In this assignment I will explore different kind of encoding techniques such as co-occurrence matrix, sparse representations via PPMI weighting, pre-trained dense word embeddings, ecc...

Effects of MRI scanner manufacturers in classification tasks with deep learning models

My thesis work. Data processing on Google Drive. Here are only scripts and key findings.

意味表現学習

🧮 Python package to construct word embeddings for small data using PMI and SVD

This project focuses on text mining "The Big Bang Theory" scripts, covering 10 seasons. Participants preprocess character dialogues, analyzing sentence/word counts, noun/person name mentions, important words per episode/season, and word co-occurrence. (Part of Evaluation of Text Mining-KUL [G00C8a])

Turn documents into vectors by decomposing a PPMI cooccurence matrix.

The project strives to predict the risk of Parkinson's Disease progression in the patient based on the evaluation of baseline motor and non-motor symptoms of the patients via machine learning approach.

In this paper we compare and evaluate two simple embedding models which can be constructed directly from a given co-occurrence matrix extracted from Twitter data; Positive Pointwise Mutual Information (PPMI), and Hellinger Principal Component Analysis (H-PCA). For each embedding model we consider three alternative metrics for word similarity: co…

Parkinson's Progression Marker Initiative data science challenge, 2016

Construct Word Embeddings with PPMI & SVD

An implementation of (Chambers and Jurafsky, 2008), using updated machine learning models, and different training data domains for an independent study at the University of Pennsylvania.

Analysis of Parkinson's Progressive Markers Initiative Data

Four word embedding models implemented in Python. Supporting arbitrary context features