![Captura de Pantalla 2022-01-03 a la(s) 19 39 26](https://user-images.githubusercontent.com/80273045/148165701-b1ae6ad4-48c3-48a4-ab2e-8f292d5dab78.jpg)
About me:
Hi, my name is Cesar Olivares Espinosa, Iโm currently enrolled on a Masterโs programme on Computer Engineering Sciences, focusing on Data Science, Computer Vision and Machine/Deep Learning, I enjoy building projects on this topics and learning everyday. I have a Biomedical Engineer Degree where I find about image processing and thats whatโs brought me here!
Projects:
-
Potato Leaves Disease Classification on Google Cloud ๐ฅ๐ฅฌ๐: On this project a Convolutional Neural Network model is trained using images from potato leaves, with three clases, healthy and two with diseases. The model achieves an accuracy around 0.95. The model is used in a React local webpage having a drag and drop function, which returns the predicted class with its accuracy. The trained model is then loaded into Google Cloud Platform, connected with the FastAPI to get predictions with a cloud server and tested with Postman.
-
Sentiment Analysis on Movies Reviews ๐๐คฌ๐ฟ: On this project a pre-trained BERT Model was taken from Hugging Face, specifically a Multilingual uncased sentiment to assign a value to a sentence given on the sentiment of the text. Data was retrieved from the popular website RottenTomatoes using the Request Dependency, processed with the BeautifulSoup library and RegEx, finally formatted on a Pandas DataFrame.
-
Bank Churners Prediction using Treeโs Algorithms ๐ฆ๐ณ๐ณ: On this project we capture data from Kaggle, perform Exploratory Data Analysis, cleaning, feature engineering, categorical encoding and use the data to train different Tree Models, use cross validation to find the one with the best performance and fine tune the parameters.
-
MNIST Digit Generator with GANs ๐ข๐ค: On this project we take the popular MNIST dataset which consists of tens of thousands images of hand written digits, design a Generative Adversarial Network, train it on the dataset to generate more realistic digits on every epoch.
- Text Generator using LSTM ๐๐ช:
On this project we train a Long-Short Term Memory Neural Network using Tensorflow on the popular book Alice In Wonderland to generate text given an input, we can see how the model is behaving on each run!
From Scratch Algorithms:
- DBSCAN Algorithm
- K-Means
- K-Modes
- SVM
- Naive Bayes
- ID3 Algorithm
Kaggle:
- TensorFlow - Help Protect the Great Barrier Reef๐ : Detect crown-of-thorns starfish in underwater image data, currently using YOLOv5 and YOLOx to train the model.
- Feedback Prize - Evaluating Student Writing Analyze๐ argumentative writing elements from students grade 6-12, currently uing a BERT BigBird Model to find the discourse type.
Technologies:
- Python
- R
- Java
- MATLAB
- GitHub
- SQL
- OpenCv
- Pandas
- Matplotlib
- SkLearn
- TensorFlow
- PyTorch
- Google Colab