Traditional learning is a process where a specific set of variables or inputs are tested against each other, and their results are analyzed and compared to determine which input or variable is better in terms of achieving a particular goal. This process can be compared to an A/B test where two versions of a variable are tested against each other to determine which version performs better.

On the other hand, Meta-learning is a more advanced technique that takes the results from traditional learning and uses them as inputs to create a learning model. In addition to using these traditional learning results, meta-learning also considers another set of data known as "dark inputs." Dark inputs are essentially data points or variables that have not been previously tested or analyzed in a traditional learning context.

By combining traditional learning results with dark inputs, meta-learning creates a more sophisticated learning model that can improve over time. This is achieved by continuously testing and analyzing the performance of the learning model against the dark inputs, and adjusting the model's algorithms and parameters to improve its performance.

An example of meta-learning is the use of machine learning algorithms in natural language processing. In this scenario, the model is trained on a specific set of data (traditional learning), and then additional data is added to the model to help it learn new patterns and structures (dark inputs). This can improve the model's ability to understand and process new types of data, leading to more accurate and efficient language processing.

<aside> 💡 Traditional learning focuses on comparing a set of known inputs to determine which one performs better, meta-learning takes this a step further by incorporating new and previously unknown data points to create a more sophisticated learning model that can adapt and improve over time.

</aside>

Our algorithms first consider publicly available data, such as research studies and industry reports, and then incorporate internal data sets.

A study conducted by Startup Genome found that startups with balanced teams, consisting of one technical founder and one business founder, are more likely to raise funds, experience user growth, and avoid premature scaling.

In addition to incorporating data from studies like this, the meta-algorithm also uses thousands of other inputs based on our team's extensive experience in the venture investment space. These "dark inputs" help the algorithm to intelligently learn which companies to invest in and when to do so, resulting in better yearly results on their deployments.

Moreover, the use of meta-algorithms allows the team to distinguish outliers and game-changers far before others in the venture space. This early identification of market opportunities can result in better pro rata rights and better market positions, giving Blueprint a competitive advantage.