Showing 51–100 of 117 results for author: Kanter, I

Chaos Pass Filter: Linear Response of Synchronized Chaotic Systems

Authors: Wolfgang Kinzel, Johannes Kestler, Ido Kanter

Abstract: The linear response of synchronized chaotic units with delayed couplings and feedback to small external perturbations is investigated in the context of communication with chaos synchronization. For iterated chaotic maps, the distribution of distances is calculated numerically and, for some special cases, analytically as well. Depending on model parameters, this distribution has power law tails l… ▽ More The linear response of synchronized chaotic units with delayed couplings and feedback to small external perturbations is investigated in the context of communication with chaos synchronization. For iterated chaotic maps, the distribution of distances is calculated numerically and, for some special cases, analytically as well. Depending on model parameters, this distribution has power law tails leading to diverging moments of distances in the region of synchronization. The corresponding linear equations have multiplicative and additive noise due to perturbations and chaos. The response to small harmonic perturbations shows resonances related to coupling and feedback delay times. For perturbation from a binary message the bit error rate is calculated. The bit error rate is not related to the transverse Lyapunov exponents, and it can be reduced when additional noise is added to the transmitted signal. For some special cases, the bit error rate as a function of coupling strength has the structure of a devil's staircase, related to an iterated function system. Finally, the security of communication is discussed by comparing uni- and bi-directional couplings. △ Less

Submitted 4 February, 2013; v1 submitted 26 June, 2008; originally announced June 2008.

Comments: Outdated by https://arxiv.boxedpaper.com/abs/1301.6929

Shannon Meets Carnot: Generalized Second Thermodynamic Law

Authors: Ori Shental, Ido Kanter

Abstract: The classical thermodynamic laws fail to capture the behavior of systems with energy Hamiltonian which is an explicit function of the temperature. Such Hamiltonian arises, for example, in modeling information processing systems, like communication channels, as thermal systems. Here we generalize the second thermodynamic law to encompass systems with temperature-dependent energy levels,… ▽ More The classical thermodynamic laws fail to capture the behavior of systems with energy Hamiltonian which is an explicit function of the temperature. Such Hamiltonian arises, for example, in modeling information processing systems, like communication channels, as thermal systems. Here we generalize the second thermodynamic law to encompass systems with temperature-dependent energy levels, $dQ=TdS+<d\mathcal{E}/dT>dT$, where $<\cdot>$ denotes averaging over the Boltzmann distribution and reveal a new definition to the basic notion of temperature. This generalization enables to express, for instance, the mutual information of the Gaussian channel as a consequence of the fundamental laws of nature - the laws of thermodynamics. △ Less

Submitted 23 June, 2008; originally announced June 2008.

Shannon Meets Carnot: Mutual Information Via Thermodynamics

Authors: Ori Shental, Ido Kanter

Abstract: In this contribution, the Gaussian channel is represented as an equivalent thermal system allowing to express its input-output mutual information in terms of thermodynamic quantities. This thermodynamic description of the mutual information is based upon a generalization of the $2^{nd}$ thermodynamic law and provides an alternative proof to the Guo-Shamai-Verdú theorem, giving an intriguing conn… ▽ More In this contribution, the Gaussian channel is represented as an equivalent thermal system allowing to express its input-output mutual information in terms of thermodynamic quantities. This thermodynamic description of the mutual information is based upon a generalization of the $2^{nd}$ thermodynamic law and provides an alternative proof to the Guo-Shamai-Verdú theorem, giving an intriguing connection between this remarkable theorem and the most fundamental laws of nature - the laws of thermodynamics. △ Less

Submitted 19 June, 2008; originally announced June 2008.

Public Channel Cryptography: Chaos Synchronization and Hilbert's Tenth Problem

Authors: Ido Kanter, Evi Kopelowitz, Wolfgang Kinzel

Abstract: The synchronization process of two mutually delayed coupled deterministic chaotic maps is demonstrated both analytically and numerically. The synchronization is preserved when the mutually transmitted signal is concealed by two commutative private filters that are placed on each end of the communication channel. We demonstrate that when the transmitted signal is a convolution of the truncated ti… ▽ More The synchronization process of two mutually delayed coupled deterministic chaotic maps is demonstrated both analytically and numerically. The synchronization is preserved when the mutually transmitted signal is concealed by two commutative private filters that are placed on each end of the communication channel. We demonstrate that when the transmitted signal is a convolution of the truncated time delayed output signals or some powers of the delayed output signals synchronization is still maintained. The task of a passive attacker is mapped onto Hilbert's tenth problem, solving a set of nonlinear Diophantine equations, which was proven to be in the class of NP-Complete problems. This bridge between two different disciplines, synchronization in nonlinear dynamical processes and the realm of the NPC problems, opens a horizon for a new type of secure public-channel protocols. △ Less

Submitted 5 June, 2008; originally announced June 2008.

Patterns of Chaos Synchronization

Authors: Johannes Kestler, Evi Kopelowitz, Ido Kanter, Wolfgang Kinzel

Abstract: Small networks of chaotic units which are coupled by their time-delayed variables, are investigated. In spite of the time delay, the units can synchronize isochronally, i.e. without time shift. Moreover, networks can not only synchronize completely, but can also split into different synchronized sublattices. These synchronization patterns are stable attractors of the network dynamics. Different… ▽ More Small networks of chaotic units which are coupled by their time-delayed variables, are investigated. In spite of the time delay, the units can synchronize isochronally, i.e. without time shift. Moreover, networks can not only synchronize completely, but can also split into different synchronized sublattices. These synchronization patterns are stable attractors of the network dynamics. Different networks with their associated behaviors and synchronization patterns are presented. In particular, we investigate sublattice synchronization, symmetry breaking, spreading chaotic motifs, synchronization by restoring symmetry and cooperative pairwise synchronization of a bipartite tree. △ Less

Submitted 3 January, 2008; originally announced January 2008.

Chaos Synchronization with Dynamic Filters: Two Way is Better Than One Way

Authors: Ido Kanter, Evi Kopelowitz, Johannes Kestler, Wolfgang Kinzel

Abstract: Two chaotic systems which interact by mutually exchanging a signal built from their delayed internal variables, can synchronize. A third unit may be able to record and to manipulate the exchanged signal. Can the third unit synchronize to the common chaotic trajectory, as well? If all parameters of the system are public, a proof is given that the recording system can synchronize as well. However,… ▽ More Two chaotic systems which interact by mutually exchanging a signal built from their delayed internal variables, can synchronize. A third unit may be able to record and to manipulate the exchanged signal. Can the third unit synchronize to the common chaotic trajectory, as well? If all parameters of the system are public, a proof is given that the recording system can synchronize as well. However, if the two interacting systems use private commutative filters to generate the exchanged signal, a driven system cannot synchronize. It is shown that with dynamic private filters the chaotic trajectory even cannot be calculated. Hence two way (interaction) is more than one way (drive). The implication of this general result to secret communication with chaos synchronization is discussed. △ Less

Submitted 17 December, 2007; originally announced December 2007.

