Скачать с ютуб Explainable AI and DevOps/MLOps for Banking - Unicredit в хорошем качестве

Explainable AI and DevOps/MLOps for Banking - Unicredit

Explainable AI in the Banking Industry - Ilaria Bordino, senior researcher at UniCredit Artificial Intelligence (AI) has come a long way from the earlier days of its inception, and it has become an integral part of our contemporary society. The improvements in computing power and the exponential increase in data availability that have characterized the last decades have helped AI grow at the speed of light, and to pervasively penetrate all major industries with countless real-world applications. However, the unprecedented rise of AI has also opened new critical challenges and obstacles. One such challenge is that current AI models are often obscure black boxes for their users, who are unable to understand the logic behind the behavior of the models, given that the size of training data and the model complexity are too big for human capabilities. With AI being increasingly used to provide automatic support to decision processes, the inability to understand the rationale beyond a decision and to obtain an explanation for it is a serious drawback that may heavily limit the trust in AI systems, and therefore their widespread adoption. The pressing need to gain better insights into how AI systems make their decision has given rise to the research field of Explainable AI (XAI), which aims at making AI systems, their behavior, and their predictions more comprehensible to the end users. In this talk we will briefly overview the state of the art on Explainable AI, and then we will discuss applications to real-world use cases in the banking industry. Application of DevOps and MLOps towards digital automation in banking use-cases - Bhaskar Chakraborty, AI Architect/ML engineer at UniCredit The banks are considered as the fundamental financial institutions that cater the pecuniary needs for a large portion of the population through several thousands of internal workflows originated via in-person (physical branches) and digital (certified emails, voice messages, chatbots) communications. These internal workflows also extend the services for the foreign customers as well. Introducing AI/ML algorithms in different stages of these workflows also known as digital automation is a crucial step to improve the end-to-end customer experience towards reliability, security and execution time. The digital automation process often results in several micro-services for each workflow-based use-case including AI/ML model development and serving layers. So, the key question arises on how to choose an optimal strategy and tools/techniques to organize these micro-services in a framework of continuous integration and deployment. Recent advancement in the area of Development Operations (DevOps) and its extension towards Machine Learning Operations (MLOps) would act as the fundamental pillars towards this end. In this presentation we would be talking about how the application of DevOps and MLOps in the digital automation process improves the development, deployment and maintenance of the production applications for different banking use-cases. Topics: #ExplainableAI #DevOps #MLOps