Descripció del projecte
In the recent years, Artificial intelligence (AI)-based systems increasingly build high-impact solutions for a wide-range of real-world problems. For example, in the health care industry, AI can be used to improve patient care and reduce costs [1]. In the financial sector, AI can be used to detect fraud and improve investment decisions [2]. In the transportation sector, AI can be included in embedded systems inside cars, flights or even used to optimize traffic flow and prevent accidents [3]. Or in retail, AI can be used to personalize product recommendations and increase sales [4].
The AI ability of learning to solve complex problems makes it an attractive option for many industries to build generically accurate forecasters. However, the requirements of real-world problems are often characterized to go beyond the need of having good predictive systems, but also to prevent jeopardized scenarios while planning safe actions or analyze risks based on the previous experience. To support such capabilities, AI requires to additionally model the surrounding uncertainty and to identify the causal impacts and effects of any potential action.
The present research project has the goal of providing new AI generative models to handle and report the intrinsically uncertainty and to simulate the causal impacts of potential interventions. This will be applied into several Pharmaceutical industrial problems where risk should be considered such as finance forecasting, competitor impact analysis or investment effect prediction but also can be extended to other domains such as computer vision problems where Safety requirements are needed. In this framework, generative models are understood as mathematical models working in high-dimensional spaces (where Deep Learning methods tend to stand out [5]) that deal with all the uncertainty sources and simulate the effect of interventions on the desired system [6]. The intervention could be anything from a change in an input price to the introduction of a new drug in the marketplace over time. Consequently, the goal of the generative model is to take into account all of the factors that could influence the outcome, considering the effects of all the uncertainty sources from a probabilistic viewpoint [7, 8]. As a result, more informed and trustable decisions about what interventions are likely to be the most effective will be produced and shown, which requires also to research in visualization techniques to properly illustrate such information and help into the decision-making process [9].
[1] Matheny, M. E., Whicher, D., & Israni, S. T. (2020). Artificial intelligence in health care: a report from the National Academy of Medicine. Jama, 323(6), 509-510.
[2] Davenport, T. H., & Ronanki, R. (2018). Artificial intelligence for the real world. Harvard business review, 96(1), 108-116.
[3] Abduljabbar, R., Dia, H., Liyanage, S., & Bagloee, S. A. (2019). Applications of artificial intelligence in transport: An overview. Sustainability, 11(1), 189.
[4] Kumar, V., Rajan, B., Venkatesan, R., & Lecinski, J. (2019). Understanding the role of artificial intelligence in personalized engagement marketing. California Management Review, 61(4), 135-155.
[5] Pang, G., Shen, C., Cao, L., & Hengel, A. V. D. (2021). Deep learning for anomaly detection: A review. ACM Computing Surveys (CSUR), 54(2), 1-38.
[6] Schwab, P., & Karlen, W. (2019). Cxplain: Causal explanations for model interpretation under uncertainty. Advances in Neural Information Processing Systems, 32.
[7] Kendall, A., & Gal, Y. (2017). What uncertainties do we need in bayesian deep learning for computer vision?. Advances in neural information processing systems, 30.
[8] Lopez, R., Regier, J., Cole, M. B., Jordan, M. I., & Yosef, N. (2018). Deep generative modeling for single-cell transcriptomics. Nature methods, 15(12), 1053-1058.
[9] Chung, Y., Char, I., Guo, H., Schneider, J., & Neiswanger, W. (2021). Uncertainty toolbox: an open-source library for assessing, visualizing, and improving uncertainty quantification. arXiv preprint arXiv:2109.10254.
