Reconstructive Reservoir Computing to Detect Anomaly in Time-series Signals

We propose reconstructive reservoir computing (RRC) for anomaly detection working for time-series signals. This paper investigates its fundamental properties with experiments employing echo state networks (ESNs). The RRC model is a reconstructor to replicate a normal input time-series signal with no delay or a certain delay (delay ≥ 0). In its anomaly detection process, we evaluate instantaneous reconstruction error defined as the difference between input and output signals at each time. Experiments with a sound dataset from industrial machines demonstrate that the error is low for normal signals while it becomes higher for abnormal ones, showing successful anomaly detection. It is notable that the RRC models’ behavior is very different from that of conventional anomaly detection models, that is, those based on forecasting (delay < 0). The error of the proposed reconstructor is explicitly lower than that of a forecaster, resulting in superior distinction between normal and abnormal states. We show that the RRC model is effective over a large range of reservoir parameters. We also illustrate the distribution of the output weights optimized through a training to discuss their roles in the reconstruction. Then, we investigate the influence of the neuronal leaking rate and the delay time shift amount on the transient response and the reconstruction error, showing high effectiveness of the reconstructor in anomaly detection. The proposed RRC will play a significant role for anomaly detection in the present and future sensor network society

Reconstructive Reservoir Computing to Detect Anomaly in Time-series Signals

We propose reconstructive reservoir computing (RRC) for anomaly detection working for time-series signals. This paper investigates its fundamental properties with experiments employing echo state networks (ESNs). The RRC model is a reconstructor to replicate a normal input time-series signal with no delay or a certain delay (delay ≥ 0). In its anomaly detection process, we evaluate instantaneous reconstruction error defined as the difference between input and output signals at each time. Experiments with a sound dataset from industrial machines demonstrate that the error is low for normal signals while it becomes higher for abnormal ones, showing successful anomaly detection. It is notable that the RRC models’ behavior is very different from that of conventional anomaly detection models, that is, those based on forecasting (delay < 0). The error of the proposed reconstructor is explicitly lower than that of a forecaster, resulting in superior distinction between normal and abnormal states. We show that the RRC model is effective over a large range of reservoir parameters. We also illustrate the distribution of the output weights optimized through a training to discuss their roles in the reconstruction. Then, we investigate the influence of the neuronal leaking rate and the delay time shift amount on the transient response and the reconstruction error, showing high effectiveness of the reconstructor in anomaly detection. The proposed RRC will play a significant role for anomaly detection in the present and future sensor network society

The distributional properties of exemplars affect category learning and generalization

What we learn about the world is affected by the input we receive. Many extant category learning studies use uniform distributions as input in which each exemplar in a category is presented the same number of times. Another common assumption on input used in previous studies is that exemplars from the same category form a roughly normal distribution. However, recent corpus studies suggest that real-world category input tends to be organized around skewed distributions. We conducted three experiments to examine the distributional properties of the input on category learning and generalization. Across all studies, skewed input distributions resulted in broader generalization than normal input distributions. Uniform distributions also resulted in broader generalization than normal input distributions. Our results not only suggest that current category learning theories may underestimate category generalization but also challenge current theories to explain category learning in the real world with skewed, instead of the normal or uniform distributions often used in experimental studies.

Development of a New Type of Geodiversity System for the Scoria Cones of the Chaîne des Puys Based on Geomorphometric Studies

