Know Center

Ensembles have revolutionized weather forecasts, by providing a probabilistic view of future weather. However, ensembles contain a large amount of information, and using this to understand the potential impacts of the forecasted weather requires a skilled user with extensive computational resources. This challenge is particularly acute for users considering the weather conditions over a geographical region, rather than at a single location. In these cases, considering every single ensemble member and its impacts may be practically unfeasible. Many such users thus simply consider the ensemble mean and some measure of ensemble spread, such as the ensemble’s standard deviation. While this facilitates the use of ensemble forecasts, it does not explore the range of possible impacts of the forecasted weather. Here, we propose a framework facilitating the use of ensemble forecasts for weather impacts. We specifically represent the ensemble by the mean and a single deviation from the mean. This deviation is defined so as to both be representative of the variability in the ensemble, and have a significant impact according to some impact metric. We determine such a deviation using a statistical method known as Directional Component Analysis, which is based on linearizing an impact metric around the ensemble mean. We provide a concrete example using 2-m temperature forecasts for continental Europe from ECMWF, and show that this approach is more robust than considering the single worst (in terms of impacts) ensemble member. This same approach can be applied to ensembles of projections of future climates. We illustrate this by deriving representative deviations for the UKCP18 and EURO-CORDEX projections of future precipitation in Europe. We conclude that the mean and representative deviation method we propose may both contribute to automated early warnings for weather impacts and support users who wish to explore the implications of longer-term climate impacts in a resource-effective fashion.

Machine Learning (ML) and AI techniques, especially methods based on Deep Learning, have long been considered as black boxes that might be good at predicting, but not explaining predictions. This has changed recently, with more techniques becoming available that explain predictions by ML models – known as Explainable AI (XAI). These have seen adaptation also in climate science, because they could have the potential to help us in understanding the physics behind phenomena in geoscience. It is, however, unclear, how large that potential really is, and how these methods can be incorporated into the scientific process. In our study, we use the exemplary research question of which aspects of the large-scale atmospheric circulation affect specific local conditions. We compare the different answers to this question obtained with a range of different methods, from the traditional approach of targeted data analysis based on physical knowledge (such as using dimensionality reduction based on physical reasoning) to purely data-driven and physics-unaware methods using Deep Learning with XAI techniques. Based on these insights, we discuss the usefulness and potential pitfalls of XAI for understanding and explaining phenomena in geosciences. 

Biases in algorithmic systems have led to discrimination against historically disadvantaged groups, including the reinforcement of outdated gender stereotypes. While a substantial body of research addresses biases in algorithms and underlying data, in this work, we study if and how users themselves reflect these biases in their interactions with systems, which expectedly leads to the further manifestation of biases. More specifically, we investigate the replication of stereotypical gender representations by users in formulating online search queries. Following prototype theory, we define the disproportionate mention of the gender that does not conform to the prototypical representative of a searched domain (e.g., "male nurse") as an indication of bias. In a pilot study with 224 US participants and a main study with 400 UK participants, we find clear evidence of gender biases in formulating search queries. We also report the effects of an educative text on user behaviour and highlight the wish of users to learn about bias-mitigating strategies in their interactions with search engines.

This document contains the Data Management Plan (DMP) of PathOS on M6 of the project. It is a living document that was updated on M18 of the project (DMP D3.5 is avalible here https://zenodo.org/doi/10.5281/zenodo.11049556). It has been created using the ARGOS service (https://argos.openaire.eu) an online tool for creating, managing and sharing DMPs and linking them with the research artifacts they correspond to. In the DMP, we describe the input (re-used) and output (created) datasets that will be used in the technical work of PathOS, following the principles of Open (whenever possible) and FAIR as outlined in the Grant Agreement.* * The numbering of the questions skips in order to allow the updating of the DMP in the future (i.e. filling out questions for which the information is currently missing) via the Argos service. * This project deliverable is currently pending approval by the EC.

Identifying hit compounds is a principal step in early-stage drug discovery. While many machine learning (ML) approaches have been proposed, in the absence of binding data, molecular docking is the most widely used option to predict binding modes and score hundreds of thousands of compounds for binding affinity to the target protein. Docking's effectiveness is critically dependent on the protein-ligand (P-L) scoring function (SF), thus re-scoring with more rigorous SFs is a common practice. In this pilot study, we scrutinize the PM6-D3H4X/COSMO semi-empirical quantum mechanical (SQM) method as a docking pose re-scoring tool on 17 diverse receptors and ligand decoy sets, totaling 1.5 million P-L complexes. We investigate the effect of explicitly computed ligand conformational entropy and ligand deformation energy on SQM P-L scoring in a virtual screening (VS) setting, as well as molecular mechanics (MM) versus hybrid SQM/MM structure optimization prior to re-scoring. Our results proclaim that there is no obvious benefit from computing ligand conformational entropies or deformation energies and that optimizing only the ligand's geometry on the SQM level is sufficient to achieve the best possible scores. Instead, we leverage machine learning (ML) to include implicitly the missing entropy terms to the SQM score using ligand topology, physicochemical, and P-L interaction descriptors. Our new hybrid scoring function, named SQM-ML, is transparent and explainable, and achieves in average 9% higher AUC-ROC than PM6-D3H4X/COSMO and 3% higher than Glide SP, but with consistent and predictable performance across all test sets, unlike the former two SFs, whose performance is considerably target-dependent and sometimes resembles that of a random classifier. The code to prepare and train SQM-ML models is available at https://github.com/tevang/sqm-ml.git and we believe that will pave the way for a new generation of hybrid SQM/ML protein-ligand scoring functions.

A summary of the current state of the art in timeseries datamining and automated imputation.

Our aim in this paper is to present experimental evidence of a side-channel attack on superconducting cloud-based quantum computers, demonstrating that a quantum circuit can be analyzed indirectly using data gathered before and after its execution on a quantum computer. We hope that this can motivate further research into hardware and system stability of quantum computers as well as into more-nuanced systems-level attacks. Although much work has been done on establishing error correcting codes to address issues of decoherence, and schemes can be utilized to leverage quantum effects for the sake of intrusion detection over quantum communication channels, there is a lack of literature in the field regarding side-channel attacks on transpiled quantum circuits, particularly those which leverage quantum effects to gather information about quantum circuits running in the cloud. We explored issues of decoherence by examining quantum circuit behavior both before and after a specified circuit was executed in order to ascertain decalibration rates and potential sources of decoherence: our data clearly shows that information on an interim circuit can be acquired in this manner. Using convenient data classification techniques and various circuit diagrams consisting of one quantum gate and several measurements, we were able to distinguish between these various circuits when executed at position n in the interim phase by training our classifier on data from probing circuits executed at queue positions n-1 and n+1. This shows that there is sufficient information residing in the data before and after a circuit is executed on a cloud-based superconducting quantum computer to allow an attacker to ascertain information about an otherwise unknown third-party circuit and shows potential to form the basis of a side-channel attack in cases where queue positions can be examined to target a third party’s transpiled circuit during experimentation. This side channel attack merits further examination and extrapolation to identify more-complex circuits, and deeper statistical consideration to determine potential sources of information leakage and mitigation strategies.

<p>State of the art performances for entity extraction tasks are achieved by supervised learning, specifically, by fine-tuning pretrained language models such as BERT. As a result, annotating application specific data is the first step in many use cases. However, no practical guidelines are available for annotation requirements. This work supports practitioners by empirically answering<br> the frequently asked questions (1) how many training samples to annotate? (2) which examples to annotate?</p>