Spiking Optical Patterns and Synchronization

Authors: Michael Rosenbluh, Yaara Aviad, Elad Cohen, Lev Khaykovich, Wolfgang Kinzel, Evi Kopelowitz, Pinhas Yoskovits, Ido Kanter

Abstract: We analyze the time resolved spike statistics of a solitary and two mutually interacting chaotic semiconductor lasers whose chaos is characterized by apparently random, short intensity spikes. Repulsion between two successive spikes is observed, resulting in a refractory period which is largest at laser threshold. For time intervals between spikes greater than the refractory period, the distribu… ▽ More We analyze the time resolved spike statistics of a solitary and two mutually interacting chaotic semiconductor lasers whose chaos is characterized by apparently random, short intensity spikes. Repulsion between two successive spikes is observed, resulting in a refractory period which is largest at laser threshold. For time intervals between spikes greater than the refractory period, the distribution of the intervals follows a Poisson distribution. The spiking pattern is highly periodic over time windows corresponding to the optical length of the external cavity, with a slow change of the spiking pattern as time increases. When zero-lag synchronization between the two lasers is established, the statistics of the nearly perfectly matched spikes are not altered. The similarity of these features to those found in complex interacting neural networks, suggests the use of laser systems as simpler physical models for neural networks. △ Less

Submitted 11 June, 2007; originally announced June 2007.

Sublattice synchronization of chaotic networks with delayed couplings

Authors: Johannes Kestler, Wolfgang Kinzel, Ido Kanter

Abstract: Synchronization of chaotic units coupled by their time delayed variables are investigated analytically. A new type of cooperative behavior is found: sublattice synchronization. Although the units of one sublattice are not directly coupled to each other, they completely synchronize without time delay. The chaotic trajectories of different sublattices are only weakly correlated but not related by… ▽ More Synchronization of chaotic units coupled by their time delayed variables are investigated analytically. A new type of cooperative behavior is found: sublattice synchronization. Although the units of one sublattice are not directly coupled to each other, they completely synchronize without time delay. The chaotic trajectories of different sublattices are only weakly correlated but not related by generalized synchronization. Nevertheless, the trajectory of one sublattice is predictable from the complete trajectory of the other one. The spectra of Lyapunov exponents are calculated analytically in the limit of infinite delay times, and phase diagrams are derived for different topologies. △ Less

Submitted 7 March, 2007; originally announced March 2007.

Dynamics of neural cryptography

Authors: Andreas Ruttor, Wolfgang Kinzel, Ido Kanter

Abstract: Synchronization of neural networks has been used for novel public channel protocols in cryptography. In the case of tree parity machines the dynamics of both bidirectional synchronization and unidirectional learning is driven by attractive and repulsive stochastic forces. Thus it can be described well by a random walk model for the overlap between participating neural networks. For that purpose… ▽ More Synchronization of neural networks has been used for novel public channel protocols in cryptography. In the case of tree parity machines the dynamics of both bidirectional synchronization and unidirectional learning is driven by attractive and repulsive stochastic forces. Thus it can be described well by a random walk model for the overlap between participating neural networks. For that purpose transition probabilities and scaling laws for the step sizes are derived analytically. Both these calculations as well as numerical simulations show that bidirectional interaction leads to full synchronization on average. In contrast, successful learning is only possible by means of fluctuations. Consequently, synchronization is much faster than learning, which is essential for the security of the neural key-exchange protocol. However, this qualitative difference between bidirectional and unidirectional interaction vanishes if tree parity machines with more than three hidden units are used, so that those neural networks are not suitable for neural cryptography. In addition, the effective number of keys which can be generated by the neural key-exchange protocol is calculated using the entropy of the weight distribution. As this quantity increases exponentially with the system size, brute-force attacks on neural cryptography can easily be made unfeasible. △ Less

Submitted 1 November, 2007; v1 submitted 21 December, 2006; originally announced December 2006.

Comments: 9 pages, 15 figures; typos corrected

Journal ref: Phys. Rev. E 75, 056104 (2007)

Synchronization of mutually coupled chaotic lasers in the presence of a shutter

Authors: Ido Kanter, Noam Gross, Einat Klein, Evi Kopelowitz, Pinhas Yoskovits, Lev Khaykovich, Wolfgang Kinzel, Michael Rosenbluh

Abstract: Two mutually coupled chaotic diode lasers exhibit stable isochronal synchronization in the presence of self feedback. When the mutual communication between the lasers is discontinued by a shutter and the two uncoupled lasers are subject to self-feedback only, the desynchronization time is found to scale as $A_dτ$ where $A_d>1$ and $τ$ corresponds to the optical distance between the lasers. Prior… ▽ More Two mutually coupled chaotic diode lasers exhibit stable isochronal synchronization in the presence of self feedback. When the mutual communication between the lasers is discontinued by a shutter and the two uncoupled lasers are subject to self-feedback only, the desynchronization time is found to scale as $A_dτ$ where $A_d>1$ and $τ$ corresponds to the optical distance between the lasers. Prior to synchronization, when the two lasers are uncorrelated and the shutter between them is opened, the synchronization time is found to be much shorter, though still proportional to $τ$. As a consequence of these results, the synchronization is not significantly altered if the shutter is opend/closed faster than the desynchronization time. Experiments in which the coupling between two chaotic-synchronized diode lasers is modulated with an electro-optic shutter are found to be consistent with the results of numerical simulations. △ Less

Submitted 15 October, 2006; originally announced October 2006.

Parallel vs. Sequential Belief Propagation Decoding of LDPC Codes over GF(q) and Markov Sources

Authors: Nadav Yacov, Hadar Efraim, Haggai Kfir, Ido Kanter, Ori Shental

Abstract: A sequential updating scheme (SUS) for belief propagation (BP) decoding of LDPC codes over Galois fields, $GF(q)$, and correlated Markov sources is proposed, and compared with the standard parallel updating scheme (PUS). A thorough experimental study of various transmission settings indicates that the convergence rate, in iterations, of the BP algorithm (and subsequently its complexity) for the… ▽ More A sequential updating scheme (SUS) for belief propagation (BP) decoding of LDPC codes over Galois fields, $GF(q)$, and correlated Markov sources is proposed, and compared with the standard parallel updating scheme (PUS). A thorough experimental study of various transmission settings indicates that the convergence rate, in iterations, of the BP algorithm (and subsequently its complexity) for the SUS is about one half of that for the PUS, independent of the finite field size $q$. Moreover, this 1/2 factor appears regardless of the correlations of the source and the channel's noise model, while the error correction performance remains unchanged. These results may imply on the 'universality' of the one half convergence speed-up of SUS decoding. △ Less

Submitted 16 May, 2006; originally announced May 2006.

Synchronization of Mutually Versus Unidirectionally Coupled Chaotic Semiconductor Lasers

Authors: Noam Gross, Wolfgang Kinzel, Ido Kanter, Michael Rosenbluh, Lev Khaykovich

Abstract: Synchronization dynamics of mutually coupled chaotic semiconductor lasers are investigated experimentally and compared to identical synchronization of unidirectionally coupled lasers. Mutual coupling shows high quality synchronization in a broad range of self-feedback and coupling strengths. It is found to be tolerant to significant parameter mismatch which for unidirectional coupling would resu… ▽ More Synchronization dynamics of mutually coupled chaotic semiconductor lasers are investigated experimentally and compared to identical synchronization of unidirectionally coupled lasers. Mutual coupling shows high quality synchronization in a broad range of self-feedback and coupling strengths. It is found to be tolerant to significant parameter mismatch which for unidirectional coupling would result in loss of synchronization. The advantages of mutual coupling are emphasized in light of its potential use in chaos communications. △ Less

Submitted 26 April, 2006; originally announced April 2006.

