This paper describes results of using machine learning model to aid reduction of number of repairs in external workshops for motor insurance company. The model predicts the customer decision based on data stored in insurance company’s database as well as additional features. We built several models, based on decision tree, random forest, gradient boost, ada boost, naive bayesian, logistic regression, neural network, then we evaluated them on real data.
Built models were tested on separate evaluation dataset provided by the insurance company. Models achieved over 0.8 area under curve ROC and thus were accepted for a pilot study in the production environment.
In this paper we evaluate the current state of the art in natural language paraphrase generation using deep learning methods. The focus is put on the entire modeling pipeline from data gathering up to model evaluation. Specifically, we list the publicly available datasets suitable for this task, assess their quality and discuss procedures connected with data preparation and model training. Finally, we discuss problems related to the currently used evaluation approaches.
Adversarial examples are deliberately crafted data points which aim to induce errors in machine learning models. This phenomenon has gained much attention recently, especially in the field of image classification, where many methods have been proposed to generate such malicious examples. In this paper we focus on defending a trained model against such attacks by introducing randomness to its inputs.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.