Unsupervised and semi-supervised clustering by message passing: soft-constraint affinity propagation

Soft-constraint affinity propagation (SCAP) is a new statistical-physics based clustering technique [M. Leone, Sumedha, M. Weigt, Bioinformatics 23, 2708 (2007)]. First we give the derivation of a simplified version of the algorithm and discuss possibilities of time- and memory-efficient implementations. Later we give a detailed analysis of the performance of SCAP on artificial data, showing that the algorithm efficiently unveils clustered and hierarchical data structures. We generalize the algorithm to the problem of semi-supervised clustering, where data are already partially labeled, and clustering assigns labels to previously unlabeled points. SCAP uses both the geometrical organization of the data and the available labels assigned to few points in a computationally efficient way, as is shown on artificial and biological benchmark data.     

Electrical Studies on Pentacene Thin Film Transistors with Different Channel Widths

In order to conduct electrical studies on organic thin film transistors, top-contact devices are fabricated by growing polycrystalline films of freshly synthesized pentacene over Si/SiO2 substrates with two different channel widths under identical conditions. Reasonable field effect mobilities in order of 10^-2-10^-3 cm^2V^-1s^-1 are obtained in these devices. An elaborative electrical characterization of all the devices is undertaken to study the variance in output saturation current, field effect mobility, and leakage current with aging under ambient conditions. As compared to the devices with longer channel width, the devices with shorter channel width exhibit better electrical performance initially. However, the former devices sustain the moderate performance much longer than the latter ones.     

Effect of Eliminating Edges on Robustness of Scale-Free Networks under Intentional Attack

We study the robustness of complex networks under edge elimination. We propose three different edge elimination strategies and investigate their effects on the robustness of scale-free networks under intentional attack. We show that deleting a proper fraction of edges connecting hub nodes and hub nodes can enhance the robustness of scale-free networks under intentional attack.     

Influence of TM and RE elements on glass formation of the ternary Al-TM-RE systems

The effects of TM and RE elements on glass formation ability (GFA) of the Al-TM-RE systems are studied systematically. The TM elements show distinct differences: critical sizes of the Al-TM-Ce systems are in the order of Ni > Co > Fe > Cu. However, the RE elements show similar effect on glass formation. These results are discussed in terms of the GFA-related factors, i.e., reduced glass transition temperature (T_rg = T_g (T_x)/T_L, where T_g, T_x and T_L are the glass transition temperature, onset crystallization temperature and liquidus temperature, respectively) and the mixing enthalpy ΔH_Al-TM. A new parameter of T_L − T_g (T_x) is potentially proposed and used to evaluate the GFA of the Al-based alloys, as well as bulk glass-formers.