- Supercapacitor Materials and Fabrication
- Hydrogen Storage and Materials
- Artificial Intelligence in Games
- Advancements in Battery Materials
- Fuel Cells and Related Materials
- Conducting polymers and applications
- Reinforcement Learning in Robotics
- Quantum, superfluid, helium dynamics
- Electrocatalysts for Energy Conversion
- Modular Robots and Swarm Intelligence
- Handwritten Text Recognition Techniques
- Video Analysis and Summarization
- Advanced Condensed Matter Physics
- Ammonia Synthesis and Nitrogen Reduction
- Advanced Battery Materials and Technologies
- Magnetic and transport properties of perovskites and related materials
- Topic Modeling
- Cellular Automata and Applications
- Educational Games and Gamification
- Evolutionary Algorithms and Applications
- X-ray Diffraction in Crystallography
- Electrochemical sensors and biosensors
- Advanced Battery Technologies Research
- Ionic liquids properties and applications
- Superconductivity in MgB2 and Alloys
University of Tartu
2016-2025
KTH Royal Institute of Technology
2009-2024
IT University of Copenhagen
2020-2022
National Institute of Advanced Industrial Science and Technology
2020
Kyoto University
2020
Paul Scherrer Institute
2020
Institut Laue-Langevin
2020
Technical University of Denmark
2012-2018
SINTEF
2004
We present CloudScan; an invoice analysis system that requires zero configuration or upfront annotation. In contrast to previous work, CloudScan does not rely on templates of layout, instead it learns a single global model invoices naturally generalizes unseen layouts. The is trained using data automatically extracted from end-user provided feedback. This automatic training extraction removes the requirement for users annotate precisely. describe recurrent neural network can capture long...
Document information extraction tasks performed by humans create data consisting of a PDF or document image input, and extracted string outputs. This end-to-end is naturally consumed produced when performing the task because it valuable in itself. It available, at no additional cost. Unfortunately, state-of-the-art word classification methods for cannot use this data, instead requiring word-level labels which are expensive to consequently not available many real life tasks. In paper we...
This paper is concerned with learning to solve tasks that require a chain of interdependent steps relational inference, like answering complex questions about the relationships between objects, or solving puzzles where smaller elements solution mutually constrain each other. We introduce recurrent network, general purpose module operates on graph representation objects. As generalization Santoro et al. [2017]'s it can augment any neural network model capacity do many-step reasoning. achieve...
For hydrogen to be successfully used as an energy carrier in a new renewable driven economy, more efficient storage technologies have found. Solid-state complex metal hydrides, such sodium alanate (NaAlH4), is well-researched candidate for this application. A series of NaAlH4/mesoporous carbon black composites, with high NaAlH4 content (50–90 wt%), prepared via ball milling demonstrated significantly lower dehydrogenation temperatures intense starting at ∼373 K compared bulk alanate's ≥ 456...
The electrochemical impedance spectroscopy, the cyclic voltammetry, and constant power charge/discharge methods have been applied to establish characteristics for electrical double–layer capacitor consisting of 1 M (C2H5)3CH3NBF4 electrolyte in acetonitrile. microporous carbon electrodes prepared from D-(+)-glucose by hydrothermal carbonization method subsequent pyrolysis dioxide activation steps. total Brunauer-Emmett-Teller specific surface area (SBET ≤ 1539 m2 g−1), micropore (Smicro 1534...
Influence of specific surface area (SDFT), total pore volume (Vtot) and other porosity characteristics on the electrochemical parameters power density two electrode electric double layer capacitors based 1 M 3-ethyl-methylammonium tetrafluoroborate (Et3MeNBF4) solution in acetonitrile 1-ethyl-3-methylimidazolium (EtMeImBF4) has been analysed. The size distribution data calculated from nitrogen, CO2 Ar adsorption isotherms using mainly Carbon 2D non-local functional theory for heterogeneous...
