Feedback alignment with weight normalization can provide a biologically plausible mechanism for learning

The mechanism by which plasticity in millions of synapses in the brain is orchestrated to achieve behavioral and cognitive goals is a fundamental question in neuroscience. In this regard, insights from learning methods in artificial neural networks (ANNs) and in particular the idea of backpropagation (BP) seem inspiring. However, the implementation of BP requires exact matching of forward and backward weights, which is unrealistic given the known connectivity pattern in the brain (known as "weight transport problem"). Notably, it is recently shown that under certain conditions, error BackPropagation Through Random backward Weights (BP-TRW), can lead to partial alignment of forward and backward weights overtime (feedback alignment or FA) and result in surprisingly good accuracies in simple classification tasks using shallow ANNs. In this work, we took a closer look at FA to find out why it occurs when using BP-TRW and explored ways to boost it for deep ANNs. We first show that the gradual alignment of forward and backward weights arises from the successive application of BP-TRW update rule on forward weights regardless of learning or loss function if error signals and outputs of neurons satisfy certain conditions such as when they are autocorrelated. Moreover, we show that FA in deeper networks can be improved significantly by applying a biologically-inspired weight normalization (WN) to the input weights of each neuron. In addition, WN can improve the performance of both BP and BP-TRW when class labels are changed across time, an under-explored phenomenon in ANNs which is crucial for flexible learning in the brain in everyday life. With WN, BP-TRW test accuracy can almost match that of BP following class label changes. Altogether, our results portray a clearer picture of the FA mechanism and provide evidence for how learning can occur using BP-like mechanisms while abiding by biological limits on synaptic weights.

What Are the Relative Intensities of the Components of NMR Spectral Multiplets from Quadrupolar Nuclei in Uniformly Anisotropic Media?

Stretched hydrogels make uniformly anisotropic environments for quadrupolar nuclei such as 2H, 23Na, and 133Cs. Such surroundings cause the partial alignment of nuclear spin bearing ions and molecules that is sufficiently pronounced to alter the nuclear magnetic resonance spectra of the guest species. In most cases, resonance splittings are directly related to the spin quantum number I. The relative intensities of the components of the resonance multiplets can be inferred from basic quantum mechanics.

The Relevance of Typology for Pattern Replication

Structuralists and generativists have insisted for a long time that the elements and structures one language could borrow from another are constrained by typological compatibility, naturalness, and other factors (cf. Thomason and Kaufman 1988: 13–34). Such constraints are still thought to apply to structural interference, or pattern replication in the terms of Matras and Sakel (2007), and the often concomitant contact-induced grammaticalisation of non-native structures. This paper suggests that a priori there are no typological constraints against pattern replication in general. It is proposed that typological differences between model and replica pattern are only of relevance during the grammaticalisation and maintenance of such patterns in the replica language; in other words, typological constraints do not apply at the stage of pattern replication. It will be argued that typology, in the form of system pressure, interacts with pattern frequency and socio-historical factors, which together determine retention, adaptation, or loss of a replicated pattern. This argument is illustrated on the basis of three short studies of partial alignment change in Old Aramaic, Classical Armenian, and North-Eastern Neo-Aramaic, all of which have been in contact with Iranian languages for extended periods. In each case, Iranian ergative alignment patterns have been replicated, adapted, grammaticalised to varying degrees, and finally ousted in favour of nominative-accusative alignment. The loss of the replica pattern in each case is shown to be dependent on both typology, extent of bilingualism, and pattern frequency.

3D extruded composite thermoelectric threads for flexible energy harvesting

Whereas the rigid nature of standard thermoelectrics limits their use, flexible thermoelectric platforms can find much broader applications, for example, in low-power, wearable energy harvesting for internet-of-things applications. Here we realize continuous, flexible thermoelectric threads via a rapid extrusion of 3D-printable composite inks (Bi2Te3n- or p-type micrograins within a non-conducting polymer as a binder) followed by compression through a roller-pair, and we demonstrate their applications in flexible, low-power energy harvesting. The thermoelectric power factors of these threads are enhanced up to 7 orders-of-magnitude after lateral compression, principally due to improved conductivity resulting from reduced void volume fraction and partial alignment of thermoelectric micrograins. This dependence is quantified using a conductivity/Seebeck vise for pressure-controlled studies. The resulting grain-to-grain conductivity is well explained with a modified percolation theory to model a pressure-dependent conductivity. Flexible thermoelectric modules are demonstrated to utilize thermal gradients either parallel or transverse to the thread direction.

Parallel calibration based on modified trim strategy

Purpose Partial alignment for 3 D point sets is a challenging problem for laser calibration and robot calibration due to the unbalance of data sets, especially when the overlap of data sets is low. Geometric features can promote the accuracy of alignment. However, the corresponding feature extraction methods are time consuming. The purpose of this paper is to find a framework for partial alignment by an adaptive trimmed strategy. Design/methodology/approach First, the authors propose an adaptive trimmed strategy based on point feature histograms (PFH) coding. Second, they obtain an initial transformation based on this partition, which improves the accuracy of the normal direction weighted trimmed iterative closest point (ICP) method. Third, they conduct a series of GPU parallel implementations for time efficiency. Findings The initial partition based on PFH feature improves the accuracy of the partial registration significantly. Moreover, the parallel GPU algorithms accelerate the alignment process. Research limitations/implications This study is applicable to rigid transformation so far. It could be extended to non-rigid transformation. Practical implications In practice, point set alignment for calibration is a technique widely used in the fields of aircraft assembly, industry examination, simultaneous localization and mapping and surgery navigation. Social implications Point set calibration is a building block in the field of intelligent manufacturing. Originality/value The contributions are as follows: first, the authors introduce a novel coarse alignment as an initial calibration by PFH descriptor similarity, which can be viewed as a coarse trimmed process by partitioning the data to the almost overlap part and the rest part; second, they reduce the computation time by GPU parallel coding during the acquisition of feature descriptor; finally, they use the weighted trimmed ICP method to refine the transformation.