Design an end-to-end ML solution to detect ads selling weapons.

Focus on outlining the entire machine learning pipeline, starting from data collection to model deployment. Consider the types of data sources you would use, such as social media platforms or online marketplaces, and how you would ensure data quality and relevance. Discuss the preprocessing steps necessary to clean and prepare the data, including how to handle unstructured text or images. Detail the model selection process, including potential algorithms and how you would train and validate the model. Finally, address how you would implement monitoring and feedback loops to continually improve the model's accuracy and adapt to new types of weapon ads.