AdOptimizer: A Self-Supervised Framework for Efficient Ad Delivery in Low-Resource Markets

In markets with few resources, effective ad delivery stays a main problem. This happens because people get little access to labeled data and audience information. We present AdOptimizer, a self supervised system. It learns advertisement plans by itself, using only a few labeled samples. The system uses contrastive learning plus clustering - this helps to find connections between ad features and how audiences respond. That method improves ad targeting also campaign setup. It does not depend much on a lot of data. The system includes a feedback loop; this loop changes ad plans as users interact with ads. The system reacts to how the market moves. Tests show big gains in click through rates and conversion rates. This shows the system works well in markets that do not get good service. AdOptimizer acts as a changing answer for advertisers. It improves how people interact as well as how much money ads bring in, especially where resources are limited. CCS Concepts: • Software and its engineering → Software prototyping.