We propose a new approach to document image layout extraction using rapid feature analysis, preclassification and predictive coding. First, a set of layout features is used to render the image profile information. The...
详细信息
We propose a new approach to document image layout extraction using rapid feature analysis, preclassification and predictive coding. First, a set of layout features is used to render the image profile information. The knowledge base is utilized to rule these early regions into layout labels. The regions found are given a classification tag and a degree of membership into background, text, picture and line drawing classes. A predictive coding method is used with the preclassification information to increase the confidence of each label, and to integrate the regional domain and the labels into a uniform class without any shape assumption. We have tested our technique using three different databases that comprise over 1000 document images. The results show a high degree of confidence in region separation and extraction. The main benefits include robust classification shape independency and rapid computation.
Lossless compression of datasets is a problem of significant theoretical and practical interest. It appears naturally in the task of storing, sending, or archiving large collections of information for scientific resea...
详细信息
Lossless compression of datasets is a problem of significant theoretical and practical interest. It appears naturally in the task of storing, sending, or archiving large collections of information for scientific research. We can greatly improve encoding bitrate if we allow the compression of the original dataset to decompress to a permutation of the data. We prove the equivalence of dataset compression to compressing a permutation-invariant structure of the data and implement such a scheme via predictive coding. We benchmark our compression procedure against state-of-the-art compression utilities on the popular machine-learning datasets MNIST and CIFAR-10 and outperform for multiple parameter sets.
We investigate predictive coding for reducing the amount of data communicated between a haptic controller and a host. This allows increased update rate, which potentially improves quality even if coding is lossy. A lo...
详细信息
We investigate predictive coding for reducing the amount of data communicated between a haptic controller and a host. This allows increased update rate, which potentially improves quality even if coding is lossy. A low-order predictive coding is investigated for a pneumatic force display. Due to human and device characteristics, some compression is possible without loss, although the technique is lossy in general. Lossy uniform and nonuniform quantizers are also investigated. An experiment was conducted to determine how much data reduction is possible before compression artifacts become detectable to users.
The current paper presents a novel recurrent neural network model, predictive multiple spatio-temporal scales RNN (P-MSTRNN), which can generate as well as recognize dynamic visual patterns in a predictive coding fram...
详细信息
ISBN:
(纸本)9781509061839
The current paper presents a novel recurrent neural network model, predictive multiple spatio-temporal scales RNN (P-MSTRNN), which can generate as well as recognize dynamic visual patterns in a predictive coding framework. The model is characterized by multiple spatio-temporal scales imposed on neural unit dynamics through which an adequate spatio-temporal hierarchy develops via learning from exemplars. The model was evaluated by conducting an experiment of learning a set of whole body human movement patterns, which was generated by following a hierarchically defined movement syntax. The analysis of the trained model clarifies what types of spatio-temporal hierarchy develops in dynamic neural activity as well as how robust generation and recognition of movement patterns can be achieved by using the error minimization principle.
Our group has explored possible neuropsychological mechanisms for social cognition by using predictive coding and active inference frameworks [1]. For the purpose of gaining better understanding, we have taken so-call...
详细信息
ISBN:
(纸本)9781450369220
Our group has explored possible neuropsychological mechanisms for social cognition by using predictive coding and active inference frameworks [1]. For the purpose of gaining better understanding, we have taken so-called the synthetic robotics approach wherein a set of experiments have been conducted for robot-human as well as robot-robot interactions. Especially, we examine the underlying mechanisms accounting for spontaneous coupling and decoupling among agents as well as autonomous shifts from one social context to another. We investigate also how can novel or creative behaviors be co-developed by robots and human tutors through their developmental interactive tutoring processes. Finally, I address phenomenological aspects in social cognition from our preliminary examinations on how human can feel intention or free will of the robots or how the robots can possibly do so for the humans in the human-in-the-robot-loop experiment.
Three dimensional display of moving images greatly enhances realism and adds a unique sense of "presence". Three dimensional video systems have been kept from widespread application by two technical problems...
详细信息
Three dimensional display of moving images greatly enhances realism and adds a unique sense of "presence". Three dimensional video systems have been kept from widespread application by two technical problems, the need for glasses, viewing hoods, or other cumbersome devices for image steering, and the high bandwidths needed for transmission. Devices that avoid the discomfort of headgear by using autostereoscopic (pseudo-holographic) displays are known, but these methods require even higher bandwidths to be effective. This paper introduces the use of digital predictive coding as a means of data compression for the transmission or storage of a set of spatially related images needed for an autostereoscopic display. (Interframe coding without frame memories.) The algorithms, implementations, and application of a new sort of predictor called Disparity Corrected Prediction are described.
predictive coding can be regarded as a function which reduces the error between an input signal and a top-down prediction. If reducing the error is equivalent to reducing the influence of stimuli from the environment,...
详细信息
ISBN:
(数字)9781728124858
ISBN:
(纸本)9781728124865
predictive coding can be regarded as a function which reduces the error between an input signal and a top-down prediction. If reducing the error is equivalent to reducing the influence of stimuli from the environment, predictive coding can be regarded as stimulation avoidance by prediction. Our previous studies showed that action and selection for stimulation avoidance emerge in spiking neural networks through spike-timing dependent plasticity (STDP). In this study, we demonstrate that spiking neural networks with random structure spontaneously learn to predict temporal sequences of stimuli based solely on STDP.
The reasons underlying the success of predictive coding are outlined. Proper attention to auditory perception and masking of the quantizing noise have resulted in coders with high quality at 0.5 bits/residual sample. ...
详细信息
The reasons underlying the success of predictive coding are outlined. Proper attention to auditory perception and masking of the quantizing noise have resulted in coders with high quality at 0.5 bits/residual sample. Ever more powerful chips have transported these basic principles into the realm of real-time reality. For sustained progress, current research should focus on 1. The effects of dynamic masking: during and immediately following rapid transitions in the speech spectrum the ear seems more tolerant to rough coding. The saving in average information rate could be substantial. 2. Better algebraically structured codes: codes are needed that are fast, simple to implement and allow the incorporation of subjective criteria. 3. Efficient coding of the parameters: because of the great success with coding the residual, sparse representation of the predictor coefficients is now paramount.
暂无评论