-
Hierarchical Light Transformer Ensembles for Multimodal Trajectory Forecasting
Authors:
Adrien Lafage,
Mathieu Barbier,
Gianni Franchi,
David Filliat
Abstract:
Accurate trajectory forecasting is crucial for the performance of various systems, such as advanced driver-assistance systems and self-driving vehicles. These forecasts allow us to anticipate events that lead to collisions and, therefore, to mitigate them. Deep Neural Networks have excelled in motion forecasting, but overconfidence and weak uncertainty quantification persist. Deep Ensembles addres…
▽ More
Accurate trajectory forecasting is crucial for the performance of various systems, such as advanced driver-assistance systems and self-driving vehicles. These forecasts allow us to anticipate events that lead to collisions and, therefore, to mitigate them. Deep Neural Networks have excelled in motion forecasting, but overconfidence and weak uncertainty quantification persist. Deep Ensembles address these concerns, yet applying them to multimodal distributions remains challenging. In this paper, we propose a novel approach named Hierarchical Light Transformer Ensembles (HLT-Ens) aimed at efficiently training an ensemble of Transformer architectures using a novel hierarchical loss function. HLT-Ens leverages grouped fully connected layers, inspired by grouped convolution techniques, to capture multimodal distributions effectively. We demonstrate that HLT-Ens achieves state-of-the-art performance levels through extensive experimentation, offering a promising avenue for improving trajectory forecasting techniques.
△ Less
Submitted 7 January, 2025; v1 submitted 26 March, 2024;
originally announced March 2024.
-
On the computation of the M{ö}bius transform
Authors:
Morgan Barbier,
Hayat Cheballah,
Jean-Marie Le Bars
Abstract:
The M{ö}bius transform is a crucial transformation into the Boolean world; it allows to change the Boolean representation between the True Table and Algebraic Normal Form. In this work, we introduce a new algebraic point of view of this transformation based on the polynomial form of Boolean functions. It appears that we can perform a new notion: the M{ö}bius computation variable by va…
▽ More
The M{ö}bius transform is a crucial transformation into the Boolean world; it allows to change the Boolean representation between the True Table and Algebraic Normal Form. In this work, we introduce a new algebraic point of view of this transformation based on the polynomial form of Boolean functions. It appears that we can perform a new notion: the M{ö}bius computation variable by variable and new computation properties. As a consequence, we propose new algorithms which can produce a huge speed up of the M{ö}bius computation for sub-families of Boolean function. Furthermore we compute directly the M{ö}bius transformation of some particular Boolean functions. Finally, we show that for some of them the Hamming weight is directly related to the algebraic degree of specific factors.
△ Less
Submitted 15 April, 2020;
originally announced April 2020.
-
An Elastic Energy Minimization Framework for Mean Contour Calculation
Authors:
Jozsef Molnar,
Michael Barbier,
Winnok H. De Vos,
Peter Horvath
Abstract:
In this paper we propose a contour mean calculation and interpolation method designed for averaging manual delineations of objects performed by experts and interpolate 3D layer stack images. The proposed method retains all visible information of the input contour set: the relative positions, orientations and size, but allows invisible quantities - parameterization and the centroid - to be changed.…
▽ More
In this paper we propose a contour mean calculation and interpolation method designed for averaging manual delineations of objects performed by experts and interpolate 3D layer stack images. The proposed method retains all visible information of the input contour set: the relative positions, orientations and size, but allows invisible quantities - parameterization and the centroid - to be changed. The chosen representation space - the position vector rescaled by square root velocity - is a real valued vector space on which the imposed L2 metric is used to define the distance function. With respect to this representation the re-parameterization group acts by isometries and the distance has well defined meaning: the sum of the central second moments of the coordinate functions. To identify the optimal re-parameterization system and proper centroid we use double energy minimization realized in a variational framework.
△ Less
Submitted 9 January, 2019;
originally announced January 2019.