The aesthetic beauty of a landscape is an integral value reflected in artistic inspiration. Science, in contrast, tries to quantify the landscape using various methods. Of these, geodiversity indices have been found to be a useful approach, and this geomorphological diversity is characterized through derivatives made from digital terrain models (DTM). While these methods are useful, they have a drawback that the value of some landscape features may be underestimated if they have regular forms. For example, the aesthetic and scientific attractiveness of our study area, the Chaîne des Puys (Auvergne, France), a UNESCO World Heritage site, is strongly related to the distinctive small volcanoes, but despite being an outstanding element of the landscape, the scoria cones do not stand out well in geodiversity indices. This is because they have almost symmetrical conical forms and regular slopes that score low in the available geodiversity methods. We explore this problem and investigate how to overcome the low geodiversity performance of these distinctive landscape elements. We propose a modified approach for scoria cones using the normal input layers but adapted to the cone geometry. The modified indices are easy to compute and consider the uniformity and symmetry of larger landscape elements that form scientifically integral and aesthetically vital components of the landscape. The method is applicable to the tens of thousands of small monogenetic volcanoes in the hundreds of volcanic fields around the world, and could be extended to other volcanic features, such as domes. It would be possible to use the method to study larger volcanoes, as they often share and replicate the small-scale monogenetic morphology considered here.

The Precompression Processing of LMS Algorithm in Noise Elimination

In this study, the authors propose a novel precompression processing (PCP) of the least mean squares (LMS) algorithm based on a regulator factor. The novelty of the PCP algorithm is that the compressed input signals vary from each other on different components at each iteration. The input signal of the improved LMS algorithm is precompressed based on the regulator factor. The precompressed input signal is not only related to the regulator factor α and the current value of the input signal at each iteration but also related to the amplitude of the input signal before this iteration. The improved algorithm can eliminate the influence of input signal mutation on the filter performance. In the numerical simulations, we compare the improved LMS algorithm and NLMS algorithm in the cases of normal input signal and input signal with mutation and the influence of different regulator factors on the noise elimination. Results show that the PCP algorithm has good noise elimination effect when the input signal changes abruptly and the regulator factor α = 0.01 can meet the requirements.

Effects of Non-Normal Input Distributions and Sampling Region on Monte Carlo Results

In order to address the risks associated with the operation of ageing pressure boundary components, many assessments incorporate probabilistic analysis tools for alleviating excessive conservatism of deterministic methodologies. In general, deterministic techniques utilize conservative bounding values for all critical parameters. Recently, various Probabilistic Fracture Mechanics (PFM) codes have been employed to identify governing parameters which could affect licensing basis margins of pressure retaining components. Moreover, these codes are used to calculate a probability of failure in order to estimate potential risks under operating and design loading conditions for the pressure retaining components experiencing plausible and active degradation mechanisms. Probabilistic approaches typically invoke the Monte-Carlo (MC) method where a set of critical input variables are randomly distributed and inserted in deterministic computer models. Estimates of results from probabilistic assessments are then compared against various assessment criteria. During the PVP-2016 conference, we investigated the assumption of normality of the Monte Carlo results utilizing a non-linear system function. In this paper, we extend the study by employing non-normal input distributions and investigating the effects of sampling region on the system function.

Underlying Mechanisms of Tinnitus: Review and Clinical Implications

Background: The study of tinnitus mechanisms has increased tenfold in the last decade. The common denominator for all of these studies is the goal of elucidating the underlying neural mechanisms of tinnitus with the ultimate purpose of finding a cure. While these basic science findings may not be immediately applicable to the clinician who works directly with patients to assist them in managing their reactions to tinnitus, a clear understanding of these findings is needed to develop the most effective procedures for alleviating tinnitus. Purpose: The goal of this review is to provide audiologists and other health-care professionals with a basic understanding of the neurophysiological changes in the auditory system likely to be responsible for tinnitus. Results: It is increasingly clear that tinnitus is a pathology involving neuroplastic changes in central auditory structures that take place when the brain is deprived of its normal input by pathology in the cochlea. Cochlear pathology is not always expressed in the audiogram but may be detected by more sensitive measures. Neural changes can occur at the level of synapses between inner hair cells and the auditory nerve and within multiple levels of the central auditory pathway. Long-term maintenance of tinnitus is likely a function of a complex network of structures involving central auditory and nonauditory systems. Conclusions: Patients often have expectations that a treatment exists to cure their tinnitus. They should be made aware that research is increasing to discover such a cure and that their reactions to tinnitus can be mitigated through the use of evidence-based behavioral interventions.