DEBI-NN: Distance-encoding biomorphic-informational neural networks for minimizing the number of trainable parameters.

Modern artificial intelligence (AI) approaches mainly rely on neural network (NN) or deep NN methodologies. However, these approaches require large amounts of data to train, given, that the number of their trainable parameters has a polynomial relationship to their neuron counts. This property renders deep NNs challenging to apply in fields operating with small, albeit representative datasets such as healthcare. In this paper, we propose a novel neural network architecture which trains spatial positions of neural soma and axon pairs, where weights are calculated by axon-soma distances of connected neurons. We refer to this method as distance-encoding biomorphic-informational (DEBI) neural network. This concept significantly minimizes the number of trainable parameters compared to conventional neural networks. We demonstrate that DEBI models can yield comparable predictive performance in tabular and imaging datasets, where they require a fraction of trainable parameters compared to conventional NNs, resulting in a highly scalable solution.

Densification: A Route towards Enhanced Thermal Conductivity of Epoxy Composites.

When an amorphous polymer is cooled under pressure from above its glass transition temperature to room temperature, and then the pressure is released, this results in a densified state of the glass. This procedure applied to an epoxy composite system filled with boron nitride (BN) particles has been shown to increase the density of the composite, reduce its enthalpy, and, most importantly, significantly enhance its thermal conductivity. An epoxy-BN composite with 58 wt% BN platelets of average size 30 µm has been densified by curing under pressures of up to 2.0 MPa and then cooling the cured sample to room temperature before releasing the pressure. It is found that the thermal conductivity is increased from approximately 3 W/mK for a sample cured at ambient pressure to approximately 7 W/mK; in parallel, the density increases from 1.55 to 1.72 ± 0.01 g/cm3. This densification process is much more effective in enhancing the thermal conductivity than is either simply applying pressure to consolidate the epoxy composite mixture before curing or applying pressure during cure but then removing the pressure before cooling to room temperature; this last procedure results in a thermal conductivity of approximately 5 W/mK. Furthermore, it has been shown that the densification and corresponding effect on the thermal conductivity is reversible; it can be removed by heating above the glass transition temperature and then cooling without pressure and can be reinstated by again heating above the glass transition temperature and then cooling under pressure. This implies that a densified state and an enhanced thermal conductivity can be induced even in a composite prepared without the use of pressure.

Remarkable Thermal Conductivity of Epoxy Composites Filled with Boron Nitride and Cured under Pressure.

This work demonstrates that the application of even moderate pressures during cure can result in a remarkable enhancement of the thermal conductivity of composites of epoxy and boron nitride (BN). Two systems have been used: epoxy-thiol and epoxy-diamine composites, filled with BN particles of different sizes and types: 2, 30 and 180 µm platelets and 120 µm agglomerates. Using measurements of density and thermal conductivity, samples cured under pressures of 175 kPa and 2 MPa are compared with the same compositions cured at ambient pressure. The thermal conductivity increases for all samples cured under pressure, but the mechanism responsible depends on the composite system: For epoxy-diamine composites, the increase results principally from a reduction in the void content; for the epoxy-thiol system with BN platelets, the increase results from an improved matrix-particle interface; for the epoxy-thiol system with BN agglomerates, which has a thermal conductivity greater than 10 W/mK at 44.7 vol.% filler content, the agglomerates are deformed to give a significantly increased area of contact. These results indicate that curing under pressure is an effective means of achieving high conductivity in epoxy-BN composites.

Thermal Conductivity and Cure Kinetics of Epoxy-Boron Nitride Composites-A Review.

Epoxy resin composites filled with thermally conductive but electrically insulating particles play an important role in the thermal management of modern electronic devices. Although many types of particles are used for this purpose, including oxides, carbides and nitrides, one of the most widely used fillers is boron nitride (BN). In this review we concentrate specifically on epoxy-BN composites for high thermal conductivity applications. First, the cure kinetics of epoxy composites in general, and of epoxy-BN composites in particular, are discussed separately in terms of the effects of the filler particles on cure parameters and the cured composite. Then, several fundamental aspects of epoxy-BN composites are discussed in terms of their effect on thermal conductivity. These aspects include the following: the filler content; the type of epoxy system used for the matrix; the morphology of the filler particles (platelets, agglomerates) and their size and concentration; the use of surface treatments of the filler particles or of coupling agents; and the composite preparation procedures, for example whether or not solvents are used for dispersion of the filler in the matrix. The dependence of thermal conductivity on filler content, obtained from over one hundred reports in the literature, is examined in detail, and an attempt is made to categorise the effects of the variables and to compare the results obtained by different procedures.

Achieving High Thermal Conductivity in Epoxy Composites: Effect of Boron Nitride Particle Size and Matrix-Filler Interface.

For the thermal management of high watt density circuit layers, it is common to use a filled epoxy system to provide an electrically insulating but thermally conducting bond to a metal substrate. An epoxy-thiol system filled with boron nitride (BN), in the form of 2, 30 and 180 µm platelets, has been investigated with a view to achieving enhanced thermal conductivity. The effect of BN content on the cure reaction kinetics has been studied by differential scanning calorimetry and the thermal conductivity of the cured samples has been measured by the Transient Hot Bridge method. The heat of reaction and the glass transition temperature of the fully cured samples are both independent of the BN content, but the cure reaction kinetics is systematically affected by both BN content and particle size. These results can be correlated with the thermal conductivity of the cured systems, which is found to increase with both BN content and particle size. For a given BN content, the thermal conductivity found here is significantly higher than most others reported in the literature; this effect is attributed to a Lewis acid-base interaction between filler and matrix.