Rest Analytics and Research

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 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]

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]

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]

Using Scikit-Learn Pipelines and Converting Them To PMML Introduction Pipelining in machine learning involves chaining all the steps involved in training a model together. The pipeline allows to assemble several steps that can be cross-validated together while setting different parameter values. It is a step closer to automating the all... [Read More]