The gradient clusteron: A model neuron that learns to solve classification tasks via dendritic nonlinearities,structural plasticity,and gradient descent

Synaptic clustering on neuronal dendrites has been hypothesized to play an important role in implementing pattern recognition. Neighboring synapses on a dendritic branch can interact in a synergistic, cooperative manner via nonlinear voltage-dependent mechanisms, such as NMDA receptors. Inspired by the NMDA receptor, the single-branch clusteron learning algorithm takes advantage of location-dependent multiplicative nonlinearities to solve classification tasks by randomly shuffling the locations of “under-performing” synapses on a model dendrite during learning (“structural plasticity”), eventually resulting in synapses with correlated activity being placed next to each other on the dendrite. We propose an alternative model, the gradient clusteron, or G-clusteron, which uses an analytically-derived gradient descent rule where synapses are "attracted to" or "repelled from" each other in an input- and location-dependent manner. We demonstrate the classification ability of this algorithm by testing it on the MNIST handwritten digit dataset and show that, when using a softmax activation function, the accuracy of the G-clusteron on the all-versus-all MNIST task (~85%) approaches that of logistic regression (~93%). In addition to the location update rule, we also derive a learning rule for the synaptic weights of the G-clusteron (“functional plasticity”) and show that a G-clusteron that utilizes the weight update rule can achieve ~89% accuracy on the MNIST task. We also show that a G-clusteron with both the weight and location update rules can learn to solve the XOR problem from arbitrary initial conditions.     

The dichotomy of yield and drought resistance: Translation challenges from basic research to crop adaptation to climate change

Global climate change and the increasing human population require crop varieties with higher yield and draught resistance. But meeting both goals is not an easy task for breeders and plant science.

The human population is increasing and so does the demand on food production. The Food and Agriculture Organization of the United Nations (FAO) predicts that in order to meet the global food demands by 2050, the production of staple cereal crops must be doubled at least (FAO, 2017), which means that the current rate of yield improvement needs to increase by at least 40%. Crop breeders are expected to cope with this challenge and come up with novel high‐yield varieties, but the prospects of even maintaining the current rate of yield improvement in light of climate change are unclear. To meet the growing demand for food and increase the yield of staple crops, we need a better understanding of how plants adapt to environmental factors that limit their productivity in terms of turning sunlight and CO₂ into tissues and seeds.

To meet the growing demand for food and increase the yield of staple crops, we need a better understanding of how plants adapt to environmental factors that limit their productivity…

Although nature provides many examples of how plants adapt to harsh environments, these are rarely suitable for use in an agronomic environment, mainly owing to the economics: Any stress‐tolerance variety must also be profitable for the farmer. If a stress response mechanism enables the plant to survive but reduces yield, it will not be economical and, therefore, not be used by farmers. Thus, understanding the key parameters limiting crop yield—plant‒environment interactions, in particular—will help us to cope with the food security challenges presented by changing environmental conditions. In particular, this knowledge helps to inform breeding programmes to more efficiently create and screen for crop varieties to meet the challenges of population growth and climate change. This is not an easy task.Plants are autotrophic; sessile organisms and their productivity completely depends on the temperature, light levels, and the availability of inorganic substances in the soil. Terrestrial plants are further, and primarily, limited by the availability of water, as the absorption of CO₂ from the air requires water: A few hundred water molecules are lost for each CO₂ molecule absorbed. Therefore, understanding the mechanisms that maintain water balance is critical for optimizing crop growth and fruit production in any given environment.     

Numerical solutions of the Lamm equation. IV. Rotor slowing experiments

This paper presents the results of a numerical solution of the Lamm equation for roton slowing specified by ω² = ω₀² exp {– λτ}, for parameters relevant to equilibrium experiments. It. is shown that in the two-component system it is theoretically possible to deduce s/D from measurements of the difference of concentration across the. cell with rotor slowing, provided that the time at which the difference reaches a maximum is known.     

