Machine learning Jobs

4 were found based on your criteria

  • Hourly – 1 to 3 months – Less than 10 hrs/week – Posted
    I need a machine learning professional with python experience with scikits-learn to design, pre-train and train and finetune a system of classification using naive bayesian filter process (Details will be discussed on skype). The work is in the context of data extraction in natural language processing. The project will span over the next 4 to 6 weeks and start very soon. The process involve back and forth so flexibility and a good ability to communicate is needed. Thank you very ...
  • Hourly – Less than 1 month – 10-30 hrs/week – Posted
    Overview: We have an existing database of thousands of documents and we want to classify them as SPAM/HAM. The objective of this job is to find the best possible model to predict if a determinate document is SPAM or HAM. The model is going to be used to classify a larger collection of documents using an Openscoring open source web service or other technology (on you suggestion). The model should be optimized to provide the best combination of TP ...
  • Hourly – Less than 1 month – Less than 10 hrs/week – Posted
    We have a data set of values recorded for 1000 patients. For each patient, we recorded around 100 potential predictors of a clinically relevant variable, which has one of three values: (1) no disorder, (2) mild disorder, (3) severe disorder. Many of these potential predictors are correlated. We would like to create a simple mechanism which can be used in clinical practice to predict whether a patient has the disorder. Ideally, this procedure should use as few predictor variables as ...
  • Hourly – Less than 1 month – Less than 10 hrs/week – Posted
    I need a learning algorithm to improve the way posts created automatically by an rss feed importer are categorized. I can get into nearer specifics later on. At the moment the plugin uses just simple keyword matching to categorize posts and we would like to implement something more intelligent. This is the way we envision it: 1) No training sets. Instead the training data should be compiled by user feedback. This should be in the form of how you see ...
loading