http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Mobeen Ur Rehman,Kil To Chong 제어로봇시스템학회 2021 제어로봇시스템학회 각 지부별 자료집 Vol.2021 No.12
The Paper focuses on detection of mitosis in breast cancer, current progression for detection are though quite efficient but as they are carried out through manual assessments of pathologists and oncologists, it munches a lot of time to ensure the results. Still there is a gap that is needed to be filled. Mitotic count is the important factor, as it gives the rate of aggressiveness of the tumor cells, how rapidly it is spreading in the human body. The detection of mitosis is based upon the dominant factors of intensity, shape and texture, which we selected as the parameters for mototic and non-mitotic cells detection. The aim of the research is to propose a model on basis of the features that play significant role in recognizing the mitosis of cancer cell. For this purpose, the features that distinguishes cancer cell from the normal cells were identified. Keeping these features as the base segmentation, feature extraction with multiple techniques and classification algorithms were implied to the source dataset of the images, whose ground truth reference was available. After the implementation of all these approaches, it is stated that the software to carry these assessments digitally produced the excellent results and accuracy up to 91% for detection of mitotic cell and 92% for the non-mitotic cells.
A Neural Network Based Computational Model for Post-transcriptional Modification Site Identification
Mobeen Ur Rehman,Kil To Chong 제어로봇시스템학회 2021 제어로봇시스템학회 국제학술대회 논문집 Vol.2021 No.10
In a variety of cellular and developmental processes, RNA alterations are important. Understanding the distributions of RNA modifications in genome sequences will lead to the discovery of their functions. In the last five years, computational methods for identifying RNA changes have been presented because experimental methods are time consuming and complex. However, both experimental and existing computational approaches have difficulties when it comes to concurrently recognizing changes on various nucleotides. Recently a machine learning based model for simultaneously identifying multiple kinds of RNA modifications was proposed however the neural networks for such problem are not explored yet. To solve this problem, we built a new predictor in this paper that can identify m6A, m5C, and m1A for alterations in Homo sapiens, Mus musculus, and Saccharomyces cerevisiae at the same time. The proposed model uses k-mer encoding scheme to encode the input sequence. The encoded sequence is used by convolution neural network which automatically learns the features and performs classification between modified and unmodified sequences. The 10-fold cross-validation results have exhibited improved results in comparison to the existing state-of-the-art results in literature.
Mobeen Ur Rehman(레흐만 모빈우르),Kil To Chong(정길도) 대한전기학회 2021 대한전기학회 학술대회 논문집 Vol.2021 No.10
DNA methylation is a modification mechanism that takes part in the number of biological functions including the normal development and proper functionality of the brain. The methylated DNA holds crucial epigenetic information. In recent years numerous research drills are carried out to propose efficient computational models for 6㎃ modification identification. Still, a great gap of improvement is available in the performance of these models. This paper proposes an efficient model for DNA N6-methyladenine modification identification in Rice genome. In proposed model one-hot encoding is applied to the input-sequence. The two convolution layers with a max-pooling layer and a dropout layer extract convolutional features and Long Short Term Memory (LSTM) gives the optimal interpretation to these features. The extracted optimal feature vector from Neural Network is embedded to the Naïve Bayes (NB) for the classification of DNA 6mA. The proposed model is evaluated on two publically available datasets of Rice genome. The proposed model has illustrated high performance when compared with other existing techniques. The high performance of the proposed model depicts its effectiveness.
Mobeen Ur Rehman,Kil To Chong 제어로봇시스템학회 2022 제어로봇시스템학회 국제학술대회 논문집 Vol.2022 No.11
Among the most pervasive gene mutations is DNA methylation. An important part of controlling chromatin shape and epigenetics is DNA 4mC modification. To comprehend biological processes, it is essential to correctly identify 4mC sites. In this paper, we introduced the i4mC-NN, a computational predictor. The model involves a unique neural network architecture that uses three encoding schemes to extract insight information about the modification. Neural features gathered from different encoding schemes are combined to get the final prediction. The multi-feature encoding approach served as the foundation for our stacked neural network model. Numerous experimental findings demonstrate that proposed predictive model may greatly enhance due to stacking features, surpassing existing approaches in benchmarking comparisons. Additionally, we discovered that our model can accurately describe the properties of 4mC sites in comparison to other widely-used feature descriptors, highlighting the model’s potent feature learning capabilities. As a result, it is reasonable to assume that i4mC-NN will be a beneficial tool for research groups of interest.