-
Properties and constructions of coincident functions
Authors:
Morgan Barbier,
Hayat Cheballah,
Jean-Marie Le Bars
Abstract:
Extensive studies of Boolean functions are carried in many fields. The Mobius transform is often involved for these studies. In particular, it plays a central role in coincident functions, the class of Boolean functions invariant by this transformation. This class -- which has been recently introduced -- has interesting properties, in particular if we want to control both the Hamming weight and th…
▽ More
Extensive studies of Boolean functions are carried in many fields. The Mobius transform is often involved for these studies. In particular, it plays a central role in coincident functions, the class of Boolean functions invariant by this transformation. This class -- which has been recently introduced -- has interesting properties, in particular if we want to control both the Hamming weight and the degree. We propose an innovative way to handle the Mobius transform which allows the composition between several Boolean functions and the use of Shannon or Reed-Muller decompositions. Thus we benefit from a better knowledge of coin-cident functions and introduce new properties. We show experimentally that for many features, coincident functions look like any Boolean functions.
△ Less
Submitted 19 July, 2015;
originally announced July 2015.
-
Image Watermaking With Biometric Data For Copyright Protection
Authors:
Morgan Barbier,
Jean-Marie Le Bars,
Christophe Rosenberger
Abstract:
In this paper, we deal with the proof of ownership or legitimate usage of a digital content, such as an image, in order to tackle the illegitimate copy. The proposed scheme based on the combination of the watermark-ing and cancelable biometrics does not require a trusted third party, all the exchanges are between the provider and the customer. The use of cancelable biometrics permits to provide a…
▽ More
In this paper, we deal with the proof of ownership or legitimate usage of a digital content, such as an image, in order to tackle the illegitimate copy. The proposed scheme based on the combination of the watermark-ing and cancelable biometrics does not require a trusted third party, all the exchanges are between the provider and the customer. The use of cancelable biometrics permits to provide a privacy compliant proof of identity. We illustrate the robustness of this method against intentional and unintentional attacks of the watermarked content.
△ Less
Submitted 16 March, 2015;
originally announced March 2015.
-
Improving success probability and embedding efficiency in code based steganography
Authors:
Morgan Barbier,
Carlos Munuera
Abstract:
For stegoschemes arising from error correcting codes, embedding depends on a decoding map for the corresponding code. As decoding maps are usually not complete, embedding can fail. We propose a method to ensure or increase the probability of embedding success for these stegoschemes. This method is based on puncturing codes. We show how the use of punctured codes may also increase the embedding eff…
▽ More
For stegoschemes arising from error correcting codes, embedding depends on a decoding map for the corresponding code. As decoding maps are usually not complete, embedding can fail. We propose a method to ensure or increase the probability of embedding success for these stegoschemes. This method is based on puncturing codes. We show how the use of punctured codes may also increase the embedding efficiency of the obtained stegoschemes.
△ Less
Submitted 8 February, 2013;
originally announced February 2013.
-
On the decoding of quasi-BCH codes
Authors:
Morgan Barbier,
Clément Pernet,
Guillaume Quintin
Abstract:
In this paper we investigate the structure of quasi-BCH codes. In the first part of this paper we show that quasi-BCH codes can be derived from Reed-Solomon codes over square matrices extending the known relation about classical BCH and Reed-Solomon codes. This allows us to adapt the Welch-Berlekamp algorithm to quasi-BCH codes. In the second part of this paper we show that quasi-BCH codes can be…
▽ More
In this paper we investigate the structure of quasi-BCH codes. In the first part of this paper we show that quasi-BCH codes can be derived from Reed-Solomon codes over square matrices extending the known relation about classical BCH and Reed-Solomon codes. This allows us to adapt the Welch-Berlekamp algorithm to quasi-BCH codes. In the second part of this paper we show that quasi-BCH codes can be seen as subcodes of interleaved Reed-Solomon codes over finite fields. This provides another approach for decoding quasi-BCH codes.
△ Less
Submitted 22 December, 2012;
originally announced December 2012.