Comments: 5 pages, 6 figures

Optimum Asymptotic Multiuser Efficiency of Pseudo-Orthogonal Randomly Spread CDMA

Authors: Ori Shental, Ido Kanter

Abstract: A $K$-user pseudo-orthogonal (PO) randomly spread CDMA system, equivalent to transmission over a subset of $K'\leq K$ single-user Gaussian channels, is introduced. The high signal-to-noise ratio performance of the PO-CDMA is analyzed by rigorously deriving its asymptotic multiuser efficiency (AME) in the large system limit. Interestingly, the $K'$-optimized PO-CDMA transceiver scheme yields an A… ▽ More A $K$-user pseudo-orthogonal (PO) randomly spread CDMA system, equivalent to transmission over a subset of $K'\leq K$ single-user Gaussian channels, is introduced. The high signal-to-noise ratio performance of the PO-CDMA is analyzed by rigorously deriving its asymptotic multiuser efficiency (AME) in the large system limit. Interestingly, the $K'$-optimized PO-CDMA transceiver scheme yields an AME which is practically equal to 1 for system loads smaller than 0.1 and lower bounded by 1/4 for increasing loads. As opposed to the vanishing efficiency of linear multiuser detectors, the derived efficiency is comparable to the ultimate CDMA efficiency achieved for the intractable optimal multiuser detector. △ Less

Submitted 21 April, 2006; originally announced April 2006.

Comments: WIC 27th Symposium on Information Theory in the Benelux, 2006

Information and multiaccess interference in a complexity-constrained vector channel

Authors: Rodrigo de Miguel, Ori Shental, Ralf R. Muller, Ido Kanter

Abstract: Rodrigo de Miguel et al 2007 J. Phys. A: Math. Theor. 40 5241-5260: A noisy vector channel operating under a strict complexity constraint at the receiver is introduced. According to this constraint, detected bits, obtained by performing hard decisions directly on the channel's matched filter output, must be the same as the transmitted binary inputs. An asymptotic analysis is carried out using ma… ▽ More Rodrigo de Miguel et al 2007 J. Phys. A: Math. Theor. 40 5241-5260: A noisy vector channel operating under a strict complexity constraint at the receiver is introduced. According to this constraint, detected bits, obtained by performing hard decisions directly on the channel's matched filter output, must be the same as the transmitted binary inputs. An asymptotic analysis is carried out using mathematical tools imported from the study of neural networks, and it is shown that, under a bounded noise assumption, such complexity-constrained channel exhibits a non-trivial Shannon-theoretic capacity. It is found that performance relies on rigorous interference-based multiuser cooperation at the transmitter and that this cooperation is best served when all transmitters use the same amplitude. △ Less

Submitted 19 May, 2009; v1 submitted 19 April, 2006; originally announced April 2006.

Comments: The author has withdrawn this submission [arXiv admin]

Public-channel cryptography based on mutual chaos pass filters

Authors: Einat Klein, Noam Gross, Evi Kopelowitz, Michael Rosenbluh, Lev Khaykovich, Wolfgang Kinzel, Ido Kanter

Abstract: We study the mutual coupling of chaotic lasers and observe both experimentally and in numeric simulations, that there exists a regime of parameters for which two mutually coupled chaotic lasers establish isochronal synchronization, while a third laser coupled unidirectionally to one of the pair, does not synchronize. We then propose a cryptographic scheme, based on the advantage of mutual-coupli… ▽ More We study the mutual coupling of chaotic lasers and observe both experimentally and in numeric simulations, that there exists a regime of parameters for which two mutually coupled chaotic lasers establish isochronal synchronization, while a third laser coupled unidirectionally to one of the pair, does not synchronize. We then propose a cryptographic scheme, based on the advantage of mutual-coupling over unidirectional coupling, where all the parameters of the system are public knowledge. We numerically demonstrate that in such a scheme the two communicating lasers can add a message signal (compressed binary message) to the transmitted coupling signal, and recover the message in both directions with high fidelity by using a mutual chaos pass filter procedure. An attacker however, fails to recover an errorless message even if he amplifies the coupling signal. △ Less

Submitted 25 April, 2006; originally announced April 2006.

Genetic attack on neural cryptography

Authors: Andreas Ruttor, Wolfgang Kinzel, Rivka Naeh, Ido Kanter

Abstract: Different scaling properties for the complexity of bidirectional synchronization and unidirectional learning are essential for the security of neural cryptography. Incrementing the synaptic depth of the networks increases the synchronization time only polynomially, but the success of the geometric attack is reduced exponentially and it clearly fails in the limit of infinite synaptic depth. This… ▽ More Different scaling properties for the complexity of bidirectional synchronization and unidirectional learning are essential for the security of neural cryptography. Incrementing the synaptic depth of the networks increases the synchronization time only polynomially, but the success of the geometric attack is reduced exponentially and it clearly fails in the limit of infinite synaptic depth. This method is improved by adding a genetic algorithm, which selects the fittest neural networks. The probability of a successful genetic attack is calculated for different model parameters using numerical simulations. The results show that scaling laws observed in the case of other attacks hold for the improved algorithm, too. The number of networks needed for an effective attack grows exponentially with increasing synaptic depth. In addition, finite-size effects caused by Hebbian and anti-Hebbian learning are analyzed. These learning rules converge to the random walk rule if the synaptic depth is small compared to the square root of the system size. △ Less

Submitted 1 June, 2006; v1 submitted 1 December, 2005; originally announced December 2005.

Comments: 8 pages, 12 figures; section 5 amended, typos corrected

Journal ref: Phys. Rev. E 73, 036121 (2006)

Stable isochronal synchronization of mutually coupled chaotic lasers

Authors: Einat Klein, Noam Gross, Michael Rosenbluh, Wolfgang Kinzel, Lev Khaykovich, Ido Kanter

Abstract: The dynamics of two mutually coupled chaotic diode lasers are investigated experimentally and numerically. By adding self feedback to each laser, stable isochronal synchronization is established. This stability, which can be achieved for symmetric operation, is essential for constructing an optical public-channel cryptographic system. The experimental results on diode lasers are well described b… ▽ More The dynamics of two mutually coupled chaotic diode lasers are investigated experimentally and numerically. By adding self feedback to each laser, stable isochronal synchronization is established. This stability, which can be achieved for symmetric operation, is essential for constructing an optical public-channel cryptographic system. The experimental results on diode lasers are well described by rate equations of coupled single mode lasers. △ Less

Submitted 28 November, 2005; originally announced November 2005.

From finite-system entropy to entropy rate for a Hidden Markov Process

Authors: Or Zuk, Eytan Domany, Ido Kanter, Michael Aizenman

Abstract: A recent result presented the expansion for the entropy rate of a Hidden Markov Process (HMP) as a power series in the noise variable $\eps$. The coefficients of the expansion around the noiseless ($\eps = 0$) limit were calculated up to 11th order, using a conjecture that relates the entropy rate of a HMP to the entropy of a process of finite length (which is calculated analytically). In this c… ▽ More A recent result presented the expansion for the entropy rate of a Hidden Markov Process (HMP) as a power series in the noise variable $\eps$. The coefficients of the expansion around the noiseless ($\eps = 0$) limit were calculated up to 11th order, using a conjecture that relates the entropy rate of a HMP to the entropy of a process of finite length (which is calculated analytically). In this communication we generalize and prove the validity of the conjecture, and discuss the theoretical and practical consequences of our new theorem. △ Less

Submitted 6 October, 2005; originally announced October 2005.

