http://science.sciencemag.org/content/sci/201/4359/899.2.full.pdf

Phenomenology of perception.

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

From the Publisher: 
The accessible presentation of this book gives both a general view of the entire computer vision enterprise and also offers sufficient detail to be able to build useful applications. Users learn techniques that have proven to be useful by first-hand experience and a wide range of mathematical methods. A CD-ROM with every copy of the text contains source code for programming practice, color images, and illustrative movies. Comprehensive and up-to-date, this book includes essential topics that either reflect practical significance or are of theoretical importance. Topics are discussed in substantial and increasing depth. Application surveys describe numerous important application areas such as image based rendering and digital libraries. Many important algorithms broken down and illustrated in pseudo code. Appropriate for use by engineers as a comprehensive reference to the computer vision enterprise.

Computer vision : a modern approach = 计算机视觉 : 一种现代的方法

The theory of affordances

/pdf/neural-networks-and-deep-learning-3ccs3ain8p.pdf

Neural Networks and Deep Learning

In this presentation, we discuss embodied cognition in the human brain from perspectives of spatial cognition, sensorimotor processing, face processing, and mobile EEG recordings. The argument is based upon experimental evidence gathered from five separate studies. First, we focus on spatial representations and demonstrate that, given time pressure, information on the spatial orientation of houses, independent of a participant's own location, is best retrieved when it directly relates to potential actions. Thus providing evidence that even spatial representations code information in a manner directly related to the action. Next, we discuss the concept of representations as such. Using the example of face processing in the human visual system, we argue that the concept of representations should be confined to cases where neuronal activity contains explicit information on the variable of interest and, in turn, that this variable explains the complete part of the explainable variance, i.e. reaches the noise limit. Next, to push towards an investigation of cognition under natural conditions we present a benchmark test of mobile and research-grade EEG systems. Specifically, we demonstrate that the variance over systems contributes a significant part to the total variance of recorded event related potentials. As a next step, using Independent Component Analysis of EEG data we demonstrate that in cognitive tasks some independent components systematically relate to sensory processing as well as to action execution. This supports theories of the common coding theory and, thus, a mechanistic part of the embodied cognition framework. Finally, we demonstrate a real world application investigating face processing in the form of the N170 event related potential during natural visual exploration in a fully mobile setup. This technique allows investigating the physiological basis of cognitive processes under real world conditions. In this presentation we argue that understanding cognitive processes will need to consider the (inter)actions in the natural environment.

Embodied cognition

The human visual system is an intricate network of brain regions that enables us to recognize the world around us. Despite its abundant lateral and feedback connections, object processing is commonly viewed and studied as a feedforward process. Here, we measure and model the rapid representational dynamics across multiple stages of the human ventral stream using time-resolved brain imaging and deep learning. We observe substantial representational transformations during the first 300 ms of processing within and across ventral-stream regions. Categorical divisions emerge in sequence, cascading forward and in reverse across regions, and Granger causality analysis suggests bidirectional information flow between regions. Finally, recurrent deep neural network models clearly outperform parameter-matched feedforward models in terms of their ability to capture the multiregion cortical dynamics. Targeted virtual cooling experiments on the recurrent deep network models further substantiate the importance of their lateral and top-down connections. These results establish that recurrent models are required to understand information processing in the human ventral stream.

/pdf/recurrence-is-required-to-capture-the-representational-37jw4p3wtq.pdf

Recurrence is required to capture the representational dynamics of the human visual system

Deepfakes: trick or treat?

The goal of computational neuroscience is to find mechanistic explanations of how the nervous system processes information to support cognitive function and behaviour. At the heart of the field are its models, i.e. mathematical and computational descriptions of the system being studied. These models typically map sensory stimuli to neural responses and/or neural to behavioural responses and range for simple to complex. Recently, deep neural networks (DNNs), using either feedforward and recurrent architectures, have come to dominate several domains of artificial intelligence (AI). As the term “neural network” suggests, these models are inspired by biological brains. However, current DNN models abstract from many details of biological neural networks. Their abstractions contribute to their computational efficiency, enabling to perform complex feats of intelligence, ranging from perceptual tasks (e.g. visual object and auditory speech recognition) to cognitive tasks (e.g. machine translation), and on to motor control tasks (e.g. playing computer games or controlling a robot arm). In addition to their ability to model complex intelligent behaviours, DNNs have been shown to predict neural responses to novel sensory stimuli that cannot be predicted with any other currently available type of model. DNNs can have millions of parameters (connection strengths), which are required to capture the domain knowledge needed for task performance. These parameters are often set by task training using stochastic gradient descent. The computational properties of the units are the result of four directly manipulable elements: input statistics, network structure, functional objective, and learning algorithm. The advances with neural nets in engineering provide the technological basis for building task-performing models of varying degrees of biological realism that promise substantial insights for computational neuroscience.

/pdf/deep-neural-networks-in-computational-neuroscience-1gm49d0aej.pdf

Deep Neural Networks in Computational Neuroscience

The purpose of this paper is to explain the technological phenomenon artificial intelligence (AI) and how it can contribute to knowledge-based marketing in B2B. Specifically, this paper describes the foundational building blocks of any artificial intelligence system and their interrelationships. This paper also discusses the implications of the different building blocks with respect to market knowledge in B2B marketing and outlines avenues for future research.,The paper is conceptual and proposes a framework to explicate the phenomenon AI and its building blocks. It further provides a structured discussion of how AI can contribute to different types of market knowledge critical for B2B marketing: customer knowledge, user knowledge and external market knowledge.,The paper explains AI from an input–processes–output lens and explicates the six foundational building blocks of any AI system. It also discussed how the combination of the building blocks transforms data into information and knowledge.,Aimed at general marketing executives, rather than AI specialists, this paper explains the phenomenon artificial intelligence, how it works and its relevance for the knowledge-based marketing in B2B firms. The paper highlights illustrative use cases to show how AI can impact B2B marketing functions.,The study conceptualizes the technological phenomenon artificial intelligence from a knowledge management perspective and contributes to the literature on knowledge management in the era of big data. It addresses calls for more scholarly research on AI and B2B marketing.

Tim C. Kietzmann

Papers

Embodied cognition

Recurrence is required to capture the representational dynamics of the human visual system

Deepfakes: trick or treat?

Deep Neural Networks in Computational Neuroscience

Artificial intelligence (AI) and its implications for market knowledge in B2B marketing