Elution properties of inhomogeneous chromatographic systems

The first two moments of residence time in an inhomogeneous column are calculated by means of first passage time theory. These can be used to find the resolving power of the column for nearly equal amounts of protein. A recently developed perturbation theory allows the calculation of approximate elution profiles.     

Kinetics of sedimentation in a density gradient

Two models of sedimentation in a density gradient are analyzed. The first is for sedimentation in cylindrical sector geometry and contains the assumption that diffusion can be neglected. The second treats sedimentation in a rectangular field and includes diffusion, although the boundaries are not treated exactly. In both of these models we approximate the time dependence of the gradient by a relaxation form. We derive exact results for both models. It is also shown that the sedimentation coefficient can be calculated from data by following the motion of the position of the maximum (or minimum) of the concentration gradient.     

Helicobacter pylori and Peptic Ulcer Disease Therapies: A Survey of Gastroenterologists in Israel

Background. Eradication of Helicobacter pylori has become a therapeutic option in the treatment of patients with peptic ulcer disease. The aim of this study was to evaluate the current management strategies of Israeli gastroenterologists in the diagnosis and treatment of H. pylori- related peptic ulcer disease, 14 years after the discovery of H. pylori.
Materials and Methods. A questionnaire was mailed to all specialists in gastroenterology, members of the Israel Gastroenterological Association (IGA). Replies were received from 60% of Israel Board-certified gastroenterology specialists.
Results. Over 89% of the gastroenterologists (89.1%) noted that they recommend anti- H. pylori treatment. 94.5% said that they treat duodenal ulcer in the first presentation with anti- H. pylori medication and 75% said that they do so in cases of recurrent duodenal ulcer. According to the replies received, there is a strong consensus towards triple treatment as the favored anti -H. pylori treatment; no one noted the use of dual treatment. Seven-day triple treatment was prescribed by 83.6% of the gastroenterologists who responded. Of these, the great majority, 89.1%, stated that they use proton pump inhibitors (PPI) in combination with any two of the following antibiotics: metronidazole (47.3%), tinidazole (29.1%), clarithromycin (61.8%), and amoxicillin (40%).
Conclusion. At the time of the survey, most Israel Board-certified gastroenterology specialists prescribed triple anti- H. pylori treatment of one-week's duration.     

Numerical solutions of the Lamm equation. III. Velocity centrifugation

We have generated solutions to the Lamm equation to examine the effects of concentration dependence on velocity experiments. Two forms of c dependence are considered: s/s₀ = 1 – kc and s/s₀ = (1 + kc)^?1. Features of these solutions are discussed. The magnitude of the errors resulting from the usual procedure of measuring the rate of movement of schlieren maxima or of the position at which the concentration is one half the plateau value have been examined. These errors are usually negligible after sufficient centrifugation time. The errors in using the half-plateau concentration are less than those using the movement of the peak. We have also examined a technique due to Fujita for determining D from boundary spreading when s/s₀ = (1+kc)^?1. This method is satisfactory when s/s₀ is actually of this form, or under certain limitations when s/s₀ = (1 + kc)^?1. Creeth has shown that under certain conditions the concentration gradient, curve remains virtually unchanged in shape after separating from the meniscus. When this occurs it is possible to estimate s/D from the data. The condition for such a steady state is that kc₀ be sufficiently large. Numerical confirmation of this method is presented in the final section.     

Species-specific ability ofChlorella strains (Chlorophyceae) to form stable symbioses withHydra viridis

46 strains ofChlorella, identified by physiological and biochemical characters, were examined for their ability to form stable symbioses with aposymbioticHydra viridis. It was found to be a species-specific characteristic. Among the 15 taxa studied, onlyC. saccharophila var.ellipsoidea, C. saccharophila var.saccharophila, C. fusca var.vacuolata, C. kessleri, C. luteoviridis, andC. protothecoides formed stable symbioses withHydra viridis. Among the 11 known physiological and biochemical characters of theseChlorella species, only acid tolerance seems to be correlated with symbiosis: All symbiotic species are capable of growing at or below pH 4.0.