maps to machine learning pattern recognition*

Over recent years, phase separation has been revealed from the nanometer to micron scales in various transition metal oxides. These maps have offered the possibility to extract and analyze cluster size, critical exponents etc… at snapshot temperatures across the transition. Tracking fine changes in the cluster dynamics at slow temperature sweep has, however, not yet been possible. To do so we have developed a 77K-600K variable temperature autofocus microscope operating in the visible range. Up to 1000 images can be typically measured crossing T_c. We will first review the experimental setup and image analysis developed to complete this study. We will then review key results obtained on Vanadium Dioxide, VO₂: critical temperature maps (T_c maps), ΔT_c hysteretic maps, memory maps, switching maps, resistor network and pattern recognition of clusters using machine learning.