separating the two ways of traffic ie ongoing and incoming vehicles in the traffic. With growing population, the vehicles used per family increases, but there is limitation in resources and leads to more number of cars on roads. In that case static road divider fixes the number of road lines on either side of road. This invites the better usage of available resources. In most of the cities, there are areas like industrial and shopping places where traffic flows only in one direction both in morning as well as in evening. In the peak hours, most of the time one road side is unutilized. It causes time loss of public and traffic jams. We aim to build a smart road divider in terms of automated road divider which moves or shift the lane directing the rush in traffic. Such type of mechanism of traffic system not only saves time but also fuel. It can add one more lane based on the traffic in the particular direction. With the smarter planet application proposed below, manual dependency and manual traffic coordination is reduced. Like this a proposal of smart traffic is built in ehich low , medium and high density of rush in the traffic will be shown in IOT server in using graph. It provides a better solution for traffic problem.

Access control is a fundamental security technique in systems in which multiple users share access to common resources. It is the process of stating and enforcing security A approach in network security for misbehavior detection system present a method for detecting malicious misbehavior activity within networks. Along with the detection, it also blocks the malicious system within the network and adds it to Blacklist. Malicious node defined as a compromised machine within the network that performs the task provided by i.e. it does not forward the legitimate message to another node in the network or sends some other message to a neighbor node. This system is based on Probabilistic threat propagation. This scheme is used in graph analysis for community detection. The proposed system enhances the prior community detection work by propagating threat probabilities across graph nodes. To demonstrate Probabilistic Threat Propagation (PTP) considers the task of detecting malicious node in the network. Proposed System also shows the relationship between PTP and loopy belief propagation.

Opportunities for improving actual power system performance is increased; since the power electronics technology is rapidly developing. This paper is a study and validation of capabilities of FACTS device, namely, STATCOM, through modelling and Simulation using Matlab/Simulink Platform. The STATCOM, is a device connected in transmission line to compensate reactive power; made of a coupled transformer and a solid-state VSI. At the point of connection, capacitive reactance or inductive reactance is emulated since the sinusoidal current is injected of varying magnitude which is almost in quadrature with the line voltage. This paper shows the functioning of the STATCOM using simulink platform verified simulation. Simulation result shows the capability of STATCOM for reactive power compensation & fast response.

This paper present a soft computing approach to form dynamic cluster by using dynamically drown pattern on behalf of assigned attribute values of entities. The proposed methodology work on a proposed trained system which could work on drown pattern from assigned attribute values those depends on user's preference to form clusters then information could retrieve and categories as per user need. This whole system could reduce much require time to retrieve specific information from the huge data.

Mobile Ad hoc Network (MANET) contains mobile nodes that can exchange information with each other without the need of framework. The development of nodes is arbitrary; thusly MANET's have a dynamic topology. Because of this, the connection breakages in these systems are something commonplace. This issue causes high information loss and deferral. So as to lessen this issue the possibility of P-AODV has been arrived. In P-AODV routes are repaired on the premise of priority. In this paper new approach Modified P-AODV is clarified in which paths are predicted on the basis of received signal strength and energy of node. The node having good RSSI and energy are given high priority and rests of the other nodes are given low priority. The accessibility of a high priority neighbor is required for the accomplishment of MP-AODV. This new mechanism decreases the delay and packet loss that occur in the original protocol.

Lift or elevator is transport devices that are used to move goods or peoples vertically in upward and downward directions. In this project, the microcontroller based lift control system is constructed to simulate as an actual lift in the real life. Most lifts are powered by electric motors that pull steel cables along the pulleys. The elevators must ensure the safe operation meeting all safety requirements and compliance standards. Backup in case of electricity failure, buzzer, fan and light in a single apparatus, actuators for door locking are the key points discussed in the project.

Hydro-meteorological measures are one of the most challenging tasks of nature. Rainfall has become the most significant and technical factor for scheduling and functional criteria of any irrigation program with profound effects on agribusiness based economy for any region. It is vital important to accurately estimate rainfall and has been remained a great challenge. To estimate medium range monthly summer monsoon rainfall, machine learning models were developed using thirteen rain-gauges’ long-term-average rainfall values of Jaipur city. The different input combinations of rain-gauges having 2, 4, 6, 8, 10 and all-input parameters were explored using rainfall values of rain-gauge stations of Viratnagar, Bassi, Chaksu, Chomu, Jaipur, Dudu, Jamuna_Ramgarh, Kotputli, Phagi, Phulera, Amber, Sanganer, Shapura in Jaipur city. Exploring various empirical and dynamic machine learning approaches on rainfall data of 54 years period [1961–2015], rain-gauge station-wise were computed to evaluate the best rain-gauge station of Jaipur city with selection of best model for medium range estimation of monthly summer monsoon rainfall of Jaipur city. In the course of escalated experimentation, it is observed that Linear Regression outperforms over rest of other machine learning techniques and Amber rain-gauge station is the best rain-gauge among all used stations. The dataset utilized for this work is accessible as supplement at http://waterresources.rajasthan.gov.in authored by Government of Rajasthan.

Metal Selenium thin film prepared as Metal Semiconductor Field Effect Thin film Transistor. Metal (Al/Cu) contact and metalloid (Se) channel used as a Field Effect Transistor on glass substrate by depositing the source, drain and top-gate Al/Cu contact. Metal gate is used to control flow of charge from source to drain. By varying the width of depletion layer channel has controlled using metal contact which modulates the thickness of conducting channel which causes flow of current between source and drain. The source and drain of the FET structure were defined by the metal layers Cu were sequentially deposited on the Se in the defined contact patterns by thermal evaporation technique at pressure of 10−5 torr using vacuum coating unit. Its electrical and structural properties were characterized. This device is used to control the flow of current by changeable potential applied between gate and source (VGS), and free charge accumulation has induced at semiconductor interface.

The users of networks are increasing rapidly due to increase in growth of internet. There is always a concern among owners of the digital products about illegal copying of their products. Important issues in multimedia applications and services today are Security and copyright protection. Digital watermarking is a technology used for copyright protection of images videos etc. that is stored on internet. Here own right information data called watermark is embedded into the digital media without affecting the perceptual quality of video. In case if any dispute arises, it is possible to detect or extract the watermark data from the media and it can be used as an ownership proof. Digital video watermarking scheme is based on Discrete Wavelet Transform and Singular Value Decomposition. Design of this scheme using Matlab is proposed. Embedded watermark is robust against various attacks that can be carried out on the watermarked video.