Search for: All records

Subspace clustering algorithms are used for understanding the cluster structure that explains the patterns prevalent in the dataset well. These methods are extensively used for data-exploration tasks in various areas of Natural Sciences. However, most of these methods fail to handle confounding attributes in the dataset. For datasets where a data sample represent multiple attributes, naively applying any clustering approach can result in undesired output. To this end, we propose a novel framework for jointly removing confounding attributes while learning to cluster data points in individual subspaces. Assuming we have label information about these confounding attributes, we regularize the clustering method by adversarially learning to minimize the mutual information between the data representation and the confounding attribute labels. Our experimental result on synthetic and real-world datasets demonstrate the effectiveness of our approach. 

Understanding the thermal stability of metallic glasses is critical to determining their safe temperatures of service. In this paper, the crystallization mechanism in spark plasma sintered Fe₄₈Cr₁₅Mo₁₄Y₂C₁₅B₆metallic glass is established by analyzing the crystal size distribution using x-ray diffraction, transmission electron microscopy andin-situsmall angle neutron scattering. Isothermal annealing at 700 °C and 725 °C for 100 min resulted in the formation of (Fe,Cr)₂₃C₆crystals, measured from transmission electron micrographs, to be from 10 to 30 nm. The small angle neutron scattering intensity measuredin-situ, over a Q-range of 0.02 to 0.3 Å⁻¹, during isothermal annealing of the sintered samples, confirmed the presence of (Fe,Cr)₂₃C₆crystals. The measured scattering intensity, fitted by the maximum entropy model, over the Q-range of 0.02 to 0.06 Å⁻¹, revealed that the crystals had radii ranging from 3 to 18 nm. The total volume fraction of crystals were estimated to be 0.13 and 0.22 upon isothermal annealing at 700 °C and 725 °C for 100 min respectively. The mechanism of crystallization in this spark plasma sintered iron based metallic glass was established to be from pre-existing nuclei as confirmed by Avrami exponents of 0.25 ± 0.01 and 0.39 ± 0.01 at the aforesaid temperatures.