Herein we demonstrate how peat, abundant and cheap biomass, can be successfully used as a precursor to synthesize peat-derived hard carbons (PDCs), applicable electrode materials for sodium-ion batteries (SIB). The PDCs were obtained by pre-pyrolysing peat at 300-800 °C, removing impurities with base-acid solution treatment thereafter post-pyrolysing the temperatures (T) from 1000 1500 °C. By modification of pre- post-pyrolysis low surface areas, optimal carbonization degree high...
Abstract Li-ion pouch cells that have experienced long-term accelerated testing at extreme temperatures exhibited electrolyte permeation through the seals of bag, leading to eventual cell failure due unwetting and impedance growth. Degradation bag material is also noticed. The rate follows an Arrhenius-type behavior can be substantial high (35 wt% y−1 70°C). indicated dependent on volatility solvent. Electrolyte loss mitigated by applying high-temperature epoxy sealant or switching form...
Strong confinement of hydrogen is important for adsorption-based storage solutions, which are vital the transition toward a hydrogen-based economy. The dynamics adsorbed in high-porosity TiC-derived carbon with relatively well-stacked graphenic layers carbide-derived investigated situ inelastic and quasi-elastic neutron scattering methods. Both para-ortho rotational elastic incoherent factor investigated. Hydrogen translationally bound at temperatures 20-80 K. At 50 80 K, exhibits localized...
Naturally abundant well-decomposed peat was used as a precursor for synthesizing several non-platinum group metal-type oxygen electrocatalysts. The materials were studied in an alkaline environment, where it discovered that the evolution (OER) and reduction (ORR) activity of catalysts can be severely influenced by changing parameters carbonization procedure. High OER achieved with minimally treated catalyst which seemed to because Co-rich FeCo alloy species. In both rotating disc electrode...
Methods for dynamic difficulty adjustment allow games to be tailored particular players maximize their engagement. However, current methods often only modify a limited set of game features such as the opponents, or availability resources. Other approaches, experience-driven Procedural Content Generation (PCG), can generate complete levels with desired properties that are neither too hard nor easy, but require many iterations. This paper presents method and search specific target in few...
Abstract In the quest for developing novel and efficient batteries, a great interest has been raised sustainable K-based honeycomb layer oxide materials, both their application in energy devices as well fundamental material properties. A key issue realization of batteries based on such compounds, is to understand K-ion diffusion mechanism. However, investigation potassium-ion (K $$^+$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msup> <mml:mrow/> <mml:mo>+</mml:mo>...
The influence of the synthesis conditions on formation porous and graphitic structure carbon materials was investigated Mo2C-derived synthesized at different temperatures (Tsyn from 700 to 1000 °C). Contrast matched small-angle neutron scattering method used obtain explicit information about pore wall carbons. It found that carbons formed lower Tsyn exhibit prevalently nano-scale cylindrical geometry an increase in promotes slit-like structures. Additionally, partial contrast matching used,...
Quick characterization methods to determine the structure of carbon materials are sought after for a wide array technical applications. In this study we present combined analysis carbide-derived carbons (CDCs) with Raman spectroscopy and wide-angle X-ray scattering (WAXS) methods. We optimal deconvolution method be used detailed data CDCs comparison corresponding WAXS results is made. For broad set both showed that average graphene layer extent increases synthesis temperature CDC, while...
The self-diffusion mechanisms for adsorbed H2 in different porous structures are investigated with situ quasi-elastic neutron scattering method at a temperature range from 50 K to 100 and various loadings. of the carbon materials have been characterized by sorption analysis four gases results correlated previous in-depth small-angle method. Thus, an investigation discussing effect pore shape size on nature is performed. It shown that nanometer-scale pores self-diffusing two distinguishable...
The suitability of small additions (from 0.005 M to 1 M) different organic solvents (acetonitrile, propylenecarbonate and γ-butyrolactone) improve the properties 1-ethyl 3-methylimidazolium tetrafluoroborate (EMImBF4) as an electrolyte in electrical double layer capacitors (EDLCs) has been studied. cyclic voltammetry electrochemical impedance spectroscopy (EIS) methods were used determine EDLCs filled with mixtures series capacitance, characteristic relaxation time, specific energy power...