A5051848285- Machine Learning in Materials Science
- Advanced Chemical Physics Studies
- Computational Drug Discovery Methods
- X-ray Diffraction in Crystallography
- Mind wandering and attention
- Perfectionism, Procrastination, Anxiety Studies
- Graph theory and applications
- History and advancements in chemistry
- Spectroscopy and Quantum Chemical Studies
- Molecular spectroscopy and chirality
- Advanced Combinatorial Mathematics
- Electron and X-Ray Spectroscopy Techniques
- Theoretical and Computational Physics
- Visual and Cognitive Learning Processes
- Landslides and related hazards
- Anomaly Detection Techniques and Applications
- Inorganic Fluorides and Related Compounds
- Online Learning and Analytics
- High-pressure geophysics and materials
- Virtual Reality Applications and Impacts
- Spatial Cognition and Navigation
- Complex Systems and Time Series Analysis
- Innovative Teaching and Learning Methods
- Human Pose and Action Recognition
- Time Series Analysis and Forecasting
University of Basel
2019-2022
Swiss Distance University of Applied Sciences
2022
École Polytechnique Fédérale de Lausanne
2021
Sharif University of Technology
2019
Atomic environment fingerprints are widely used in computational materials science, from machine learning potentials to the quantification of similarities between atomic configurations. Many approaches construction such fingerprints, also called structural descriptors, have been proposed. In this work, we compare performance based on Overlap Matrix(OM), Smooth Positions (SOAP), Behler-Parrinello atom-centered symmetry functions (ACSF), modified (MBSF) ANI-1ccx potential and...
Atomic fingerprints are commonly used for the characterization of local environments atoms in machine learning and other contexts. In this work, we study behavior two widely fingerprints, namely, smooth overlap atomic positions (SOAP) atom-centered symmetry functions (ACSFs), under finite changes demonstrate existence manifolds quasi-constant fingerprints. These found numerically by following eigenvectors sensitivity matrix with quasi-zero eigenvalues. The such ACSF SOAP causes a failure to...
Procrastination, the irrational delay of tasks, is a common occurrence in online learning. Potential negative consequences include higher risk drop-outs, increased stress, and reduced mood. Due to rise learning management systems analytics, indicators such behavior can be detected, enabling predictions future procrastination other dilatory behavior. However, research focusing on scarce. Moreover, studies involving different types predictors comparisons between predictive performance various...
Human pose forecasting involves complex spatiotemporal interactions between body parts (e.g., arms, legs, spine). State-of-the-art approaches use Long Short-Term Memories (LSTMs) or Variational AutoEncoders (VAEs) to solve the problem. Yet, they do not effectively predict human motions when both global trajectory and local movements exist. We propose learn decoupled representations for tasks. also show that it is better stop prediction uncertainty in motion increases. Our model outperforms...
The (100) surface of CaF<sub>2</sub>exhibits a large number nearly degenerate reconstructions as well superionicity.
Using fingerprints used mainly in machine learning schemes of the potential energy surface, we detect a fully algorithmic way long range effects on local physical properties simple covalent system carbon atoms. The fact that these exist for many configurations implies atomistic simulation methods, such as force fields or modern schemes, are based locality assumptions, limited accuracy. We show basic driving mechanism is charge transfer. If transfer known, can be recovered certain quantities...
Fingerprint distances, which measure the similarity of atomic environments, are commonly calculated from environment fingerprint vectors. In this work, we present simplex method that can perform inverse operation, i.e., calculating vectors distances. The found in way point to corners a simplex. For large dataset fingerprints, find particular largest simplex, whose dimension gives effective vector space. We show correspond landmark environments be used fully automatic analyze structures. way,...
Recent research has produced mixed results regarding the effectiveness of learning in VR. It been suggested that rich multisensory input VR may induce cognitive overload impedes process. Cognitive load is typically measured by administering questionnaires. Although questionnaires are easily used, they imply need to interrupt students during or assess retrospect. In this work-in-progress paper, we argue motion tracking data potential provide unobtrusive, yet valid measures load. We report...
Using methods borrowed from machine learning we detect in a fully algorithmic way long range effects on local physical properties simple covalent system of carbon atoms. The fact that these exist for many configurations implies atomistic simulation methods, such as force fields or modern schemes, are based locality assumptions, limited accuracy. We show the basic driving mechanism is charge transfer. If transfer known, can be recovered certain quantities band structure energy.
Procrastination, the irrational delay of tasks, is a common occurrence in online learning. Potential negative consequences include higher risk drop-outs, increased stress, and reduced mood. Due to rise learning management systems analytics, indicators such behavior can be detected, enabling predictions future procrastination other dilatory behavior. However, research focusing on scarce. Moreover, studies involving different types predictors comparisons between predictive performance various...