Journal ref: IEEE Signal Processing Letters 13,517 (2006).

Taylor series expansions for the entropy rate of Hidden Markov Processes

Authors: Or Zuk, Eytan Domany, Ido Kanter, Michael Aizenman

Abstract: Finding the entropy rate of Hidden Markov Processes is an active research topic, of both theoretical and practical importance. A recently used approach is studying the asymptotic behavior of the entropy rate in various regimes. In this paper we generalize and prove a previous conjecture relating the entropy rate to entropies of finite systems. Building on our new theorems, we establish series ex… ▽ More Finding the entropy rate of Hidden Markov Processes is an active research topic, of both theoretical and practical importance. A recently used approach is studying the asymptotic behavior of the entropy rate in various regimes. In this paper we generalize and prove a previous conjecture relating the entropy rate to entropies of finite systems. Building on our new theorems, we establish series expansions for the entropy rate in two different regimes. We also study the radius of convergence of the two series expansions. △ Less

Submitted 3 October, 2005; originally announced October 2005.

Journal ref: Proceedings 2006 IEEE International Conference on Communications (ICC 2006).

On the Achievable Information Rates of CDMA Downlink with Trivial Receivers

Authors: Ori Shental, Ido Kanter, Anthony J. Weiss

Abstract: A noisy CDMA downlink channel operating under a strict complexity constraint on the receiver is introduced. According to this constraint, detected bits, obtained by performing hard decisions directly on the channel's matched filter output, must be the same as the transmitted binary inputs. This channel setting, allowing the use of the simplest receiver scheme, seems to be worthless, making relia… ▽ More A noisy CDMA downlink channel operating under a strict complexity constraint on the receiver is introduced. According to this constraint, detected bits, obtained by performing hard decisions directly on the channel's matched filter output, must be the same as the transmitted binary inputs. This channel setting, allowing the use of the simplest receiver scheme, seems to be worthless, making reliable communication at any rate impossible. However, recently this communication paradigm was shown to yield valuable information rates in the case of a noiseless channel. This finding calls for the investigation of this attractive complexity-constrained transmission scheme for the more practical noisy channel case. By adopting the statistical mechanics notion of metastable states of the renowned Hopfield model, it is proved that under a bounded noise assumption such complexity-constrained CDMA channel gives rise to a non-trivial Shannon-theoretic capacity, rigorously analyzed and corroborated using finite-size channel simulations. For unbounded noise the channel's outage capacity is addressed and specifically described for the popular additive white Gaussian noise. △ Less

Submitted 16 September, 2005; originally announced September 2005.

Comments: Submitted to IEEE Transactions on Information Theory

Capacity of Complexity-Constrained Noise-Free CDMA

Authors: Ori Shental, Ido Kanter, Anthony J. Weiss

Abstract: An interference-limited noise-free CDMA downlink channel operating under a complexity constraint on the receiver is introduced. According to this paradigm, detected bits, obtained by performing hard decisions directly on the channel's matched filter output, must be the same as the transmitted binary inputs. This channel setting, allowing the use of the simplest receiver scheme, seems to be worth… ▽ More An interference-limited noise-free CDMA downlink channel operating under a complexity constraint on the receiver is introduced. According to this paradigm, detected bits, obtained by performing hard decisions directly on the channel's matched filter output, must be the same as the transmitted binary inputs. This channel setting, allowing the use of the simplest receiver scheme, seems to be worthless, making reliable communication at any rate impossible. We prove, by adopting statistical mechanics notion, that in the large-system limit such a complexity-constrained CDMA channel gives rise to a non-trivial Shannon-theoretic capacity, rigorously analyzed and corroborated using finite-size channel simulations. △ Less

Submitted 16 September, 2005; originally announced September 2005.

Comments: To appear in IEEE Communications Letters

The Entropy of a Binary Hidden Markov Process

Authors: O. Zuk, I. Kanter, E. Domany

Abstract: The entropy of a binary symmetric Hidden Markov Process is calculated as an expansion in the noise parameter epsilon. We map the problem onto a one-dimensional Ising model in a large field of random signs and calculate the expansion coefficients up to second order in epsilon. Using a conjecture we extend the calculation to 11th order and discuss the convergence of the resulting series. The entropy of a binary symmetric Hidden Markov Process is calculated as an expansion in the noise parameter epsilon. We map the problem onto a one-dimensional Ising model in a large field of random signs and calculate the expansion coefficients up to second order in epsilon. Using a conjecture we extend the calculation to 11th order and discuss the convergence of the resulting series. △ Less

Submitted 23 July, 2005; originally announced July 2005.

A framework for public-channel cryptography using chaotic lasers

Authors: Noam Gross, Einat Klein, Michael Rosenbluh, Wolfgang Kinzel, Lev Khaykovich, Ido Kanter

Abstract: Two mutually coupled chaotic diode lasers with individual external feedback, are shown to establish chaos synchronization in the low-frequency fluctuations regime. A third laser with identical external feedback but coupled unidirectionally to one of the pair does not synchronize. Both experiments and simulations reveal the existence of a window of parameters for which synchronization by mutual c… ▽ More Two mutually coupled chaotic diode lasers with individual external feedback, are shown to establish chaos synchronization in the low-frequency fluctuations regime. A third laser with identical external feedback but coupled unidirectionally to one of the pair does not synchronize. Both experiments and simulations reveal the existence of a window of parameters for which synchronization by mutual coupling is possible but synchronization by unidirectional coupling is not. This parameter space forms the basis of a proposed public-channel cryptographic scheme and is robust to various possible attacks. △ Less

Submitted 24 July, 2005; originally announced July 2005.

Public-channel cryptography using chaos synchronization

Authors: Einat Klein, Rachel Mislovaty, Ido Kanter, Wolfgang Kinzel

Abstract: We present a key-exchange protocol that comprises two parties with chaotic dynamics that are mutually coupled and undergo a synchronization process, at the end of which they can use their identical dynamical state as an encryption key. The transferred coupling-signals are based non-linearly on time-delayed states of the parties, and therefore they conceal the parties' current state and can be tr… ▽ More We present a key-exchange protocol that comprises two parties with chaotic dynamics that are mutually coupled and undergo a synchronization process, at the end of which they can use their identical dynamical state as an encryption key. The transferred coupling-signals are based non-linearly on time-delayed states of the parties, and therefore they conceal the parties' current state and can be transferred over a public channel. Synchronization time is linear in the number of synchronized digits alpha, while the probability for an attacker to synchronize with the parties drops exponentially with alpha. To achieve security with finite alpha we use a network. △ Less

Submitted 29 March, 2005; v1 submitted 22 December, 2004; originally announced December 2004.

