Volume 14, Issue 11, 11 2023 Edition - IJSER Journal Publication

Publication for Volume 14, Issue 11, 11 2023 Edition - IJSER Journal Publication

IJSER Research Group https://www.ijser.org/forum/index.php Register for IJSER Research Forum      IJSER Xplore IJSER Xplore Research Paper Database

Pages   [1]

Optimizing Tech Sales Talent - A Data-Driven Approach []

This paper examines an integrated data-driven methodology for optimizing talent planning implemented within five technology sales organizations seeking accelerated growth. The approach combines AI prospecting, predictive hiring analytics, and immersive sales training techniques to enhance end-to-end planning. Results indicate consistent and significant improvements versus conventional methods for specialized sales role staffing velocity (57-95% faster), quality of hire (66% higher sales capability), and capability building (60%+ motivation lift). The integrated techniques offer technology sales leaders solutions to urgent deficits inhibiting growth.


Attendance is one of the most important elements of any educational institution because it explains the academic performance of a student and in relation to the NUC benchmark of 75% policy, it states the exact students and number that are eligible to write an examination thereby, acting as a guide in producing question paper. Based on these, a good system that accurately captures the attendance of students is of utmost need to any institutional administrator. Radio Frequency Identification (RFID) based class attendance system is a facial pattern recognition system that automatically captures the student’s attendance. It coordinates the hardware and software design through handshaking the data communications between RFID tag and RFID reader whereby the student passes through a RFID reader with the tag on them or swipes the tag on the RFID reader for the web camera to capture the tag’s electronic product code (EPC) and images respectively. It does eigenface pattern matching with the existing patterns in the database (MySQL) for proper validation and authentication. The tools used in achieving this system are C variant, Arduino based programming language (the programming languages for the microcontroller AT89C51 on Arduino development board), and the Arduino 1.6.5 IDE. Java programming language was used for the serial communication, eigenfaces training, detection and recognition, and the application interface. It provided security and privacy such that the product codes and identification were not compromised. The system performed better than the manual method and RFID with finger print in terms of time and accuracy.

Vision Based Vehicle Detection System Using Deep Learning []

Recently, there was a change observed in deep learning architectures for better application in vehicular traffic control systems. In TensorFlow, the pre-trained model is extremely efficient and may be transferred easily to unravel other similar problems. However, thanks to inconsistency between the first dataset utilized in the pre-trained model and therefore the target dataset for testing, this will cause low-accuracy detection and hinder vehicle counting performance. One major obstacle in retraining deep learning architectures is that the network requires an outsized corpus training dataset to secure good results. Therefore, we propose to perform data annotation and transfer learning from an existing model to construct a replacement model for vehicle detection and counting within the world urban traffic scenes. Then, the new model is compared with the experimental data to verify the validity of the new model. Besides, this paper reports some experimental results, comprising a group of innovative tests to spot the simplest detection algorithm and system performance. Furthermore, an easy vehicle tracking method is proposed to assist the vehicle counting process in challenging illumination and traffic conditions. The results showed a big improvement of the proposed system with the typical vehicle counting of 80.90%.

Experimental Results for Breakdown Voltage around Various Contaminating Particles with Presence of Spacers under D.C Voltage []

The combination of metallic particle contamination and spacers has shown to be the most critical dielectric design factor in the case of gas-insulated transmission lines (GITL) and gas-insulated substations (GIS) systems. In this paper, the breakdown voltage experiments are carried out in a test chamber measuring 12 cm in diameter and 50 cm in length and capable of with-standing pressures of up to 5 bars. The electrode system is mounted in a room-temperature chamber filled with SF6 gas or air. The electrode system utilized in this experiment consists of two parallel plane electrodes connected by an Epoxy Resin spacer within the test tank. The epoxy resin material is utilized in the production of spacers. Epoxy resin is made up of two compo-nents: Bisphenol-A and an anhydride hardener. In this paper, we create different concentrations of Bisphenol and hardener, which are subsequently combined to generate an epoxy resin substance. The optimal concentration of these materials is then chosen to form various spacer forms such as disc and conical spacers. The influence of single and multiple contaminants, such as spherical and wire particles, on breakdown voltage values is investigated. The influence of different spacer forms, such as disc and conical spacers, as well as the presence of various contaminating particles, on recorded breakdown voltage values is also investigated. A comparison of measured and calculated breakdown voltage values is performed.


