Impact of Perceived Agent Expertise on Trust in Computer Agent Recommendations

Computer agents are frequently anthropomorphized, giving them appearances and responses similar to humans. Research has demonstrated that users tend to apply social norms and expectations to such computer agents, and that people interact with computer agents in a similar fashion as they would another human. Perceived expertise has been shown to affect trust in human-human relationships, but the literature investigating how this influences trust in computer agents is limited. The current study investigated the effect of computer agent perceived level of expertise and recommendation reliability on subjective (rated) and objective (compliance) trust during a pattern recognition task. Reliability of agent recommendations had a strong effect on both subjective and objective trust. Expert agents started with higher subjective trust, but showed less trust repair. Agent expertise had little impact on objective trust resiliency or repair.

An artificial sensory neuron with visual-haptic fusion

Human behaviors are extremely sophisticated, relying on the adaptive, plastic and event-driven network of sensory neurons. Such neuronal system analyzes multiple sensory cues efficiently to establish accurate depiction of the environment. Here, we develop a bimodal artificial sensory neuron to implement the sensory fusion processes. Such a bimodal artificial sensory neuron collects optic and pressure information from the photodetector and pressure sensors respectively, transmits the bimodal information through an ionic cable, and integrates them into post-synaptic currents by a synaptic transistor. The sensory neuron can be excited in multiple levels by synchronizing the two sensory cues, which enables the manipulating of skeletal myotubes and a robotic hand. Furthermore, enhanced recognition capability achieved on fused visual/haptic cues is confirmed by simulation of a multi-transparency pattern recognition task. Our biomimetic design has the potential to advance technologies in cyborg and neuromorphic systems by endowing them with supramodal perceptual capabilities.

On the Use of Concentrated Time–Frequency Representations as Input to a Deep Convolutional Neural Network: Application to Non Intrusive Load Monitoring

Since decades past, time–frequency (TF) analysis has demonstrated its capability to efficiently handle non-stationary multi-component signals which are ubiquitous in a large number of applications. TF analysis us allows to estimate physics-related meaningful parameters (e.g., F0, group delay, etc.) and can provide sparse signal representations when a suitable tuning of the method parameters is used. On another hand, deep learning with Convolutional Neural Networks (CNN) is the current state-of-the-art approach for pattern recognition and allows us to automatically extract relevant signal features despite the fact that the trained models can suffer from a lack of interpretability. Hence, this paper proposes to combine together these two approaches to take benefit of their respective advantages and addresses non-intrusive load monitoring (NILM) which consists of identifying a home electrical appliance (HEA) from its measured energy consumption signal as a “toy” problem. This study investigates the role of the TF representation when synchrosqueezed or not, used as the input of a 2D CNN applied to a pattern recognition task. We also propose a solution for interpreting the information conveyed by the trained CNN through different neural architecture by establishing a link with our previously proposed “handcrafted” interpretable features thanks to the layer-wise relevant propagation (LRP) method. Our experiments on the publicly available PLAID dataset show excellent appliance recognition results (accuracy above 97%) using the suitable TF representation and allow an interpretation of the trained model.

A System of Mining Semantic Trajectory Patterns from GPS Data of Real Users

Positioning devices allow users’ movement to be recorded. The GPS (Global Positioning System) trajectory data typically consists of spatiotemporal points, which make up the major part of the big data concerning urban life. Existing knowledge extraction methods about the trajectory share a general limitation—they only investigate data from a spatiotemporal aspect, but fail to take the semantic information of trajectories into consideration. Therefore, extracting the semantic information of trajectories with the context of big data is challenging pattern recognition task that has practical application prospects. In this paper, a system is proposed to extract the semantic trajectory patterns of positioning device users. Firstly, a spatiotemporal threshold and clustering based pre-processing model is proposed to process the raw data. Then, we design a probabilistic generative model to annotate the semantic information of each trajectory after the pre-processing procedure. Finally, we apply the PrefixSpan algorithm to mine the semantic trajectory patterns. We verify our system on a large dataset of users’ real trajectories over a period of 5 years in Beijing, China. The results of the experiment indicate that our system produces meaningful patterns.

