We propose a generalization of neural network sequence models. Instead of predicting one symbol at a time, our multi-scale model makes predictions over multiple, potentially overlapping multi-symbol tokens. A variation of the byte-pair encoding (BPE) compression algorithm is used to learn the dictionary of tokens that the model is trained with. When applied to language modelling, our model has ...

Algorithms for encoding and decoding finite strings over a finite alphabet are described. The coding operations are arithmetic involving rational numbers li as parameters such that Zi2"i 5 2". This coding technique requires no blocking, and the per-symbol length of the encoded string approaches the associated entropy within E . The coding speed is comparable to that of conventional coding metho...

6 A novel method for efficient encoding human body motion, extracted from image sequences is presented. 7 Optical flow field is calculated from sequential images, and the part of the flow field containing a person is 8 subdivided into six segments. For each of the segments, a two dimensional, eight-bin histogram of optical 9 flow is calculated. A symbol is generated, corresponding to the bin wi...

This paper studies dual-hop amplify-and-forward relaying system employing differential encoding and decoding over time-varying Rayleigh fading channels. First, the convectional “two-symbol” differential detection (CDD) is theoretically analysed in terms of the bit-error-rate (BER). The obtained analysis clearly shows that performance of two-symbol differential detection severely degrades in fas...

The communication capacity of Gaussian bosonic channels with memory has recently attracted much interest. Here, we investigate a method to prepare the multimode entangled input symbol states for encoding classical information into these channels. In particular, we study the usefulness of a Gaussian matrix-product state (GMPS) as an input symbol state, which can be sequentially generated althoug...

We present a simple but accurate parser which exploits both large tree fragments and symbol refinement. We parse with all fragments of the training set, in contrast to much recent work on tree selection in data-oriented parsing and tree-substitution grammar learning. We require only simple, deterministic grammar symbol refinement, in contrast to recent work on latent symbol refinement. Moreover...

The detection problem in the Gaussian interference channel is addressed, when each user employs an encoding scheme that is designed for the single-user Gaussian channel. A structure consisting of a separate symbol-by-symbol detector and a hard decoder is considered that is suboptimal compared to the maximum likelihood (ML) receiver, but also less complex. Given this structure, an optimal detect...

We present a simple but accurate parser which exploits both large tree fragments and symbol refinement. We parse with all fragments of the training set, in contrast to much recent work on tree selection in data-oriented parsing and treesubstitution grammar learning. We require only simple, deterministic grammar symbol refinement, in contrast to recent work on latent symbol refinement. Moreover,...

A blind equalizer attempts to compensate the inter-symbol interference caused by a communication channel without the knowledge of the transmitted sequence. Linear prediction-error lters (PEF) can be used to obtain the equalized symbols. These equalizers are derived assuming a white information sequence. In real communication systems, channel encoding is commonly used to enable the detection and...

Fountain codes are becoming increasingly important for data transferring over dedicated high-speed long-distance network. However, the encoding and decoding complexity of traditional fountain codes such as LT and Raptor codes are still high. In this paper, a new fountain codes named LRF (Loss Rate Based Fountain) codes for data transfer is proposed. In order to improve the performance of encodi...