Statistical Methods for Data Drift Detection and Monitoring

Data Drift and Concept Drift

Statistical Methods For Data Drift Detection Introduction Data quality issues account for majority of machine learning model flaws. Monitoring and detecting changes in data is commonly used to identify causes of model performance degradation, model drift, decay or staleness. One of the most important components of machine learning operations is... [Read More]

Bayesian Optimization

Bayesian Hyperparameter Tuning

Bayesian optimization Suppose we want to optimize a function such as validation error (which is a function of hyperparameters) but it’s really complicated. We can approximate it with a simple function, called the surrogate function. After we have queried a certain number of points, we can condition on these point... [Read More]

Fine-Tuning XLNet for Sentiment Analysis

Sequence Classification of Fake and Real News

Introduction XLNet is a generalized autoregressive Transformer that enables learning bidirectional contexts by maximizing the expected likelihood of a sequence w.r.t. all possible permutations of the factorization order. XLNet employs Transformer-XL autoregressive model into pre-training but without the limitation of the fixed forward or backward factorization order of autoregressive models.... [Read More]

Fake and Real News Detector

Classification and Sentiment Analysis of Fake and Real News

Introduction Fake news include false news stories,disinformation and misinformation with the intent of misleading people. The proliferation of fake news in recent years poses great danger to the safety and security of many people around the world. Some people have acted on fake news to commit certain crimes which are... [Read More]