Explorando os dois lados do cérebro: a percepção holística da fala reversa (PHFR) sob a perspectiva forense

Quando se reverte um sinal de fala, observam-se distorções em vários níveis linguísticos, bem como drásticas mudanças em suas características perceptivas. Informações fonéticas das estruturas segmentais e suprassegmentais são definitivamente perdidas. Contornos de energia e curvas de F0 são revertidas e, como resultado, o conteúdo da enunciação é obscurecido de tal forma que o que resta em nada lembra a língua original onde foram feitos os enunciados. Apesar das distorções, muitos parâmetros são mantidos, como, por exemplo, F0 médio, contorno de F0, valores médios dos formantes, espectro de longo termo e qualidade de voz. De acordo com os resultados da primeira pesquisa internacional sobre práticas de Comparação Forense de Locutor, a frequencia fundamental e a qualidade de voz são os parâmetros suprassegmentais mais usados e, além disso, a qualidade de voz é considerada o parâmetro de maior poder discriminativo. A tarefa de reconhecimento da voz é um processo de interação entre a análise fonétca componencial e uma tarefa de reconhecimento de padrão. Em virtude de se reter parâmetros importantes para tarefa de discriminação sem nenhum benefício de sugestões semanticas ou contextuais, a comparação de amostras reversas de fala é uma tarefa cujos processos psyco – perceptivos subjacentes operam de forma holística . Nesse artigo, abordamos a percepção holística da fala reversa, de acordo com os dados cientificos disponíveis , mostrando que a natureza ambígua desse estímulo, ao manter parametros altamente discriminativos, possibilita o seu uso na etapa perceptiva dos exames de Fonética Forense como procedimento alternativo em testes cegos por agrupamento.AbstractWhen a speech signal is reversed, there are distortions at various linguistic levels, as well as drastic changes in its perceptual characteristics. Phonetic information of the segmental and suprasgmental structures are definitely lost. Energy contours and F0 curves are reversed, and as a result, the content of the enunciation is obscured in such a way that what remains in no way resembles the original language in which the utterances were made.. Despite the distortions, many parameters are maintained, such as F0 average, F0 contour, mean values of formants, long term spectrum and voice quality. According to the results of the first international survey on Forensic Speech Comparison practices, fundamental frequency and voice quality are the most widely used suprassegmental parameters and in addition voice quality is considered the parameter of highest discriminative power for almost the totality of the research participants. The task of voice recognition is a process of interaction between componencial phonetic analysis and a pattern recognition task. Because retaining important parameters for discrimination task with no benefit from semantic or contextual cues, comparison of speech samples is a pattern recognition task whose underlying psycho - perceptual processes operate in a holistic way (Gestalt processing). In this article, we approach the holistic perception of reverse speech, according to available scientific data, showing that the ambiguous nature of this stimulus, by maintaining highly discriminative parameters, can be used in Forensic Speaking Comparison caseworks, as an alternative procedure in blind grouping.

Simulation of the Spiking Neural Network based on Practical Memristor

In order to gain a better understanding of the brain and explore biologically-inspired computation, significant attention is being paid to research into the spike-based neural computation. Spiking neural network (SNN), which is inspired by the understanding of observed biological structure, has been increasingly applied to pattern recognition task. In this work, a single layer SNN architecture based on the characteristics of spiking timing dependent plasticity (STDP) in accordance with the actual test of the device data has been proposed. The device data is derived from the Ag/GeSe/TiN fabricated memristor. The network has been tested on the MNIST dataset, and the classification accuracy attains 90.2%. Furthermore, the impact of device instability on the SNN performance has been discussed, which can propose guidelines for fabricating memristors used for SNN architecture based on STDP characteristics.

POTENTIAL FUNCTION METHOD APPROACH TO PATTERN RECOGNITION APPLICATIONS

Potential function method was originally offered to solve the pattern recognition tasks, then it was generalized to a wider range of tasks, which were associated with the function approximation. Potential function method algorithms are based on the hypothesis of the nature of the function that separates sets according to different classes of patterns. Geometrical interpretation of pattern recognition task includes display of patterns in the form of vector in the space of input signal that allows to perceive the learning as approximation task. The paper describes the essence of potential function method and the learning procedure is shown that is based on practical application of potential methods. Pattern recognition applications with the help of examples of potential functions and company bankruptcy data analysis with the help of potential functions are given.

A semiautomatic method to tie well logs to seismic data

We evaluated a semiautomatic method for well-to-seismic tying to improve correlation results and reproducibility of the procedure. In the manual procedure, the interpreter first creates a synthetic trace from edited well logs, determines the most appropriate bulk time shift and polarity, and then applies a minimum amount of stretching and squeezing to best match the observed data. The last step resembles a visual pattern recognition task, which often requires some experience. We replaced the last step with a constrained dynamic time-warping technique, to help guide the interpreter. The method automatically determined the appropriate amount of local stretching and squeezing to produce the highest correlation between the original data and the created synthetic trace. The constraint ensured that stretching and squeezing were kept within reasonable bounds, as determined by the interpreter. Results compared well with the manual method, leading to ties along the entire trace length in contrast to the shorter analysis window in the conventional method. Yet, we advise against unsupervised applications because the method is intended as a guide instead of a fully automated blind approach.