mboukabous's Stars
plotly/dash-sample-apps
Open-source demos hosted on Dash Gallery
amaiya/ktrain
ktrain is a Python library that makes deep learning and AI more accessible and easier to apply
RockyXu66/Faster_RCNN_for_Open_Images_Dataset_Keras
Faster R-CNN for Open Images Dataset by Keras
linogaliana/python-datascientist
Dépôt associé au cours Python pour data scientists (ENSAE 2e année)
mboukabous/Security-Intelligence-on-Exchanged-Multimedia-Messages-Based-on-Deep-Learning
Deep learning (DL) approaches use various processing layers to learn hierarchical representations of data. Recently, many methods and designs of natural language processing (NLP) models have shown significant development, especially in text mining and analysis. For learning vector-space representations of text, there are famous models like Word2vec, GloVe, and fastText. In fact, NLP took a big step forward when BERT and recently GTP-3 came out. Deep Learning algorithms are unable to deal with textual data in their natural language data form which is typically unstructured information; they require special representation of data as inputs instead. Usually, natural language text data needs to be converted into internal representations form that DL algorithms can read such as feature vectors, hence the necessity to use representation learning models. These models have shown a big leap during the last years. Their set ranges from the methods that embed words into distributed representations and use the language modeling objective to adjust them as model parameters (like Word2vec, fastText, and GloVe), to recently transfer learning models (like ELMo, BERT, ULMFiT, XLNet, ALBERT, RoBERTa, and GPT-2). These last use larger corpora, more parameters, more computing resources, and instead of assigning each word with a fixed vector, they use multilayer neural networks to calculate dynamic representations for the words according to their context, which is especially useful for the words with multiple meanings.