Comments: 4 figures

An Efficient Joint Source-Channel Decoder with Dynamical Block Priors

Authors: Ido Kanter, Haggai Kfir, Shahar Keren

Abstract: An efficient joint source-channel (s/c) decoder based on the side information of the source and on the MN-Gallager algorithm over Galois fields is presented. The dynamical block priors (DBP) are derived either from a statistical mechanical approach via calculation of the entropy for the correlated sequences, or from the Markovian transition matrix. The Markovian joint s/c decoder has many advant… ▽ More An efficient joint source-channel (s/c) decoder based on the side information of the source and on the MN-Gallager algorithm over Galois fields is presented. The dynamical block priors (DBP) are derived either from a statistical mechanical approach via calculation of the entropy for the correlated sequences, or from the Markovian transition matrix. The Markovian joint s/c decoder has many advantages over the statistical mechanical approach. In particular, there is no need for the construction and the diagonalization of a qXq matrix and for a solution to saddle point equations in q dimensions. Using parametric estimation, an efficient joint s/c decoder with the lack of side information is discussed. Besides the variant joint s/c decoders presented, we also show that the available sets of autocorrelations consist of a convex volume, and its structure can be found using the Simplex algorithm. △ Less

Submitted 21 December, 2004; originally announced December 2004.

Comments: 13 pages, to appear in "Progress in Theoretical Physics Supplement", May 2005

An Efficient MN-Algorithm for Joint Source-Channel Coding

Authors: Haggai Kfir, Eyal Shpilman, Ido Kanter

Abstract: Belief Propagation (BP) decoding of LDPC codes is extended to the case of Joint Source-Channel coding. The uncompressed source is treated as a Markov process, characterized by a transition matrix, T, which is utilized as side information for the Joint scheme. The method is based on the ability to calculate a Dynamical Block Prior (DBP), for each decoded symbol separately, and re-estimate this pr… ▽ More Belief Propagation (BP) decoding of LDPC codes is extended to the case of Joint Source-Channel coding. The uncompressed source is treated as a Markov process, characterized by a transition matrix, T, which is utilized as side information for the Joint scheme. The method is based on the ability to calculate a Dynamical Block Prior (DBP), for each decoded symbol separately, and re-estimate this prior after every iteration of the BP decoder. We demonstrate the implementation of this method using MacKay and Neel's LDPC algorithm over GF(q), and present simulation results indicating that the proposed scheme is comparable with Separation scheme, even when advanced compression algorithms (such as AC, PPM) are used. The extension to 2D (and higher) arrays of symbols is straight-forward. The possibility of using the proposed scheme without side information is briefly sketched. △ Less

Submitted 21 December, 2004; originally announced December 2004.

Journal ref: Proceedings of ISITA-2004, pp. 200-205, Parma, Italy, October 2004

Neural cryptography with queries

Authors: Andreas Ruttor, Wolfgang Kinzel, Ido Kanter

Abstract: Neural cryptography is based on synchronization of tree parity machines by mutual learning. We extend previous key-exchange protocols by replacing random inputs with queries depending on the current state of the neural networks. The probability of a successful attack is calculated for different model parameters using numerical simulations. The results show that queries restore the security again… ▽ More Neural cryptography is based on synchronization of tree parity machines by mutual learning. We extend previous key-exchange protocols by replacing random inputs with queries depending on the current state of the neural networks. The probability of a successful attack is calculated for different model parameters using numerical simulations. The results show that queries restore the security against cooperating attackers. The success probability can be reduced without increasing the average synchronization time. △ Less

Submitted 1 March, 2005; v1 submitted 15 November, 2004; originally announced November 2004.

Comments: 12 pages, 8 figures; typos corrected

Journal ref: J. Stat. Mech. (2005) P01009

Is a multiple excitation of a single atom equivalent to a single excitation of an ensemble of atoms?

Authors: Ido Kanter, Aviad Frydman, Assaf Ater

Abstract: Recent technological advances have enabled to isolate, control and measure the properties of a single atom, leading to the possibility to perform statistics on the behavior of single quantum systems. These experiments have enabled to check a question which was out of reach previously: Is the statistics of a repeatedly excitation of an atom N times equivalent to a single excitation of an ensemble… ▽ More Recent technological advances have enabled to isolate, control and measure the properties of a single atom, leading to the possibility to perform statistics on the behavior of single quantum systems. These experiments have enabled to check a question which was out of reach previously: Is the statistics of a repeatedly excitation of an atom N times equivalent to a single excitation of an ensemble of N atoms? We present a new method to analyze quantum measurements which leads to the postulation that the answer is most probably no. We discuss the merits of the analysis and its conclusion. △ Less

Submitted 16 May, 2004; originally announced May 2004.

Comments: 3 pages, 3 figures

Journal ref: Europhysics Letters, 69 874 (2005)

Utilizing hidden Markov processes as a new tool for experimental physics

Authors: Ido Kanter, Aviad Frydman, Asaf Ater

Abstract: A hidden Markov process is a well known concept in information theory and is used for a vast range of applications such as speech recognition and error correction. We bridge between two disciplines, experimental physics and advanced algorithms, and propose to use a physically oriented hidden Markov process as a new tool for analyzing experimental data. This tool enables one to extract valuable i… ▽ More A hidden Markov process is a well known concept in information theory and is used for a vast range of applications such as speech recognition and error correction. We bridge between two disciplines, experimental physics and advanced algorithms, and propose to use a physically oriented hidden Markov process as a new tool for analyzing experimental data. This tool enables one to extract valuable information on physical parameters of complex systems. We demonstrate the usefulness of this technique on low dimensional electronic systems which exhibit time dependent resistance noise. This method is expected to become a standard technique in experimental physics. △ Less

Submitted 22 February, 2004; originally announced February 2004.

Comments: 4 pages, 4 figures

Journal ref: Europhysics Letters, 69, 798 (2005)

Cooperating Attackers in Neural Cryptography

Authors: L. N. Shacham, E. Klein, R. Mislovaty, I. Kanter, W. Kinzel

Abstract: A new and successful attack strategy in neural cryptography is presented. The neural cryptosystem, based on synchronization of neural networks by mutual learning, has been recently shown to be secure under different attack strategies. The advanced attacker presented here, named the ``Majority-Flipping Attacker'', is the first whose success does not decay with the parameters of the model. This ne… ▽ More A new and successful attack strategy in neural cryptography is presented. The neural cryptosystem, based on synchronization of neural networks by mutual learning, has been recently shown to be secure under different attack strategies. The advanced attacker presented here, named the ``Majority-Flipping Attacker'', is the first whose success does not decay with the parameters of the model. This new attacker's outstanding success is due to its using a group of attackers which cooperate throughout the synchronization process, unlike any other attack strategy known. An analytical description of this attack is also presented, and fits the results of simulations. △ Less

Submitted 2 December, 2003; originally announced December 2003.

Neural cryptography with feedback

Authors: Andreas Ruttor, Wolfgang Kinzel, Lanir Shacham, Ido Kanter

Abstract: Neural cryptography is based on a competition between attractive and repulsive stochastic forces. A feedback mechanism is added to neural cryptography which increases the repulsive forces. Using numerical simulations and an analytic approach, the probability of a successful attack is calculated for different model parameters. Scaling laws are derived which show that feedback improves the securit… ▽ More Neural cryptography is based on a competition between attractive and repulsive stochastic forces. A feedback mechanism is added to neural cryptography which increases the repulsive forces. Using numerical simulations and an analytic approach, the probability of a successful attack is calculated for different model parameters. Scaling laws are derived which show that feedback improves the security of the system. In addition, a network with feedback generates a pseudorandom bit sequence which can be used to encrypt and decrypt a secret message. △ Less

Submitted 15 July, 2004; v1 submitted 26 November, 2003; originally announced November 2003.