-
Re-encoding reformulation and application to Welch-Berlekamp algorithm
Authors:
Morgan Barbier
Abstract:
The main decoding algorithms for Reed-Solomon codes are based on a bivariate interpolation step, which is expensive in time complexity. Lot of interpolation methods were proposed in order to decrease the complexity of this procedure, but they stay still expensive. Then Koetter, Ma and Vardy proposed in 2010 a technique, called re-encoding, which allows to reduce the practical running time. However…
▽ More
The main decoding algorithms for Reed-Solomon codes are based on a bivariate interpolation step, which is expensive in time complexity. Lot of interpolation methods were proposed in order to decrease the complexity of this procedure, but they stay still expensive. Then Koetter, Ma and Vardy proposed in 2010 a technique, called re-encoding, which allows to reduce the practical running time. However, this trick is only devoted for the Koetter interpolation algorithm. We propose a reformulation of the re-encoding for any interpolation methods. The assumption for this reformulation permits only to apply it to the Welch-Berlekamp algorithm.
△ Less
Submitted 11 July, 2014; v1 submitted 21 December, 2012;
originally announced December 2012.
-
Ensuring message embedding in wet paper steganography
Authors:
Daniel Augot,
Morgan Barbier,
Caroline Fontaine
Abstract:
Syndrome coding has been proposed by Crandall in 1998 as a method to stealthily embed a message in a cover-medium through the use of bounded decoding. In 2005, Fridrich et al. introduced wet paper codes to improve the undetectability of the embedding by nabling the sender to lock some components of the cover-data, according to the nature of the cover-medium and the message. Unfortunately, almost a…
▽ More
Syndrome coding has been proposed by Crandall in 1998 as a method to stealthily embed a message in a cover-medium through the use of bounded decoding. In 2005, Fridrich et al. introduced wet paper codes to improve the undetectability of the embedding by nabling the sender to lock some components of the cover-data, according to the nature of the cover-medium and the message. Unfortunately, almost all existing methods solving the bounded decoding syndrome problem with or without locked components have a non-zero probability to fail. In this paper, we introduce a randomized syndrome coding, which guarantees the embedding success with probability one. We analyze the parameters of this new scheme in the case of perfect codes.
△ Less
Submitted 9 November, 2011;
originally announced November 2011.
-
On Quasi-Cyclic Codes as a Generalization of Cyclic Codes
Authors:
Morgan Barbier,
Christophe Chabot,
Guillaume Quintin
Abstract:
In this article we see quasi-cyclic codes as block cyclic codes. We generalize some properties of cyclic codes to quasi-cyclic ones such as generator polynomials and ideals. Indeed we show a one-to-one correspondence between l-quasi-cyclic codes of length m and ideals of M_l(Fq)[X]/(X^m-1). This permits to construct new classes of codes, namely quasi-BCH and quasi-evaluation codes. We study the pa…
▽ More
In this article we see quasi-cyclic codes as block cyclic codes. We generalize some properties of cyclic codes to quasi-cyclic ones such as generator polynomials and ideals. Indeed we show a one-to-one correspondence between l-quasi-cyclic codes of length m and ideals of M_l(Fq)[X]/(X^m-1). This permits to construct new classes of codes, namely quasi-BCH and quasi-evaluation codes. We study the parameters of such codes and propose a decoding algorithm up to half the designed minimum distance. We even found one new quasi-cyclic code with better parameters than known [189, 11, 125]_F4 and 48 derivated codes beating the known bounds as well.
△ Less
Submitted 25 May, 2012; v1 submitted 18 August, 2011;
originally announced August 2011.
-
Wet paper codes and the dual distance in steganography
Authors:
Carlos Munuera,
Morgan Barbier
Abstract:
In 1998 Crandall introduced a method based on coding theory to secretly embed a message in a digital support such as an image. Later Fridrich et al. improved this method to minimize the distortion introduced by the embedding; a process called wet paper. However, as previously emphasized in the literature, this method can fail during the embedding step. Here we find sufficient and necessary conditi…
▽ More
In 1998 Crandall introduced a method based on coding theory to secretly embed a message in a digital support such as an image. Later Fridrich et al. improved this method to minimize the distortion introduced by the embedding; a process called wet paper. However, as previously emphasized in the literature, this method can fail during the embedding step. Here we find sufficient and necessary conditions to guarantee a successful embedding by studying the dual distance of a linear code. Since these results are essentially of combinatorial nature, they can be generalized to systematic codes, a large family containing all linear codes. We also compute the exact number of solutions and point out the relationship between wet paper codes and orthogonal arrays.
