IBM (France)

Conventional equalization and carrier recovery algorithms for minimizing mean-square error in digital communication systems generally require an initial training period during which a known data sequence is transmitted and properly synchronized at the receiver. This paper solves the general problem of adaptive channel equalization without resorting to a known training sequence or to conditions of limited distortion. The criterion for equalizer adaptation is the minimization of a new class of nonconvex cost functions which are shown to characterize intersymbol interference independently of carrier phase and of the data symbol constellation used in the transmission system. Equalizer convergence does not require carrier recovery, so that carrier phase tracking can be carried out at the equalizer output in a decision-directed mode. The convergence properties of the self-recovering algorithms are analyzed mathematically and confirmed by computer simulation.

International audience

The problems of fast channel estimation and fast start-up equalization in synchronous digital communication systems are considered from the viewpoint of the optimization of the training sequence to be transmitted. Various types of periodic sequences having uniform discrete power spectra are studied. Some of them are new and may be generated with data sets commonly used in phase modulation systems. As a consequence of their power spectra being flat, these sequences ensure maximum protection against noise when initial equalizer settings are computed via channel estimates and noniterative techniques.

A recursive algorithm is implemented to give high computational speeds in the solution of a cutting-stock problem. Optimal edge-to-edge cutting is shown to be achieved more easily by recursive programming than by conventional methods. The technique features preliminary discretization, which lowers the memory requirements in the computational procedure. A comparison is made between this recursive algorithm and two iterative algorithms previously given by Gilmore-Gomory. The limitations of the algorithms are discussed and some numerical results given.

Artificial intelligence (AI) has made impressive progress over the past few years, including many applications in medical imaging. Numerous commercial solutions based on AI techniques are now available for sale, forcing radiology practices to learn how to properly assess these tools. While several guidelines describing good practices for conducting and reporting AI-based research in medicine and radiology have been published, fewer efforts have focused on recommendations addressing the key questions to consider when critically assessing AI solutions before purchase. Commercial AI solutions are typically complicated software products, for the evaluation of which many factors are to be considered. In this work, authors from academia and industry have joined efforts to propose a practical framework that will help stakeholders evaluate commercial AI solutions in radiology (the ECLAIR guidelines) and reach an informed decision. Topics to consider in the evaluation include the relevance of the solution from the point of view of each stakeholder, issues regarding performance and validation, usability and integration, regulatory and legal aspects, and financial and support services. KEY POINTS: • Numerous commercial solutions based on artificial intelligence techniques are now available for sale, and radiology practices have to learn how to properly assess these tools. • We propose a framework focusing on practical points to consider when assessing an AI solution in medical imaging, allowing all stakeholders to conduct relevant discussions with manufacturers and reach an informed decision as to whether to purchase an AI commercial solution for imaging applications. • Topics to consider in the evaluation include the relevance of the solution from the point of view of each stakeholder, issues regarding performance and validation, usability and integration, regulatory and legal aspects, and financial and support services.

This paper introduces new quadrature mirror filter (QMF) structures for the frequency domain analysis and synthesis of digital signals. The conventional QMF technique is first extended to cover complex quadrature mirror filters (CQMF) in which a digital signal is split into N adjacent complex subbands where the real and imaginary parts are subsampled by 1/2N with respect to the original signal. The computational complexity of QMF banks is then analyzed and a new scheme which reduces the computational complexity by about a factor of two over conventional QMF implementations is proposed. Finally, the filter design tradeoffs are discussed and the microprogramed implementation of QMF banks is evaluated.

This paper relates different kinds of language modeling methods that can be applied to the linguistic decoding part of a speech recognition system with a very large vocabulary. These models are studied experimentally on a pseudophonetic input arising from French stenotypy. We propose a model which combines the advantages of a statistical modeling with information theoretic tools, and those of a grammatical approach.

We have recently introduced new transforms, called polynomial transforms, which are defined in rings of polynomials and give efficient algorithms for the computation of multidimensional DFT's and convolutions. In this paper we present a method for computing one-dimensional convolutions by polynomial transforms. We show that this method is computationally efficient, even for large convolutions, and can be implemented with FFT-type algorithms, while avoiding the use of trigonometric functions and complex arithmetic. We then extend this technique to complex convolutions and to multidimensional convolutions.