Comments: 8 pages, 10 figures; abstract changed, references updated

Journal ref: Phys. Rev. E 69, 046110 (2004)

An efficient joint source-channel coding for a D-dimensional array

Authors: Ido Kanter, Haggai Kfir, Shahar Keren

Abstract: An efficient joint source-channel (S/C) decoder based on the side information of the source and on the MN-Gallager Code over Galois fields, $q$, is presented. The dynamical posterior probabilities are derived either from the statistical mechanical approach for calculation of the entropy for the correlated sequences, or by the Markovian joint S/C algorithm. The Markovian joint S/C decoder has man… ▽ More An efficient joint source-channel (S/C) decoder based on the side information of the source and on the MN-Gallager Code over Galois fields, $q$, is presented. The dynamical posterior probabilities are derived either from the statistical mechanical approach for calculation of the entropy for the correlated sequences, or by the Markovian joint S/C algorithm. The Markovian joint S/C decoder has many advantages over the statistical mechanical approach, among them: (a) there is no need for the construction and the diagonalization of a $q \times q$ matrix and for a solution to saddle point equations in $q$ dimensions; (b) a generalization to a joint S/C coding of an array of two-dimensional bits (or higher dimensions) is achievable; (c) using parametric estimation, an efficient joint S/C decoder with the lack of side information is discussed. Besides the variant joint S/C decoders presented, we also show that the available sets of autocorrelations consist of a convex volume, and its structure can be found using the Simplex algorithm. △ Less

Submitted 15 August, 2003; originally announced August 2003.

Comments: 19 pages and 12 figures

Statistical mechanical aspects of joint source-channel coding

Authors: Ido Kanter, Haggai Kfir

Abstract: An MN-Gallager Code over Galois fields, $q$, based on the Dynamical Block Posterior probabilities (DBP) for messages with a given set of autocorrelations is presented with the following main results: (a) for a binary symmetric channel the threshold, $f_c$, is extrapolated for infinite messages using the scaling relation for the median convergence time, $t_{med} \propto 1/(f_c-f)$; (b) a degradat… ▽ More An MN-Gallager Code over Galois fields, $q$, based on the Dynamical Block Posterior probabilities (DBP) for messages with a given set of autocorrelations is presented with the following main results: (a) for a binary symmetric channel the threshold, $f_c$, is extrapolated for infinite messages using the scaling relation for the median convergence time, $t_{med} \propto 1/(f_c-f)$; (b) a degradation in the threshold is observed as the correlations are enhanced; (c) for a given set of autocorrelations the performance is enhanced as $q$ is increased; (d) the efficiency of the DBP joint source-channel coding is slightly better than the standard gzip compression method; (e) for a given entropy, the performance of the DBP algorithm is a function of the decay of the correlation function over large distances. △ Less

Submitted 11 May, 2003; v1 submitted 3 March, 2003; originally announced March 2003.

Comments: 6 pages

Public channel cryptography by synchronization of neural networks and chaotic maps

Authors: Rachel Mislovaty, Einat Klein, Ido Kanter, Wolfgang Kinzel

Abstract: Two different kinds of synchronization have been applied to cryptography: Synchronization of chaotic maps by one common external signal and synchronization of neural networks by mutual learning. By combining these two mechanisms, where the external signal to the chaotic maps is synchronized by the nets, we construct a hybrid network which allows a secure generation of secret encryption keys over… ▽ More Two different kinds of synchronization have been applied to cryptography: Synchronization of chaotic maps by one common external signal and synchronization of neural networks by mutual learning. By combining these two mechanisms, where the external signal to the chaotic maps is synchronized by the nets, we construct a hybrid network which allows a secure generation of secret encryption keys over a public channel. The security with respect to attacks, recently proposed by Shamir et al, is increased by chaotic synchronization. △ Less

Submitted 5 February, 2003; originally announced February 2003.

Comments: 4 pages

Communication near the channel capacity with an absence of compression: Statistical Mechanical Approach

Authors: Ido Kanter, Hanan Rosemarin

Abstract: The generalization of Shannon's theory to include messages with given autocorrelations is presented. The analytical calculation of the channel capacity is based on the transfer matrix method of the effective 1D Hamiltonian. This bridge between statistical physics and information theory leads to efficient Low-Density Parity-Check Codes over Galois fields that nearly saturate the channel capacity.… ▽ More The generalization of Shannon's theory to include messages with given autocorrelations is presented. The analytical calculation of the channel capacity is based on the transfer matrix method of the effective 1D Hamiltonian. This bridge between statistical physics and information theory leads to efficient Low-Density Parity-Check Codes over Galois fields that nearly saturate the channel capacity. The novel idea of the decoder is the dynamical updating of the prior block probabilities which are derived from the transfer matrix solution and from the posterior probabilities of the neighboring blocks. Application and possible extensions are discussed, specifically the possibility of achieving the channel capacity without compression of the data. △ Less

Submitted 1 January, 2003; originally announced January 2003.

Comments: error correction, data compression

Mutual learning in a tree parity machine and its application to cryptography

Authors: Michal Rosen-Zvi, Einat Klein, Ido Kanter, Wolfgang Kinzel

Abstract: Mutual learning of a pair of tree parity machines with continuous and discrete weight vectors is studied analytically. The analysis is based on a mapping procedure that maps the mutual learning in tree parity machines onto mutual learning in noisy perceptrons. The stationary solution of the mutual learning in the case of continuous tree parity machines depends on the learning rate where a phase… ▽ More Mutual learning of a pair of tree parity machines with continuous and discrete weight vectors is studied analytically. The analysis is based on a mapping procedure that maps the mutual learning in tree parity machines onto mutual learning in noisy perceptrons. The stationary solution of the mutual learning in the case of continuous tree parity machines depends on the learning rate where a phase transition from partial to full synchronization is observed. In the discrete case the learning process is based on a finite increment and a full synchronized state is achieved in a finite number of steps. The synchronization of discrete parity machines is introduced in order to construct an ephemeral key-exchange protocol. The dynamic learning of a third tree parity machine (an attacker) that tries to imitate one of the two machines while the two still update their weight vectors is also analyzed. In particular, the synchronization times of the naive attacker and the flipping attacker recently introduced in [1] are analyzed. All analytical results are found to be in good agreement with simulation results. △ Less

Submitted 10 September, 2002; originally announced September 2002.

Neural Cryptography

Authors: Wolfgang Kinzel, Ido Kanter

Abstract: Two neural networks which are trained on their mutual output bits show a novel phenomenon: The networks synchronize to a state with identical time dependent weights. It is shown how synchronization by mutual learning can be applied to cryptography: secret key exchange over a public channel. Two neural networks which are trained on their mutual output bits show a novel phenomenon: The networks synchronize to a state with identical time dependent weights. It is shown how synchronization by mutual learning can be applied to cryptography: secret key exchange over a public channel. △ Less

Submitted 23 August, 2002; originally announced August 2002.

