Random Compressed Coding with Neurons

Classical models of efficient coding in neurons assume simple mean responses--'tuning curves'--such as bell shaped or monotonic functions of a stimulus feature. Real neurons, however, can be more complex: grid cells, for example, exhibit periodic responses which impart the neural population code with high accuracy. But do highly accurate codes require fine tuning of the response properties? We address this question with the use of a benchmark model: a neural network with random synaptic weights which result in output cells with irregular tuning curves. Irregularity enhances the local resolution of the code but gives rise to catastrophic, global errors. For optimal smoothness of the tuning curves, when local and global errors balance out, the neural network compresses information from a high-dimensional representation to a low-dimensional one, and the resulting distributed code achieves exponential accuracy. An analysis of recordings from monkey motor cortex points to such 'compressed efficient coding'. Efficient codes do not require a finely tuned design--they emerge robustly from irregularity or randomness.

Comparison of Numerical HMT Codes to Simulate MBV Test of Hemp-Earth Composites

Bio-based and earth materials are growingly used for the building envelopes because of their numerous benefits such as slight environmental impact, great hygrothermal performances, effective regulation of the perceived indoor air quality and human comfort. In such materials, the phenomenon of mass transfer is complex and has a great impact on the performance of building envelope. Therefore, it is important to identify and understand the hygrothermal phenomena to be able to simulate accurately the envelope behavior. Nevertheless, the classical models that depict hygric transport within building materials seem not accurate enough for bio-based materials as they are simplified on several points of view. The correlation that exists between water content and relative humidity is mostly simplified and is modeled by a single curve, the hygric storage capacity is often overstated and the hysteresis is neglected. This paper deals with numerical study of hygric transfer within hemp-earth building material by using WUFI® Pro 6.5, a commercial software, and TMC code developed at the LGCGM (Moissette and Bart, 2009) . This code was validated regarding EN 15026 standard (Moissette and Bart, 2009) and has evolved over the years by integrating the hysteresis phenomena (Aït-Oumeziane et al., 2015). Thus, a significant enhancement of the numerical simulations on desorption phase was shown. This study investigates the simulation of MBV test performed on a hemp-earth material for which only the adsorption curve is known as input. Missing parameters (water vapor permeability and desorption curve) are fitted considering the first cycle of MBV test with TMC code. Then, MBV test is simulated with WUFI® Pro 6.5 and TMC code without and with hysteresis. The results highlight the need to include hysteresis to accurately simulate dynamic hygric phenomena, and show that it is possible to find missing parameters by fitting dynamic solicitations.

Modeling of Supervised Machine Learning using Mechanism of Quantum Computing.

Mechanism of quantum computing helps to propose several task of machine learning in quantum technology. Quantum computing is enriched with quantum mechanics such as superposition and entanglement for making new standard of computation which will be far different than classical computer. Qubit is sole of quantum technology and help to use quantum mechanism for several tasks. Tasks which are non-computable by classical machine can be solved by quantum technology and these tasks are classically hard to compute and categorised as complex computations. Machine learning on classical models is very well set but it has more computational requirements based on complex and high-volume data processing. Supervised machine learning modelling using quantum computing deals with feature selection, parameter encoding and parameterized circuit formation. This paper highlights on integration of quantum computation and machine learning which will make sense on quantum machine learning modeling. Modelling of quantum parameterized circuit, Quantum feature set design and implementation for sample data is discussed. Supervised machine learning using quantum mechanism such as superposition and entanglement are articulated. Quantum machine learning helps to enhance the various classical machine learning methods for better analysis and prediction using complex measurement.

Rose-Patisserie and Coffee House

This case traces the life of a new endeavor, starting with a small patisserie and coffeehouse and the subsequent development of the business, considering three alternatives, namely optimizing the concept, expanding through a franchise network, and building a network of company-owned stores. The story of Rui and Joana raises a wide range of issues that managers need to address. After reading and working through the case, students will be able to evaluate the product portfolio, based on actual sales data, and to evaluate and propose strategic options using classical models.

