ISSN:2349-2058
T R A C K P A P E R 
| S.No. | Title & Authors | Page No | View | ||
| 1 |
Title : Overview of the Video Quality Measurement Techniques Authors : Chena Ram, Subhash Panwar
Abstract :
Various network or channel errors in the network environment will result in damage or loss of video information during transmission or storage. Since very high compression has usually been applied to the compressed bit stream, any damage in the reconstructed signal at the decoder would likely lead to unpleasant visual distortion. To control the inception of potentially visible artifacts, a proper mechanized video quality measurement method is needed. This paper gives a systematic overview of available subjective and objective assessment techniques for video quality, together with performance comparisons of objective assessment methods for quality. The previous results demonstrated that the high-performance video quality predictors of perceived video quality are SSIM, MS-SSIM, RR-VQM, and VIF |
1-6 |
|
||
| 2 |
Title : An Ensemble Technique of CNN for Lung Cancer Identification Authors : Ishrat Khan, Md. Shafiur Rahman Khan, H. M. Abdul Fattah
Abstract :
Lung cancer is one of the leading contributing factors to the mortality rate. The prevailing types of Non-small Cell Lung Cancer (NSCLC) include Adenocarcinoma, Large Cell Carcinoma and Squamous Cell Carcinoma. Studies have shown that 18% of the mortality rate stems from this disease. This is attributable to substandard diagnosis techniques and inefficient treatments available to cure metastasis. Hence, this paper opts to employ transfer learning techniques by using different state-of-the-art, pre-trained models to detect lung cancer and classify it into four groups, namely, Adenocarcinoma, Large Cell Carcinoma, Squamous Cell Carcinoma and Normal using chest CT scan images, followed by conducting a comparative analysis of their performances. The models implemented are Resnet101, VGG16 and InceptionV3 and DenseNet169. Moreover, the paper proposes a new CNN model using the ensemble technique which is an amalgamation of ResNet101 and InceptionV3 that are employed initially. In addition to that, it introduces another 11 layered CNN model built from the outset. The dataset used for this study is called Chest CT-Scan images which is retrieved from Kaggle. The models’ performances are analysed utilizing different metrics like accuracy, precision, recall, F1-Score and AUC. The Ensemble of ResNet101 and InceptionV3 models has achieved the highest accuracy of 93.7%, on this particular dataset. |
7-11 |
|
||
| 3 |
Title : Predicting Drug Combination Synergies Based On Transformer Networks Authors : Yiding Zhang
Abstract :
As research into tumours continues, a large number of targeted drugs for the treatment of tumours are being devised. However, most of the drugs currently available for cancer treatment only target a single type of cancer, and single-use drugs have low drug utilisation and are prone to drug resistance. Combinations can be an excellent way to address these issues with single medicines, but because there are so many different kinds of anti-cancer medications on the market, it is inevitable that using just experimental approaches to screen combinations will be ineffective. Deep learning techniques are required to forecast the effects of combination medications, which are subsequently tested experimentally, in order to solve this issue. In this paper, we propose to use a model of transformer arithmetic to predict the effect of drug combinations on cell lines. After being fed into a multi-layer feed-forward neural network, information on the drug's molecular structure and the proteomics of the cancer cell line is used as input to the model, and this feature information is then used as input to the TRANSFORMER to predict the action fraction of the drug on a particular cancer cell line. And our model is compared with other machine learning and deep learning models. On the independent test set, our model outperforms the other models. |
12-14 |
|
||
| 4 |
Title : Extremely Efficient Convolutional Neural Network for Mobile Devices using FFT Authors : Weizhi Cui
Abstract :
Building deeper and larger convolutional neural network (CNN) is a primary trend for a wide range of applications such as image recognition, nature language processing. However, the most accurate CNNs usually have hundreds of layers and thousands of channels, thus requiring large computation and power consumption. The deployment of deep CNN in power-constrained and performance-limited scenarios remains challenging due to substantial requirements for computing resources and energy needed. In this paper, we propose a novel approach designed efficient CNN using FFT. The implementationand optimization on low-power zynq platform has been presented. Our empirical results show our method reduces computational time by a factor 2 times. |
15-18 |
|
||
| 5 |
Title : Factors Affecting the Persistence of Profit and Their Impact on the Quality of Earnings Authors : Hari Setiyawati, Marsudi Sukmono
Abstract :
One way to be able to see the company's performance is from the annual profit that the company generates. If the company has a persistent profit, it can be said that the company's performance is good. By looking at earnings persistence, stakeholders can evaluate events in previous, present and future years. Earnings with a high level of persistence are very useful for predicting future earnings. This study aims to examine the factors that influence earnings persistence and its impact on earnings quality. The population in this study are food & beverage companies listed on the Indonesia Stock Exchange. There are 16 companies that meet the criteria by using purposive sampling. The research observation period is five years so that the total research data is eighty. Earnings quality is proxied by operating cash flow ratified by the company's net profit, profit persistence is proxied by profit before tax ratified by average total assets, sales volatility is proxied by sales in a ratio of total assets, operating cash flow volatility is proxied to total operating cash flows ratioed with total assets and leverage proxied by total debt in a ratio of total assets. The results showed that sales volatility had a significant effect on earnings persistence in a positive direction, operating cash flow volatility had a significant effect on earnings persistence in a positive direction, while leverage had no effect on earnings persistence. The results of this study also show that earnings persistence has an impact on earnings quality. |
19-26 |
|
Call for Papers
