Cultured Cortical Neurons Can Perform Blind Source Separation According to the Free-Energy Principle.
ABSTRACT: Blind source separation is the computation underlying the cocktail party effect--a partygoer can distinguish a particular talker's voice from the ambient noise. Early studies indicated that the brain might use blind source separation as a signal processing strategy for sensory perception and numerous mathematical models have been proposed; however, it remains unclear how the neural networks extract particular sources from a complex mixture of inputs. We discovered that neurons in cultures of dissociated rat cortical cells could learn to represent particular sources while filtering out other signals. Specifically, the distinct classes of neurons in the culture learned to respond to the distinct sources after repeating training stimulation. Moreover, the neural network structures changed to reduce free energy, as predicted by the free-energy principle, a candidate unified theory of learning and memory, and by Jaynes' principle of maximum entropy. This implicit learning can only be explained by some form of Hebbian plasticity. These results are the first in vitro (as opposed to in silico) demonstration of neural networks performing blind source separation, and the first formal demonstration of neuronal self-organization under the free energy principle.
Project description:Computed Tomography (CT) is a powerful method for non-destructive testing (NDT) and metrology awakes with expanding application fields. To improve the spatial resolution of high energy CT, a micro-spot gamma-ray source based on bremsstrahlung from a laser wakefield accelerator was developed. A high energy CT using the source was performed, which shows that the resolution of reconstruction can reach 100??m at 10% contrast. Our proof-of-principle demonstration indicates that laser driven micro-spot gamma-ray sources provide a prospective way to increase the spatial resolution and toward to high energy micro CT. Due to the advantage in spatial resolution, laser based high energy CT represents a large potential for many NDT applications.
Project description:Deep learning has become a widespread tool in both science and industry. However, continued progress is hampered by the rapid growth in energy costs of ever-larger deep neural networks. Optical neural networks provide a potential means to solve the energy-cost problem faced by deep learning. Here, we experimentally demonstrate an optical neural network based on optical dot products that achieves 99% accuracy on handwritten-digit classification using ~3.1 detected photons per weight multiplication and ~90% accuracy using ~0.66 photons (~2.5 × 10<sup>-19</sup> J of optical energy) per weight multiplication. The fundamental principle enabling our sub-photon-per-multiplication demonstration-noise reduction from the accumulation of scalar multiplications in dot-product sums-is applicable to many different optical-neural-network architectures. Our work shows that optical neural networks can achieve accurate results using extremely low optical energies.
Project description:Animals need to adjust their inferences according to the context they are in. This is required for the multi-context blind source separation (BSS) task, where an agent needs to infer hidden sources from their context-dependent mixtures. The agent is expected to invert this mixing process for all contexts. Here, we show that a neural network that implements the error-gated Hebbian rule (EGHR) with sufficiently redundant sensory inputs can successfully learn this task. After training, the network can perform the multi-context BSS without further updating synapses, by retaining memories of all experienced contexts. This demonstrates an attractive use of the EGHR for dimensionality reduction by extracting low-dimensional sources across contexts. Finally, if there is a common feature shared across contexts, the EGHR can extract it and generalize the task to even inexperienced contexts. The results highlight the utility of the EGHR as a model for perceptual adaptation in animals.
Project description:The hippocampus and amygdala are key brain structures of the medial temporal lobe, involved in cognitive and emotional processes as well as pathological states such as epilepsy. Despite their importance, it is still unclear whether their neural activity can be recorded non-invasively. Here, using simultaneous intracerebral and magnetoencephalography (MEG) recordings in patients with focal drug-resistant epilepsy, we demonstrate a direct contribution of amygdala and hippocampal activity to surface MEG recordings. In particular, a method of blind source separation, independent component analysis, enabled activity arising from large neocortical networks to be disentangled from that of deeper structures, whose amplitude at the surface was small but significant. This finding is highly relevant for our understanding of hippocampal and amygdala brain activity as it implies that their activity could potentially be measured non-invasively.
Project description:Both CO and H2 can be utilized as energy sources during the autotrophic growth of Clostridium ljungdahlii. In principle, CO is a more energetically and thermodynamically favorable energy source for gas fermentation in comparison to H2. Therefore, metabolism may vary during growth under different energy sources. In this study, C. ljungdahlii was fed with CO and/or CO2/H2 at pH 6.0 with a gas pressure of 0.1 MPa. C. ljungdahlii primarily produced acetate in the presence of H2 as an energy source, but produced alcohols with CO as an energy source under the same fermentation conditions. A key enzyme activity assay, metabolic flux analysis, and comparative transcriptomics were performed for investigating the response mechanism of C. ljungdahlii under different energy sources. A CO dehydrogenase and an aldehyde:ferredoxin oxidoreductase were found to play important roles in CO utilization and alcohol production. Based on these findings, novel metabolic schemes are proposed for C. ljungdahlii growing on CO and/or CO2/H2. These schemes indicate that more ATP is produced during CO-fermentation than during H2-fermentation, leading to increased alcohol production.