Comments: 9th International Conference on Neural Information Processing, Singapore, Nov. 2002

Parallel versus sequential updating for Belief Propagation decoding

Authors: H. Kfir, I. Kanter

Abstract: sequential updating scheme (SUS) for the belief propagation algorithm is proposed, and is compared with the parallel (regular) updating scheme (PUS). Simulation results on various codes indicate that the number of iterations of the belief algorithm for the SUS is about one half of the required iterations for the PUS, where both decoding algorithms have the same error correction properties. The c… ▽ More sequential updating scheme (SUS) for the belief propagation algorithm is proposed, and is compared with the parallel (regular) updating scheme (PUS). Simulation results on various codes indicate that the number of iterations of the belief algorithm for the SUS is about one half of the required iterations for the PUS, where both decoding algorithms have the same error correction properties. The complexity per iteration for both schemes is similar, resulting in a lower total complexity for the SUS. The explanation of this effect is related to the inter-iteration information sharing, which is a property of only the SUS, and which increases the "correction gain" per iteration △ Less

Submitted 7 July, 2002; originally announced July 2002.

Comments: 15 pages, 3 figures, submitted to Phys. Rev. E june 2002

A secure key-exchange protocol with an absence of injective functions

Authors: R. Mislovaty, Y. Perchenok, Ido Kanter, Wolfgang Kinzel

Abstract: The security of neural cryptography is investigated. A key-exchange protocol over a public channel is studied where the parties exchanging secret messages use multilayer neural networks which are trained by their mutual output bits and synchronize to a time dependent secret key. The weights of the networks have integer values between $\pm L$. Recently an algorithm for an eavesdropper which could… ▽ More The security of neural cryptography is investigated. A key-exchange protocol over a public channel is studied where the parties exchanging secret messages use multilayer neural networks which are trained by their mutual output bits and synchronize to a time dependent secret key. The weights of the networks have integer values between $\pm L$. Recently an algorithm for an eavesdropper which could break the key was introduced by Shamir et al. [adi]. We show that the synchronization time increases with $L^2$ while the probability to find a successful attacker decreases exponentially with $L$. Hence for large $L$ we find a secure key-exchange protocol which depends neither on number theory nor on injective trapdoor functions used in conventional cryptography. △ Less

Submitted 12 June, 2002; originally announced June 2002.

Comments: 4 pages, 4 figures

Interacting neural networks and cryptography

Authors: Wolfgang Kinzel, Ido Kanter

Abstract: Two neural networks which are trained on their mutual output bits are analysed using methods of statistical physics. The exact solution of the dynamics of the two weight vectors shows a novel phenomenon: The networks synchronize to a state with identical time dependent weights. Extending the models to multilayer networks with discrete weights, it is shown how synchronization by mutual learning c… ▽ More Two neural networks which are trained on their mutual output bits are analysed using methods of statistical physics. The exact solution of the dynamics of the two weight vectors shows a novel phenomenon: The networks synchronize to a state with identical time dependent weights. Extending the models to multilayer networks with discrete weights, it is shown how synchronization by mutual learning can be applied to secret key exchange over a public channel. △ Less

Submitted 1 March, 2002; originally announced March 2002.

Comments: Invited talk for the meeting of the German Physical Society

Learning and predicting time series by neural networks

Authors: Ansgar Freking, Wolfgang Kinzel, Ido Kanter

Abstract: Artificial neural networks which are trained on a time series are supposed to achieve two abilities: firstly to predict the series many time steps ahead and secondly to learn the rule which has produced the series. It is shown that prediction and learning are not necessarily related to each other. Chaotic sequences can be learned but not predicted while quasiperiodic sequences can be well predic… ▽ More Artificial neural networks which are trained on a time series are supposed to achieve two abilities: firstly to predict the series many time steps ahead and secondly to learn the rule which has produced the series. It is shown that prediction and learning are not necessarily related to each other. Chaotic sequences can be learned but not predicted while quasiperiodic sequences can be well predicted but not learned. △ Less

Submitted 28 February, 2002; originally announced February 2002.

Comments: 5 pages

Cryptography based on neural networks - analytical results

Authors: Michal Rosen-Zvi, Ido Kanter, Wolfgang Kinzel

Abstract: Mutual learning process between two parity feed-forward networks with discrete and continuous weights is studied analytically, and we find that the number of steps required to achieve full synchronization between the two networks in the case of discrete weights is finite. The synchronization process is shown to be non-self-averaging and the analytical solution is based on random auxiliary variab… ▽ More Mutual learning process between two parity feed-forward networks with discrete and continuous weights is studied analytically, and we find that the number of steps required to achieve full synchronization between the two networks in the case of discrete weights is finite. The synchronization process is shown to be non-self-averaging and the analytical solution is based on random auxiliary variables. The learning time of an attacker that is trying to imitate one of the networks is examined analytically and is found to be much longer than the synchronization time. Analytical results are found to be in agreement with simulations. △ Less

Submitted 20 February, 2002; originally announced February 2002.

Secure exchange of information by synchronization of neural networks

Authors: I. Kanter, W. Kinzel, E. Kanter

Abstract: A connection between the theory of neural networks and cryptography is presented. A new phenomenon, namely synchronization of neural networks is leading to a new method of exchange of secret messages. Numerical simulations show that two artificial networks being trained by Hebbian learning rule on their mutual outputs develop an antiparallel state of their synaptic weights. The synchronized weig… ▽ More A connection between the theory of neural networks and cryptography is presented. A new phenomenon, namely synchronization of neural networks is leading to a new method of exchange of secret messages. Numerical simulations show that two artificial networks being trained by Hebbian learning rule on their mutual outputs develop an antiparallel state of their synaptic weights. The synchronized weights are used to construct an ephemeral key exchange protocol for a secure transmission of secret data. It is shown that an opponent who knows the protocol and all details of any transmission of the data has no chance to decrypt the secret message, since tracking the weights is a hard problem compared to synchronization. The complexity of the generation of the secure channel is linear with the size of the network. △ Less

Submitted 7 February, 2002; originally announced February 2002.

Comments: 11 pages, 5 figures

Journal ref: Europhys. Lett. 57, pp. 141-147 (2002

Statistical properties of contact vectors

Authors: A. Kabakcioglu, I. Kanter, M. Vendruscolo, E. Domany

Abstract: We study the statistical properties of contact vectors, a construct to characterize a protein's structure. The contact vector of an N-residue protein is a list of N integers n_i, representing the number of residues in contact with residue i. We study analytically (at mean-field level) and numerically the amount of structural information contained in a contact vector. Analytical calculations reve… ▽ More We study the statistical properties of contact vectors, a construct to characterize a protein's structure. The contact vector of an N-residue protein is a list of N integers n_i, representing the number of residues in contact with residue i. We study analytically (at mean-field level) and numerically the amount of structural information contained in a contact vector. Analytical calculations reveal that a large variance in the contact numbers reduces the degeneracy of the mapping between contact vectors and structures. Exact enumeration for lengths up to N=16 on the three dimensional cubic lattice indicates that the growth rate of number of contact vectors as a function of N is only 3% less than that for contact maps. In particular, for compact structures we present numerical evidence that, practically, each contact vector corresponds to only a handful of structures. We discuss how this information can be used for better structure prediction. △ Less

Submitted 4 September, 2001; originally announced September 2001.