The shift from a paper-based to an electronic-based society has dramatically reduced the cost of collecting, storing and processing individuals' personal information. As a result, it is becoming more common for businesses to "profile" individuals in order to present more personalized offers as part of their business strategy. While such profiles can be helpful and improve efficiency, they can also govern opaque decisions about an individual's access to services such as credit or an employment position. In many cases, profiling of personal data is done without the consent of the target individual.In the past decade, the European Union and its member states have implemented a legal framework to provide guidance on processing of personal data with the specific aim to restore the citizens' control over their data. To complement the legal framework, the prime (Privacy and Identity Management for Europe) project [14] has implemented a technical framework for processing personal data. prime's vision is to give individuals sovereignty over their personal data so that:

Abstract An algorithm is developed that finds the optimal orientation of a rigid molecular structure, represented by N reference sites, with respect to the same number of sites in an observed structure. The optimal orientation is found by minimizing the weighted sum of squared deviations of the rotated reference site positions from the observed site positions. The rotation is parametrized by a quaternion whose components, written as a column vector, are shown to be an eigenvector of a characteristic matrix which is defined in terms of the coordinate sets to be superimposed. The presented algorithm is particularly useful with respect to the calculation of orientational correlations of molecular structures.

This paper discusses the representational and algorithmic power of the conceptual graph model for natural language semantics and knowledge processing. Also described is a Prolog-like resolution method for conceptual graphs, which allows one to perform deduction on very large semantic domains. The interpreter that we have developed is similar to a Prolog interpreter in which the terms are any conceptual graphs and in which the unification algorithm is replaced by a specialized algorithm for conceptual graphs.

Abstract A dilute polyamic acid solution was spin‐deposited onto gold substrates and XPS analyses were performed after deposition. Evidence of the presence of organic films on the top surface of the substrates was found. A chemical characterization allowed us to identify these organic films as being polyamic acid after a 125°C curing and polyimide after a 240°C curing. From angle‐dependent data we have shown that the gold substrates were coated quasi‐uniformly by ultra‐thin films. This conclusion was confirmed by an Auger analysis. The computed thickness were found in a range of 1.3 to 2.9 nm depending on the sample preparation conditions. Adsorption was supposed to occur just after the deposition. Some possible mechanisms of the interfacial interaction between the polyamic acid and the surface were proposed. A thermally‐activated chain cleavage was supposed to occur at the interface or near the interface.

Experiments on the use of a probabilistic model to tag English text, that is, to assign to each word the correct tag (part of speech) in the context of the sentence, are presented. A simple triclass Markov model is used, and the best way to estimate the parameters of this model, depending on the kind and amount of training data that is provided, is found. Two approaches are compared: the use of text that has been tagged by hand and comparing relative frequency counts; and use text without tags and training the model as a hidden Markov process, according to a maximum likelihood principle. Experiments show that the best training is obtained by using as much tagged text as is available, a maximum likelihood training may improve the accuracy of the tagging.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

In 1991 a digital mobile radio system will be introduced in Europe. The speech codec to be used as the standard is presented. The coding scheme which has been selected by the CEPT Groupe-Speciale-Mobile (GSM) as a result of formal subjective listening tests, is based on the regular-pulse excitation LPC technique (RPE-LPC) combined with long-term prediction (LTP). The so-called RPE-LTP codec has a net bit rate of 13 kbit/s. The algorithm and the experimental implementations based on different VLSI signal processors are described and demonstrated.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Two-dimensional time-dependent buoyancy-induced flows above a horizontal line heat source inside rectangular vessels, with adiabatic sidewalls and top and bottom walls maintained at uniform temperature, are studied numerically. Transitions to unsteady flows are performed by direct simulations for various depths of immersion of the source in the central vertical plane of air-filled vessels. For a square vessel and a line source near the bottom wall, the numerical solutions exhibit a sequence of instabilities, called natural swaying motion of confined plumes, beginning with a periodic regime having a high fundamental frequency followed by a two-frequency locked regime. Then, broadband components appearing in the spectra indicate chaotic behaviour and a weakly turbulent motion arises via an intermittent route to chaos. For rectangular vessels of aspect ratio greater than 2 and depths of immersion greater than the width, the flow undergoes a pitchfork bifurcation. This symmetry breaking is driven by the destabilization of an upper unstable layer of stagnant fluid above the plume. Then a subcritical Hopf bifurcation occurs. On the other hand, if the depth of immersion is lower than the width of the vessel, a stable layer of fluid is at rest below the line source. Then penetrative convection sets the whole air-filled vessel in motion and an oscillatory motion of very low frequency arises through supercritical Hopf bifurcation followed by a two-frequency locked state.