Project description:Spike timing is precise in the auditory system and it has been argued that it conveys information about auditory stimuli, in particular about the location of a sound source. However, beyond simple time differences, the way in which neurons might extract this information is unclear and the potential computational advantages are unknown. The computational difficulty of this task for an animal is to locate the source of an unexpected sound from two monaural signals that are highly dependent on the unknown source signal. In neuron models consisting of spectro-temporal filtering and spiking nonlinearity, we found that the binaural structure induced by spatialized sounds is mapped to synchrony patterns that depend on source location rather than on source signal. Location-specific synchrony patterns would then result in the activation of location-specific assemblies of postsynaptic neurons. We designed a spiking neuron model which exploited this principle to locate a variety of sound sources in a virtual acoustic environment using measured human head-related transfer functions. The model was able to accurately estimate the location of previously unknown sounds in both azimuth and elevation (including front/back discrimination) in a known acoustic environment. We found that multiple representations of different acoustic environments could coexist as sets of overlapping neural assemblies which could be associated with spatial locations by Hebbian learning. The model demonstrates the computational relevance of relative spike timing to extract spatial information about sources independently of the source signal.
Project description:In the paper, we proposed a new approach to producing the qubits in electron transport in low-dimensional structures such as double quantum wells or double quantum wires (DQW). The qubit could arise as a result of quantum entanglement of two specific states of electrons in DQW structure. These two specific states are the symmetric and antisymmetric (with respect to inversion symmetry) states arising due to tunneling across the structure, while entanglement could be produced and controlled by means of the source of nonclassical light. We examined the possibility to produce quantum entanglement in the framework of Jaynes-Cummings model and have shown that at least in principle, the entanglement can be achieved due to series of "revivals" and "collapses" in the population inversion due to the interaction of a quantized single-mode EM field with a two-level system.
Project description:With the advent of the big data era, applications are more data-centric and energy efficiency issues caused by frequent data interactions, due to the physical separation of memory and computing, will become increasingly severe. Emerging technologies have been proposed to perform analog computing with memory to address the dilemma. Ferroelectric memory has become a promising technology due to field-driven fast switching and non-destructive readout, but endurance and miniaturization are limited. Here, we demonstrate the ?-In<sub>2</sub>Se<sub>3</sub> ferroelectric semiconductor channel device that integrates non-volatile memory and neural computation functions. Remarkable performance includes ultra-fast write speed of 40?ns, improved endurance through the internal electric field, flexible adjustment of neural plasticity, ultra-low energy consumption of 234/40 fJ per event for excitation/inhibition, and thermally modulated 94.74% high-precision iris recognition classification simulation. This prototypical demonstration lays the foundation for an integrated memory computing system with high density and energy efficiency.
Project description:Symmetry is an organizational principle that is ubiquitous throughout the visual world. However, this property can also be detected through non-visual modalities such as touch. The role of prior visual experience on detecting tactile patterns containing symmetry remains unclear. We compared the behavioral performance of early blind and sighted (blindfolded) controls on a tactile symmetry detection task. The tactile patterns used were similar in design and complexity as in previous visual perceptual studies. The neural correlates associated with this behavioral task were identified with functional magnetic resonance imaging (fMRI). In line with growing evidence demonstrating enhanced tactile processing abilities in the blind, we found that early blind individuals showed significantly superior performance in detecting tactile symmetric patterns compared to sighted controls. Furthermore, comparing patterns of activation between these two groups identified common areas of activation (e.g. superior parietal cortex) but key differences also emerged. In particular, tactile symmetry detection in the early blind was also associated with activation that included peri-calcarine cortex, lateral occipital (LO), and middle temporal (MT) cortex, as well as inferior temporal and fusiform cortex. These results contribute to the growing evidence supporting superior behavioral abilities in the blind, and the neural correlates associated with crossmodal neuroplasticity following visual deprivation.
Project description:Oil/water mixtures are a potentially major source of environmental pollution if efficient separation technology is not employed during processing. A large volume of oil/water mixtures is produced via many manufacturing operations in food, petrochemical, mining, and metal industries and can be exposed to water sources on a regular basis. To date, several techniques are used in practice to deal with industrial oil/water mixtures and oil spills such as <i>in situ</i> burning of oil, bioremediation, and solidifiers, which change the physical shape of oil as a result of chemical interaction. Physical separation of oil/water mixtures is in industrial practice; however, the existing technologies to do so often require either dissipation of large amounts of energy (such as in cyclones and hydrocyclones) or large residence times or inventories of fluids (such as in decanters). Recently, materials with selective wettability have gained attention for application in separation of oil/water mixtures and surfactant stabilized emulsions. For example, a superhydrophobic material is selectively wettable toward oil while having a poor affinity for the aqueous phase; therefore, a superhydrophobic porous material can easily adsorb the oil while completely rejecting the water from an oil/water mixture, thus physically separating the two components. The ease of separation, low cost, and low-energy requirements are some of the other advantages offered by these materials over existing practices of oil/water separation. The present review aims to focus on the surface engineering aspects to achieve selectively wettability in materials and its their relationship with the separation of oil/water mixtures with particular focus on emulsions, on factors contributing to their stability, and on how wettability can be helpful in their separation. Finally, the challenges in application of superwettable materials will be highlighted, and potential solutions to improve the application of these materials will be put forward.