Comments: 20 pages, 6 figures

Multilayer neural networks with extensively many hidden units

Authors: Michal Rosen-Zvi, Andreas Engel, Ido Kanter

Abstract: The information processing abilities of a multilayer neural network with a number of hidden units scaling as the input dimension are studied using statistical mechanics methods. The mapping from the input layer to the hidden units is performed by general symmetric Boolean functions whereas the hidden layer is connected to the output by either discrete or continuous couplings. Introducing an over… ▽ More The information processing abilities of a multilayer neural network with a number of hidden units scaling as the input dimension are studied using statistical mechanics methods. The mapping from the input layer to the hidden units is performed by general symmetric Boolean functions whereas the hidden layer is connected to the output by either discrete or continuous couplings. Introducing an overlap in the space of Boolean functions as order parameter the storage capacity if found to scale with the logarithm of the number of implementable Boolean functions. The generalization behaviour is smooth for continuous couplings and shows a discontinuous transition to perfect generalization for discrete ones. △ Less

Submitted 1 April, 2001; originally announced April 2001.

Comments: 4 pages, 2 figures

Low autocorrelated multi-phase sequences

Authors: Liat Ein-Dor, Ido Kanter, Wolfgang KJinzel

Abstract: The interplay between the ground state energy of the generalized Bernasconi model to multi-phase, and the minimal value of the maximal autocorrelation function, $C_{max}=\max_K{|C_K|}$, $K=1,..N-1$, is examined analytically and the main results are: (a) The minimal value of $\min_N{C_{max}}$ is $0.435\sqrt{N}$ significantly smaller than the typical value for random sequences… ▽ More The interplay between the ground state energy of the generalized Bernasconi model to multi-phase, and the minimal value of the maximal autocorrelation function, $C_{max}=\max_K{|C_K|}$, $K=1,..N-1$, is examined analytically and the main results are: (a) The minimal value of $\min_N{C_{max}}$ is $0.435\sqrt{N}$ significantly smaller than the typical value for random sequences $O(\sqrt{\log{N}}\sqrt{N})$. (b) $\min_N{C_{max}}$ over all sequences of length N is obtained in an energy which is about 30% above the ground-state energy of the generalized Bernasconi model, independent of the number of phases m. (c) The maximal merit factor $F_{max}$ grows linearly with m. (d) For a given N, $\min_N{C_{max}}\sim\sqrt{N/m}$ indicating that for m=N, $\min_N{C_{max}}=1$, i.e. a Barker code exits. The analytical results are confirmed by simulations. △ Less

Submitted 8 March, 2001; originally announced March 2001.

Comments: 4 pages, 4 figures

Training a perceptron in a discrete weight space

Authors: Michal Rosen-Zvi, Ido Kanter

Abstract: On-line and batch learning of a perceptron in a discrete weight space, where each weight can take $2 L+1$ different values, are examined analytically and numerically. The learning algorithm is based on the training of the continuous perceptron and prediction following the clipped weights. The learning is described by a new set of order parameters, composed of the overlaps between the teacher and… ▽ More On-line and batch learning of a perceptron in a discrete weight space, where each weight can take $2 L+1$ different values, are examined analytically and numerically. The learning algorithm is based on the training of the continuous perceptron and prediction following the clipped weights. The learning is described by a new set of order parameters, composed of the overlaps between the teacher and the continuous/clipped students. Different scenarios are examined among them on-line learning with discrete/continuous transfer functions and off-line Hebb learning. The generalization error of the clipped weights decays asymptotically as $exp(-K α^2)$/$exp(-e^{|λ| α})$ in the case of on-line learning with binary/continuous activation functions, respectively, where $α$ is the number of examples divided by N, the size of the input vector and $K$ is a positive constant that decays linearly with 1/L. For finite $N$ and $L$, a perfect agreement between the discrete student and the teacher is obtained for $α\propto \sqrt{L \ln(NL)}$. A crossover to the generalization error $\propto 1/α$, characterized continuous weights with binary output, is obtained for synaptic depth $L > O(\sqrt{N})$. △ Less

Submitted 27 February, 2001; originally announced February 2001.

Comments: 10 pages, 5 figs., submitted to PRE

Generation of unpredictable time series by a Neural Network

Authors: Richard Metzler, Wolfgang Kinzel, Liat Ein-Dor, Ido Kanter

Abstract: A perceptron that learns the opposite of its own output is used to generate a time series. We analyse properties of the weight vector and the generated sequence, like the cycle length and the probability distribution of generated sequences. A remarkable suppression of the autocorrelation function is explained, and connections to the Bernasconi model are discussed. If a continuous transfer functi… ▽ More A perceptron that learns the opposite of its own output is used to generate a time series. We analyse properties of the weight vector and the generated sequence, like the cycle length and the probability distribution of generated sequences. A remarkable suppression of the autocorrelation function is explained, and connections to the Bernasconi model are discussed. If a continuous transfer function is used, the system displays chaotic and intermittent behaviour, with the product of the learning rate and amplification as a control parameter. △ Less

Submitted 3 April, 2001; v1 submitted 17 November, 2000; originally announced November 2000.

Comments: 11 pages, 14 figures; slightly expanded and clarified, mistakes corrected; accepted for publication in PRE

Secure and linear cryptosystems using error-correcting codes

Authors: I. Kanter, E. Kanter, L. Ein-Dor

Abstract: A public-key cryptosystem, digital signature and authentication procedures based on a Gallager-type parity-check error-correcting code are presented. The complexity of the encryption and the decryption processes scale linearly with the size of the plaintext Alice sends to Bob. The public-key is pre-corrupted by Bob, whereas a private-noise added by Alice to a given fraction of the ciphertext of… ▽ More A public-key cryptosystem, digital signature and authentication procedures based on a Gallager-type parity-check error-correcting code are presented. The complexity of the encryption and the decryption processes scale linearly with the size of the plaintext Alice sends to Bob. The public-key is pre-corrupted by Bob, whereas a private-noise added by Alice to a given fraction of the ciphertext of each encrypted plaintext serves to increase the secure channel and is the cornerstone for digital signatures and authentication. Various scenarios are discussed including the possible actions of the opponent Oscar as an eavesdropper or as a disruptor. △ Less

Submitted 27 August, 2000; originally announced August 2000.

Journal ref: Europhys. Lett. 51 (15 July), 244 (2000)

On time's arrow in Ehrenfest models with reversible deterministic dynamics

Authors: R. Metzler, W. Kinzel, I. Kanter

Abstract: We introduce a deterministic, time-reversible version of the Ehrenfest urn model. The distribution of first-passage times from equilibrium to non-equilibrium states and vice versa is calculated. We find that average times for transition to non-equilibrium always scale exponentially with the system size, whereas the time scale for relaxation to equilibrium depends on microscopic dynamics. To illu… ▽ More We introduce a deterministic, time-reversible version of the Ehrenfest urn model. The distribution of first-passage times from equilibrium to non-equilibrium states and vice versa is calculated. We find that average times for transition to non-equilibrium always scale exponentially with the system size, whereas the time scale for relaxation to equilibrium depends on microscopic dynamics. To illustrate this, we also look at deterministic and stochastic versions of the Ehrenfest model with a distribution of microscopic relaxation times. △ Less

Submitted 25 July, 2000; originally announced July 2000.

Comments: 6 pages, 7 figures, revtex