Ashad Ahmad Email & Phone Number

India

Pune, Maharashtra, India

• To search different electrolytes based on their chemical and physical properties and tried to find the different QSPR relationships
• Used different platforms like PubChem and the DrugBank database to obtain different properties to calculate the coordination energy
• Performed exploratory data analysis on datasets and then developed different machine learning models Lasso, Es-LiR, GA-SVR, etc
• Finally obtained result with an accuracy of 67.45% using GA-SVR model for regression used to calculate E-corr of different electrolytes

Kharagpur, West Bengal, India

• Objective: To create Xprize rapid Covid testing to develop faster, cheaper and easier to use Covid-19 testing ML methods
• Datasets from many nations with various metrics such as GDP, daily cases, age group ratios, sex ratios, and so on has been collected
• Performed the exploratory data analysis on the datasets with various parameters using Numpy, Pandas, and the other Python packages
• Created the distinct correlations between parameters, by using spectral and K-means clustering which used to form groups of nations
• RNN-LSTM (time series) model used to train the datasets and acquire an f-score of 0.89.63, as well as to predict Covid cases using models

Kharagpur, West Bengal, India

• Objective: To create different machine learning models and to understand the water quality and turbidity in Hong Cong city
• Performed exploratory data analysis on datasets containing different parameters using different python libraries like Numpy, Pandas etc.
• Developed different the machine learning models like SVM, ANN etc. to find the relation between different parameters and the turbidity
• Achieved the accuracy of 92% on highly imbalanced dataset by implementing using recurrent neural network- long short term memory
• Identified the most accurate model and has written research paper for water quality measurement under the guidance of Prof. Saud Afzal

Varanasi, Uttar Pradesh, India

• Image Captioning | Computer visionObjective: To generate caption from images by using different MSCOCO dataset by using different deep learning tools
• Developed a merged model architecture that combines both the CNN and the RNN model to output textual description from an image
• Performed data cleaning steps on captions and generated a vocabulary by selecting words with a threshold of ten using the NLTK library
• Utilized Transfer Learning in the model, like ResNet50 to create image vectors and Glove6B50D as embedding layer to create captions