Learning in Convolutional Neural Networks Accelerated by Transfer Entropy

Recently, there is a growing interest in applying Transfer Entropy (TE) in quantifying the effective connectivity between artificial neurons. In a feedforward network, the TE can be used to quantify the relationships between neuron output pairs located in different layers. Our focus is on how to include the TE in the learning mechanisms of a Convolutional Neural Network (CNN) architecture. We introduce a novel training mechanism for CNN architectures which integrates the TE feedback connections. Adding the TE feedback parameter accelerates the training process, as fewer epochs are needed. On the flip side, it adds computational overhead to each epoch. According to our experiments on CNN classifiers, to achieve a reasonable computational overhead–accuracy trade-off, it is efficient to consider only the inter-neural information transfer of the neuron pairs between the last two fully connected layers. The TE acts as a smoothing factor, generating stability and becoming active only periodically, not after processing each input sample. Therefore, we can consider the TE is in our model a slowly changing meta-parameter.

An analysis of interdecadal variations in the perturbational feedback parameter based on a MIROC6 piControl simulation

The climate feedback parameter is a useful indicator for estimating climate sensitivity relating to anthropogenic forcing. This study defines a new feedback parameter, the Perturbational Feedback Parameter (PFP), and the impacts of internally-generated climate variations are clarified using the MIROC piControl simulation. PFP values are found to vary significantly on interdecadal timescales. The equatorial sea surface temperature (SST) has a positive anomaly in the eastern Pacific and a negative anomaly in the western Pacific, and the thermocline tilts more gently than usual when the PFP is large. The statistical properties of the interannual fluctuations also simultaneously vary, and they correspond to the background state. For example, there is an increase in the El Niño Southern Oscillation (ENSO) amplitude relative to the global mean surface temperature rise, and the equatorial high SST more effectively contributes to the southward shift of the Intertropical Convergence Zone (ITCZ). In addition, a decadal fluctuation that dominates over the extratropical northern Pacific also plays an important role in PFP variations. These fluctuations on broad timescales cooperatively induce increases in lower clouds within the subtropics by strengthening the descending flow and static stability, and the consequential net downward radiation flux change through increases in reflection enhances the PFP. In summary, internal changes in both tropical and extratropical variability corresponding to the background state control the strength of the climate feedback on interdecadal timescales.

The Third-Order Shear Deformation Theory for Modeling the Static Bending and Dynamic Responses of Piezoelectric Bidirectional Functionally Graded Plates

This work is the first exploration of the static bending and dynamic response analyses of piezoelectric bidirectional functionally graded plates by combining the third-order shear deformation theory of Reddy and the finite element approach, which can numerically model mechanical relations of the structure. The present approach and mechanical model are confirmed through the verification examples. The geometrical and material study is conducted to evaluate the effects of the feedback coefficients, volume fraction parameter, and constraint conditions on the static and dynamic behaviors of piezoelectric bidirectional functionally graded structures, and this work presents a wide variety of static and dynamic behaviors of the plate with many interesting results. There are many meanings that have not been mentioned by any work, especially the working performance of the structure is better than that when the feedback parameter of the piezoelectric component is added, that is, the piezoelectric layer increases the working efficiency. Numerical investigations are the important basis for calculating and designing related materials and structures in technical practice.

A small climate-amplifying effect of climate-carbon cycle feedback

The climate-carbon cycle feedback is one of the most important climate-amplifying feedbacks of the Earth system, and is quantified as a function of carbon-concentration feedback parameter (β) and carbon-climate feedback parameter (γ). However, the global climate-amplifying effect from this feedback loop (determined by the gain factor, g) has not been quantified from observations. Here we apply a Fourier analysis-based carbon cycle feedback framework to the reconstructed records from 1850 to 2017 and 1000 to 1850 to estimate β and γ. We show that the β-feedback varies by less than 10% with an average of 3.22 ± 0.32 GtC ppm−1 for 1880–2017, whereas the γ-feedback increases from −33 ± 14 GtC K−1 on a decadal scale to −122 ± 60 GtC K−1 on a centennial scale for 1000–1850. Feedback analysis further reveals that the current amplification effect from the carbon cycle feedback is small (g is 0.01 ± 0.05), which is much lower than the estimates by the advanced Earth system models (g is 0.09 ± 0.04 for the historical period and is 0.15 ± 0.08 for the RCP8.5 scenario), implying that the future allowable CO2 emissions could be 9 ± 7% more. Therefore, our findings provide new insights about the strength of climate-carbon cycle feedback and about observational constraints on models for projecting future climate.