This research discussed about how to utilize generative pretrained transformer-2 (GPT-2) to automatically generate Indonesian poetry. The base model of GPT-2 will be finetuned for the task of generating Indonesian poetry. This model uses a transformer type of neural network. GPT-2 is a type of generative AI model based on transformer mechanism. The finetuned GPT-2 will be trained using a dataset containing list of Indonesian poetry that contains a title, author, and the content of poetry. The model will accept title as an input and will generate a new poetry with similar format as the training dataset. This research was done to know how to utilize GPT-2 to generate Indonesian poetry and evaluate the perplexity and ROUGE-1 score. Experimental results show that on average, finetuned GPT-2 model has a perplexity score of 1.18 which is around 20.27% higher than encoder-decoder RNN model and ROUGE-1 score of 0.13 which is around 44.44% increase in comparison to the base model of GPT-2.

Simply Irrational []

Irrational numbers, though uncommon, are indispensable for certain scientific calculations. Rational numbers can be represented in multiple forms, but representing rational numbers in irrational form was never tried before. In this paper, we look into the different aspects of this representation.

A Smart Data-Driven Approach to Improving Public Fiscal Performance []

Modern technologies, such as (Big Data analytics, AI, Blockchain, etc.) have emerged to keep pace with the needs of society and companies that provide products and services of various kinds. The interest in data quality and employment information to extract practical knowledge has increased to help transform people’s lives in businesses, industry, government, and services. In developing countries, the financial sector has already applied (FMIS) and Performance-based Budgets to ensure proper scrutiny of budget estimates and to ensure (the accuracy, effectiveness, and efficiency) of government revenues and expenditures. However, it cannot face the financial obstacles resulting from external crises. Therefore, this research aims to investigate emerging artificial intelligence techniques in a financial analysis prediction to improve enterprise and local institutes' performance. we explored the studies that provide a novel early warning system to strengthen financial management by exploiting Data Science and AI techniques to face these difficulties. To accomplish the research objective EKB databases including Science Direct, Springer-Link Journal, IEEE Xplore, and Emerald are employed. Our study will shed light on the various research in this field and provide a pathway for us to analyze recent financial distress prediction models to enhance financial management decisions. Finally, we suggested a smart FMIS by emerging a data-driven AI-Based approach Based on a comparative analysis of recent research related to financial prediction. The contribution of artificial intelligence to finance has several forms, including ANN and its branches, integrated forecasting and prediction models with ANN, decision trees, sentiment analysis, and an AI-explainable approach. We discovered that the best techniques for predicting financial distress are Integrated Z-score, MLP, ANN, and Hybrid CNN, LSTM, and AM model. This article is the first study that explores opportunities for applying AI techniques to local government financial information systems.

Analyse de l’occupation du Sol dans le bassin versant de la dépression de la cuesta de Thiès à travers les images Sentinel-2 de 2017 à 2021 []

Abstract—La tranche temporelle des images obtenues concerne les dates 2017, 2018, 2019, 2020 et 2021. Les résultats des traitements des images Sentinel dans le bassin-versant ont permis d’obtenir sept principales classes d’occupation du sol notamment les arbres, les cultures, l’eau, le sol nu, les pâturages libres, la végétation inondée, les zones bâties. L’analyse diachronique du paysage peut servir à montrer l’effet de l’homme et de ses activités ou tout simplement des changements climatiques sur les modifications de l’occupation du sol (Hobbs, 1990 ; Kpedenou et al., 2016). La présente étude a pour objectif de cartographier à travers les données sentinelles, l’occupation du sol et de quantifier son évolution de 2017 à 2021. Les résultats ont permis de constater que l’occupation du sol est en mutation. Les pâturages libres qui représentaient en 2017, 87.9 % de la superficie du bassin versant ont connu une régression de 6.4% sur la période de 5 ans. Par contre, il a été observé une progression respectivement des zones de cultures 4.4%, des zones bâties 2% et des arbres 0.8%. En somme, l’étude montre que le paysage naturel du bassin versant est marqué par une anthropisation.">La tranche temporelle des image

Pages   [1]