Search

We have applied non-negative matrix factorization to a database of leaf hyperspectral reflectance and DMSO chlorophyll extract absorption measurements from grapevine canopies at seven vineyards in Dalmatia, Croatia. Extracted spectral signatures show various monthly changes in grapevine production of chlorophyll. Our signatures represent maximal absorption values within a specific colour spectrum - blue, red, near infrared, green and yellow to orange. Association of signatures and chlorophyll indices vary through time. Signature S4 is the best chlorophyll proxy. Here we show that the same amount of chlorophyll can be produced by plants using multiple internal processes or absorption of different spectrums of light. Changes in these processes can be better understood by studying extracted reflection signatures instead of just chlorophyll concentration or vegetation indices. This study shows that non-negative matrix factorization applied to hyperspectral reflectance measurements can be a powerful tool in monitoring the shortterm changes in physiology of plants thus could be applied in precision viticulture. We also tested this model on canopy hyperspectral reflectance measured at 0.5 m distance from the canopy. This study shows that NMF can be a powerful tool in monitoring the seasonal grapevine changes in physiology of plants thus could be applied in precision viticulture.

In recent decades, the overpopulation on our planet has had an impact on environmental change. Addressing and understanding such changes in a reliable manner requires contemporary data-science knowledge and methodologies since the amount of information available online has been increasing at an exponential rate. Additionally, to draw meaningful conclusions and predict certain events in terms of environmental changes, one must be familiar with existing natural processes. Thus, combining data science with knowledge about global conditions in the world and how they impact Earth’s biosphere, would highly benefit the understanding of possible consequences in the environment. Since our planet is a complex, dynamic system that is likely to change, one must be knowledgeable about the global climate system as a whole. As well, one must understand its processes and interactions with the environment to be able to understand and work on preventing the potential inevitable drastic circumstances for living. This paper addresses the issue of environmental changes and presents a solution in the form of academic education, which is currently in progress, and also related to the COVID-19 issues. As shown in the results section, this type of education aims to teach students about the basics of data science and how to interpret and visualize information related to current environmental changes.

Foundation models represent a recent paradigm shift in deep learning, where a single large-scale model trained on vast amounts of data can serve as the foundation for various downstream tasks. Foundation models are generally trained using self-supervised learning and excel in reducing the demand for training samples in downstream applications. This is especially important in medicine, where large labeled datasets are often scarce. Here, we developed a foundation model for imaging biomarker discovery by training a convolutional encoder through self-supervised learning using a comprehensive dataset of 11,467 radiographic lesions. The foundation model was evaluated in distinct and clinically relevant applications of imaging-based biomarkers. We found that they facilitated better and more efficient learning of imaging biomarkers and yielded task-specific models that significantly outperformed their conventional supervised counterparts on downstream tasks. The performance gain was most prominent when training dataset sizes were very limited. Furthermore, foundation models were more stable to input and inter-reader variations and showed stronger associations with underlying biology. Our results demonstrate the tremendous potential of foundation models in discovering novel imaging biomarkers that may extend to other clinical use cases and can accelerate the widespread translation of imaging biomarkers into clinical settings., Competing Interests: COMPETING INTERESTS The authors declare no competing interests.