Learning systems deliver valuable results, but their way there is often intransparent / ACTICO’s Intelligent Automation Platform offers this transparency, which is for example crucial for audits
Immenstaad: August 20, 2019 – Banks and financial service providers turn to ACTICO’s intelligent automation platform to assess the risk of corporate customers defaulting on loans or to ensure their advertising reaches the right target groups. Its ability to learn means the ACTICO system can optimize rule-based processes. In the latest version, ACTICO has met one of the market’s key requirements by ensuring the machine learning component of its system is audit-proof. ACTICO’S AI is able to justify decisions by explaining how they were made – something that is not a given in many AI systems. This high degree of transparency also helps to build acceptance of technologies such as machine learning (ML) and artificial intelligence among users and their customers.
ACTICO Platform enables companies to generate new knowledge from existing data and thus automate their business processes. This is known as intelligent automation. The machine-learning-based component helps to identify new patterns and areas for improvement. A special feature of ACTICO’s system is the total transparency of the decisions taken by its machine learning algorithms.
“This high degree of transparency allows firms to provide their customers with convincing arguments for their decisions about bank loans or insurance categories”, says ACTICO’s CEO Thomas Cotic. “Knowing how a decision has been made is not only relevant for customers, but also for regulators in tightly controlled industries, such as banking, insurance and healthcare.”
The perfect combination of rules and machine learning
ACTICO’s approach is different because it combines two aspects: expert knowledge in the form of predefined rules and data insights generated by machine learning. “Other systems don’t normally have this combination of humans and machines but tend to be restricted to just one aspect”, says Cotic. With ACTICO Platform, companies can begin with one scenario and then gradually expand and adapt as required. Another benefit of this scalability is that it allows them to make effective use of their IT budget.
A multitude of applications
Numerous companies are already using ACTICO’s Intelligent Automation Platform.
- For example, banks and other financial institutions are using it to estimate the default risk of loans to corporate customers. Parameters such as payment history, market position, sector, the company’s customer base and economic data all form part of the ML system’s decision-making process.
- Targeted advertising is another area where the ACTICO Platform brings real benefits. For example, when customers visit their online bank’s website it means they only see advertisements for relevant financial products and services. And the ML system can make real-time adjustments to the adverts to reflect any changes in their financial situation or risk tolerance.
- The ACTICO Platform can also be used to prevent fraudulent activities in online banking and commerce. The system can identify the legitimacy of a transaction much more quickly and accurately than humans.
- On the other hand, retailers can use machine learning to help them optimize their pricing strategies. It means they can determine their initial price, optimal price structure and discount strategy based on factors like demand, availability of goods, weather conditions or whether a product is “in”.
Easy implementation, huge benefits
The ACTICO Platform and its rules can be installed in just a few days and extending a rule-based solution to include a machine learning function can be done in two to four weeks.
The length of this implementation phase depends on a number of factors, such as the quality of the data sets provided and the particular use case. Experience with initial projects reveals that ACTICO’s system brings customers clear benefits, such as a 50% increase in the effectiveness of advertising banners with product recommendations. In another project, ACTICO Platform used special training procedures to achieve a 30-40% reduction in the false positive rate in fraud detection.