Attention Deeplabv3 model and its application into gear pitting measurement

An efficient visual detection method is explored in this study to address the low accuracy and efficiency of manual detection for irregular gear pitting. The results of gear pitting detection are enhanced by embedding two attention modules into Deeplabv3 + to obtain an improved segmentation model called attention Deeplabv3. The attention mechanism of the proposed model endows the latter with an enhanced ability for feature representation of small and irregular objects and effectively improves the segmentation performance of Deeplabv3. The segmentation ability of attention Deeplabv3+ is verified by comparing its performance with those of other typical segmentation networks using two public datasets, namely, Cityscapes and Voc2012. The proposed model is subsequently applied to segment gear pitting and tooth surfaces simultaneously, and the pitting area ratio is calculated. Experimental results show that attention Deeplabv3 has higher segmentation performance and measurement accuracy compared with the existing classical models under the same computing speed. Thus, the proposed model is suitable for measuring various gear pittings.

The role of pseudo-overdominance in maintaining inbreeding depression

Classical models ignoring linkage predict that deleterious recessive mutations purge or fix within inbred populations, yet these often retain moderate to high segregating load. True overdominance generates balancing selection that sustains inbreeding depression even in inbred populations but is rare. In contrast, arrays of mildly deleterious recessives linked in repulsion may occur commonly enough to generate pseudo-overdominance and sustain segregating load. We used simulations to explore how long pseudo-overdominant regions (POD's) persist following their creation via hybridization between populations fixed for alternative mutations at linked loci. Balancing haplotype loads, tight linkage, and moderate to strong cumulative selective effects serve to maintain POD's, suggesting that POD's may most often arise and persist in low recombination regions (e.g., inversions). Selection and drift unbalance the load, eventually eliminating POD's, but this process is very slow when pseudo-overdominance is strong. Background selection across the genome accelerates the loss of weak POD's but reinforces strong POD's in inbred populations by disfavoring homozygotes. Further modeling and studies of POD dynamics within populations could help us understand how POD's affect persistence of the load and how inbred mating systems evolve.

ἀφθόνητος αἶνος Su tre lemmi pindarici dell’Antiatticista

Among other peculiarities, the 2nd-century CE Atticist lexicon that goes under the name of Antiatticista contains seven entries exemplified with references to Pindar (not an Attic author), a fact that sets it apart from other Atticist lexica of the same period. This paper tackles the verbal adjective ἀφθόνητος and the irregular comparatives ἀφθονέστερος and ἀρχαιέστερος in order to show that two criteria guided the inclusion of these Pindaric words into the lexicon. The first, and more superficial, criterion concerns the word-formation of verbal adjectives and comparatives, and their relation with other (often more regular or more frequent) forms. The second criterion concerns semantic change, and especially the use of certain words in post-Classical and Byzantine Greek vis-à-vis the Classical models. The consideration of both criteria allows a more fine-grained interpretation of the Antiatticista’s methodology and its recourse to a wide range of Classical authors to illustrate, and defend, developments of post-Classical Greek.

Local dendritic balance enables learning of efficient representations in networks of spiking neurons

How can neural networks learn to efficiently represent complex and high-dimensional inputs via local plasticity mechanisms? Classical models of representation learning assume that feedforward weights are learned via pairwise Hebbian-like plasticity. Here, we show that pairwise Hebbian-like plasticity works only under unrealistic requirements on neural dynamics and input statistics. To overcome these limitations, we derive from first principles a learning scheme based on voltage-dependent synaptic plasticity rules. Here, recurrent connections learn to locally balance feedforward input in individual dendritic compartments and thereby can modulate synaptic plasticity to learn efficient representations. We demonstrate in simulations that this learning scheme works robustly even for complex high-dimensional inputs and with inhibitory transmission delays, where Hebbian-like plasticity fails. Our results draw a direct connection between dendritic excitatory–inhibitory balance and voltage-dependent synaptic plasticity as observed in vivo and suggest that both are crucial for representation learning.

Simple dynamics in non-monotone Kolmogorov systems

In this paper we analyse the global dynamical behaviour of some classical models in the plane. Informally speaking we prove that the folkloric criteria based on the relative positions of the nullclines for Lotka–Volterra systems are also valid in a wide class of discrete systems. The method of proof consists of dividing the plane into suitable positively invariant regions and applying the theory of translation arcs in a subtle manner. Our approach allows us to extend several results of the theory of monotone systems to nonmonotone systems. Applications in models with weak Allee effect, population models for pioneer-climax species, and predator–prey systems are given.