Polynomial transforms, defined in rings of polynomials, have been introduced recently and have been shown to give efficient algorithms for the computation of two-dimensional convolutions. In this paper we present two methods for computing discrete Fourier transforms (DFT) by polynomial transforms. We show that these techniques are particularly well adapted to multidimensional DFT's as well as to some one-dimensional DFT's and yield algorithms that are, in many instances, more efficient than the fast Fourier transform (FFT) or the Winograd Fourier Transform (WFTA). We also describe new split nesting and split prime factor techniques for computing large DFT's from a small set of short DFT's with a minimum number of operations.

A universal quantum computer can simulate diverse quantum systems, with electronic structure for chemistry offering challenging problems for practical use cases around the hundred-qubit mark. Although current quantum processors have reached this size, deep circuits and a large number of measurements lead to prohibitive runtimes for quantum computers in isolation. Here, we demonstrate the use of classical distributed computing to offload all but an intrinsically quantum component of a workflow for electronic structure simulations. Using a Heron superconducting processor and the supercomputer Fugaku, we simulate the ground-state dissociation of N 2 and the ground state properties of [2Fe-2S] and [4Fe-4S] clusters, with circuits up to 77 qubits and 10,570 gates. The proposed algorithm processes quantum samples to produce upper bounds for the ground-state energy and sparse approximations to the ground-state wave functions. Our results suggest that, for current error rates, a quantum-centric supercomputing architecture can tackle challenging chemistry problems beyond sizes amenable to exact diagonalization.

Gate induced drain leakage (GIDL) is frequently described by band-to-band tunneling. This mechanism is insensitive to temperature and occurs only under strong electric fields. Under the condition of small electric fields, however, GIDL exhibits a strong dependence on temperature, which is due to trap-assisted generation of electron hole pairs. This generation mechanism is based on the Shockley-Read-Hall (SRH) equation involving field dependent emission probabilities due to Fowler-Nordheim (FN) and Poole-Frenkel effect. The proposed model of an acceptor-like interface trap is able to reproduce the experimental results. Temperature and voltage dependencies for a p-MOSFET are correctly calculated for one single fitting parameter, i.e., the interface trap density corresponding to N/sub t/=1/spl times/10/sup 14/(1/eV m/sup 2/).

The speech coding scheme which will be used as the standard for the European mobile radio system has been selected by the CEPT Groupe Special-Mobile (GSM) as a result of formal subjective listening tests. It is based on the regular-pulse excitation linear predictive coding technique (RPE-LPC) combined with long-term prediction (LTP). The solution is called the RPE-LTP codec. The codec algorithm and the error protection scheme are presented. The net bit rate is 13.0 kb/s, and the gross bit rate, including error protection, is 22.8 kb/s. The experimental implementation based on VLSI signal processors is described. The speech quality obtained with the technique considered is far superior to that obtainable with present-day analog mobile radio systems. A duplex speech codec including error protection can be implemented with two VLSI sign processors with external data memories of about 1 K*16 b.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

This paper presents the first 150 GHz amplifier in a digital 65 nm CMOS technology. Design techniques to preserve raw transistor gain near f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">max</sub> include layout optimization, dummy-prefilled microstrip lines (MSL) for design-rule compliance, and matching topologies which minimize passive element losses. To the authors' knowledge, the measured 8.3 dB gain, 6.3 dBm saturated output power (P <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">sat</sub> ), 1.5 dBm P <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1dB</sub> , 25.5 mW DC dissipation (PDC), and 27 GHz 3 dB BW are among the best in either CMOS or SiGe beyond 110GHz.