In this work, we propose a novel method in order to rank the most relevant legal principle citations in law-cases to support a certain motion. The first score relies on feature importance metrics, where each law article is a feature supplied to a classifier for the decision outcome. The second score is based on word embeddings text similarity. As a result, our method outperforms the baseline techniques based on feature importance selection and Information Retrieval methods in the ranking evaluation relevance criteria.

An extension to a recently introduced architecture of clique-based neural networks is presented. This extension makes it possible to store sequences with high efficiency. To obtain this property, network connections are provided with orientation and with flexible redundancy carried by both spatial and temporal redundancies, a mechanism of anticipation being introduced in the model. In addition to the sequence storage with high efficiency, this new scheme also offers biological plausibility. In order to achieve accurate sequence retrieval, a double-layered structure combining heteroassociation and autoassociation is also proposed.

Coded recurrent neural networks with three levels of sparsity are introduced. The first level is related to the size of messages that are much smaller than the number of available neurons. The second one is provided by a particular coding rule, acting as a local constraint in the neural activity. The third one is a characteristic of the low final connection density of the network after the learning phase. Though the proposed network is very simple since it is based on binary neurons and binary connections, it is able to learn a large number of messages and recall them, even in presence of strong erasures. The performance of the network is assessed as a classifier and as an associative memory.

Thanks to the probabilistic message passing performed between its component decoders, a turbo decoder is able to provide strong error correction close to the theoretical limit. However, the minimum Hamming distance (dmin) of a turbo code may not be sufficiently large to ensure large asymptotic gains at very low error rates (the so-called flattening effect). Increasing the dminof a turbo code may involve using component encoders with a large number of states, devising more sophisticated internal permutations, or increasing the number of component encoders. This paper addresses the latter option and proposes a modified turbo code in which a fraction of the parity bits are encoded by a rate-1, third encoder. The result is a noticeably increased dmin, which improves turbo decoder performance at low error rates. Performance comparisons with turbo codes and serially concatenated convolutional codes are given.

An improved turbo-equalization process is proposed for linear precoding with iterative decoding. Using recent results in turbo-equalization, we derive the equalization coefficients for an optimized preceding decoder thanks to the SINR criterion. The proposed method keeps a low complexity even with large size precoders and high order modulations. Theoretical and simulation results show that full diversity is achieved after three or four iterations over a flat Rayleigh fading channel. An application of this transmission scheme is presented in the context of the OFDM wireless LAN 802.11a. We observe a gain of several decibels compared to the conventional unprecoded transmission scheme.

The original turbo codes (TCs), presented in 1993 by Berrou et al., consist of the parallel concatenation of two rate-1/2 binary recursive systematic convolutional (RSC) codes. This paper explains how replacing rate-1/2 binary component codes by rate-m/(m+1) binary RSC codes can lead to better global performance. The encoding scheme can be designed so that decoding can be achieved closer to the theoretical limit, while showing better performance in the region of low error rates. These results are illustrated with some examples based on double-binary (m=2) 8-state and 16-state TCs, easily adaptable to a large range of data block sizes and coding rates. The double-binary 8-state code has already been adopted in several telecommunication standards.

The design of the turbo code internal permutation is crucial regarding the minimum Hamming distance (MHD) and the related achievable asymptotic gain. After having set down the permutation issues in a synthetic way, this paper presents a generic model which allows us to obtain large MHDs for turbo codes and also offers large possibilities of natural parallelism in the decoder architecture.

This article describes a new concatenation scheme for turbo codes, a hybrid model between serial and parallel turbo codes, which offers improved asymptotic performance. We review the classical regular interleaving problems and we show how this new hybrid model allows us to keep a simple regular interleaving while outperforming more complex permutations at low error rates. Simulation results are presented and compared with well established turbo codes.

A new method for computing the minimum distances of linear error-correcting codes is proposed and justified. Unlike classical techniques that rely on exhaustive or partial enumeration of codewords, this new method is based on the ability of the Soft-In decoder to overcome Error Impulse input patterns. It is shown that the maximum magnitude of the Error Impulse that can be corrected by the decoder is directly related to the minimum distance. This leads to a very fast algorithm to obtain minimum distances of any linear code whatever the block size and the code rate considered. In particular, the method can be advantageously worked out for turbo-like concatenated codes.

A new method for computing minimum distances, based on the ability of the soft-in decoder (if it exists) to overcome error impulse patterns, was recently proposed. This error impulse method, which delivers the results in record time, makes it possible to easily predict the asymptotic behaviour of the code considered. In this paper, we address the association of a turbo code and an M-PSK modulation, with the so-called pragmatic or bit-interleaved coded modulation (BICM) approach, and we develop a rationale to enable the performance prediction of such a coded scheme.

Non-binary turbo codes are built from recursive systematic convolutional (RSC) component codes with m inputs (m/spl ges/2). This construction offers better performance than classical binary turbo codes, especially at very low BER and high coding rate. Some examples with rate 2/3 are given on 8-state and 16-state duo-binary turbo code associated with QPSK and 8-PSK modulation.

This paper presents a new family of convolutional codes, nicknamed turbo-codes, built from a particular concatenation of two recursive systematic codes, linked together by nonuniform interleaving. Decoding calls on iterative processing in which each component decoder takes advantage of the work of the other at the previous step, with the aid of the original concept of extrinsic information. For sufficiently large interleaving sizes, the correcting performance of turbo-codes, investigated by simulation, appears to be close to the theoretical limit predicted by Shannon.

Technological evolution and the ever-increasing demand for higher-quality services give broadcasters a strong incentive to completely digitize their broadcasting networks. This digitization, which is already well advanced in many program production areas and transmission links, now has to be extended to complete the last link in the broadcast chain; i.e., from broadcast transmitter to consumer receivers. It is therefore necessary to develop wholly new techniques for the broadcasting of digitally coded TV programmes. Thus an efficient baseband digital coding must be combined with a robust digital modulation and channel coding scheme that can meet the requirements of every mode of broadcast reception. This article presents the research work related to the coded orthogonal frequency division multiplex (COFDM) technology, which has now been completed in the field of digital radio (DAB), and which is under progress in the field of digital terrestrial TV.>

The co-occurrence matrices method, a statistical approach in texture analysis, has been shown to be useful in discriminating between textures. However, one limitation to this method is the computation throughput, since the matrix size is the square of the total number of grey levels. This affects the processing speed, but it can be significantly improved by parallel processing and pipelining, using specialized VLSI chips. A device using semisystolic architecture for fast co-occurrence matrix computation is presented. This device will allow the operational exploitation of statistical methods for discrimination between different textures in image processing.>