△ Less
Submitted 11 April, 2011;
originally announced April 2011.
-
Key Reduction of McEliece's Cryptosystem Using List Decoding
Authors:
Morgan Barbier,
Barreto S. L. M. Paulo
Abstract:
Different variants of the code-based McEliece cryptosystem were pro- posed to reduce the size of the public key. All these variants use very structured codes, which open the door to new attacks exploiting the underlying structure. In this paper, we show that the dyadic variant can be designed to resist all known attacks. In light of a new study on list decoding algorithms for binary Goppa codes, w…
▽ More
Different variants of the code-based McEliece cryptosystem were pro- posed to reduce the size of the public key. All these variants use very structured codes, which open the door to new attacks exploiting the underlying structure. In this paper, we show that the dyadic variant can be designed to resist all known attacks. In light of a new study on list decoding algorithms for binary Goppa codes, we explain how to increase the security level for given public keysizes. Using the state-of-the-art list decoding algorithm instead of unique decoding, we exhibit a keysize gain of about 4% for the standard McEliece cryptosystem and up to 21% for the adjusted dyadic variant.
△ Less
Submitted 15 November, 2011; v1 submitted 13 February, 2011;
originally announced February 2011.
-
List-decoding of binary Goppa codes up to the binary Johnson bound
Authors:
Daniel Augot,
Morgan Barbier,
Alain Couvreur
Abstract:
We study the list-decoding problem of alternant codes, with the notable case of classical Goppa codes. The major consideration here is to take into account the size of the alphabet, which shows great influence on the list-decoding radius. This amounts to compare the \emph{generic} Johnson bound to the \emph{$q$-ary} Johnson bound. This difference is important when $q$ is very small. Essentially, t…
▽ More
We study the list-decoding problem of alternant codes, with the notable case of classical Goppa codes. The major consideration here is to take into account the size of the alphabet, which shows great influence on the list-decoding radius. This amounts to compare the \emph{generic} Johnson bound to the \emph{$q$-ary} Johnson bound. This difference is important when $q$ is very small. Essentially, the most favourable case is $q=2$, for which the decoding radius is greatly improved, notably when the relative minimum distance gets close to 1/2. Even though the announced result, which is the list-decoding radius of binary Goppa codes, is new, it can be rather easily made up from previous sources (V. Guruswami, R. M. Roth and I. Tal, R .M. Roth), which may be a little bit unknown, and in which the case of binary Goppa codes has apparently not been thought at. Only D. J. Bernstein treats the case of binary Goppa codes in a preprint. References are given in the introduction. We propose an autonomous treatment and also a complexity analysis of the studied algorithm, which is quadratic in the blocklength $n$, when decoding at some distance of the relative maximum decoding radius, and in $O(n^7)$ when reaching the maximum radius.
△ Less
Submitted 15 December, 2010;
originally announced December 2010.
-
New Set of Codes for the Maximum-Likelihood Decoding Problem
Authors:
Morgan Barbier
Abstract:
The maximum-likelihood decoding problem is known to be NP-hard for general linear and Reed-Solomon codes. In this paper, we introduce the notion of A-covered codes, that is, codes that can be decoded through a polynomial time algorithm A whose decoding bound is beyond the covering radius. For these codes, we show that the maximum-likelihood decoding problem is reachable in polynomial time in the c…
▽ More
The maximum-likelihood decoding problem is known to be NP-hard for general linear and Reed-Solomon codes. In this paper, we introduce the notion of A-covered codes, that is, codes that can be decoded through a polynomial time algorithm A whose decoding bound is beyond the covering radius. For these codes, we show that the maximum-likelihood decoding problem is reachable in polynomial time in the code parameters. Focusing on bi- nary BCH codes, we were able to find several examples of A-covered codes, including two codes for which the maximum-likelihood decoding problem can be solved in quasi-quadratic time.
△ Less
Submitted 12 November, 2010;
originally announced November 2010.
