Refine
Year of publication
Document Type
- Preprint (739)
- Article (392)
- Working Paper (114)
- Doctoral Thesis (54)
- Conference Proceeding (37)
- Report (20)
- Part of a Book (9)
- Book (5)
- Master's Thesis (5)
- Bachelor Thesis (3)
Language
- English (1382) (remove)
Has Fulltext
- yes (1382) (remove)
Is part of the Bibliography
- no (1382)
Keywords
- Lambda-Kalkül (20)
- Heavy Ion Experiments (19)
- Formale Semantik (11)
- Hadron-Hadron Scattering (11)
- Hadron-Hadron scattering (experiments) (10)
- lambda calculus (9)
- LHC (8)
- Operationale Semantik (8)
- Heavy-ion collision (7)
- Kongress (7)
Institute
- Informatik (1382) (remove)
One of the most interesting domains of feedforward networks is the processing of sensor signals. There do exist some networks which extract most of the information by implementing the maximum entropy principle for Gaussian sources. This is done by transforming input patterns to the base of eigenvectors of the input autocorrelation matrix with the biggest eigenvalues. The basic building block of these networks is the linear neuron, learning with the Oja learning rule. Nevertheless, some researchers in pattern recognition theory claim that for pattern recognition and classification clustering transformations are needed which reduce the intra-class entropy. This leads to stable, reliable features and is implemented for Gaussian sources by a linear transformation using the eigenvectors with the smallest eigenvalues. In another paper (Brause 1992) it is shown that the basic building block for such a transformation can be implemented by a linear neuron using an Anti-Hebb rule and restricted weights. This paper shows the analog VLSI design for such a building block, using standard modules of multiplication and addition. The most tedious problem in this VLSI-application is the design of an analog vector normalization circuitry. It can be shown that the standard approaches of weight summation will not give the convergence to the eigenvectors for a proper feature transformation. To avoid this problem, our design differs significantly from the standard approaches by computing the real Euclidean norm. Keywords: minimum entropy, principal component analysis, VLSI, neural networks, surface approximation, cluster transformation, weight normalization circuit.
Let b1, . . . , bm 2 IRn be an arbitrary basis of lattice L that is a block Korkin Zolotarev basis with block size ¯ and let ¸i(L) denote the successive minima of lattice L. We prove that for i = 1, . . . ,m 4 i + 3 ° 2 i 1 ¯ 1 ¯ · kbik2/¸i(L)2 · ° 2m i ¯ 1 ¯ i + 3 4 where °¯ is the Hermite constant. For ¯ = 3 we establish the optimal upper bound kb1k2/¸1(L)2 · µ3 2¶m 1 2 1 and we present block Korkin Zolotarev lattice bases for which this bound is tight. We improve the Nearest Plane Algorithm of Babai (1986) using block Korkin Zolotarev bases. Given a block Korkin Zolotarev basis b1, . . . , bm with block size ¯ and x 2 L(b1, . . . , bm) a lattice point v can be found in time ¯O(¯) satisfying kx vk2 · m° 2m ¯ 1 ¯ minu2L kx uk2.
Parallel FFT-hashing
(1994)
We propose two families of scalable hash functions for collision resistant hashing that are highly parallel and based on the generalized fast Fourier transform (FFT). FFT hashing is based on multipermutations. This is a basic cryptographic primitive for perfect generation of diffusion and confusion which generalizes the boxes of the classic FFT. The slower FFT hash functions iterate a compression function. For the faster FFT hash functions all rounds are alike with the same number of message words entering each round.
Black box cryptanalysis applies to hash algorithms consisting of many small boxes, connected by a known graph structure, so that the boxes can be evaluated forward and backwards by given oracles. We study attacks that work for any choice of the black boxes, i.e. we scrutinize the given graph structure. For example we analyze the graph of the fast Fourier transform (FFT). We present optimal black box inversions of FFT-compression functions and black box constructions of collisions. This determines the minimal depth of FFT-compression networks for collision-resistant hashing. We propose the concept of multipermutation, which is a pair of orthogonal latin squares, as a new cryptographic primitive that generalizes the boxes of the FFT. Our examples of multipermutations are based on the operations circular rotation, bitwise xor, addition and multiplication.
We generalize the concept of block reduction for lattice bases from l2-norm to arbitrary norms. This extends the results of Schnorr. We give algorithms for block reduction and apply the resulting enumeration concept to solve subset sum problems. The deterministic algorithm solves all subset sum problems. For up to 66 weights it needs in average less then two hours on a HP 715/50 under HP-UX 9.05.
We study the following problem: given x element Rn either find a short integer relation m element Zn, so that =0 holds for the inner product <.,.>, or prove that no short integer relation exists for x. Hastad, Just Lagarias and Schnorr (1989) give a polynomial time algorithm for the problem. We present a stable variation of the HJLS--algorithm that preserves lower bounds on lambda(x) for infinitesimal changes of x. Given x \in {\RR}^n and \alpha \in \NN this algorithm finds a nearby point x' and a short integer relation m for x'. The nearby point x' is 'good' in the sense that no very short relation exists for points \bar{x} within half the x'--distance from x. On the other hand if x'=x then m is, up to a factor 2^{n/2}, a shortest integer relation for \mbox{x.} Our algorithm uses, for arbitrary real input x, at most \mbox{O(n^4(n+\log \alpha))} many arithmetical operations on real numbers. If x is rational the algorithm operates on integers having at most \mbox{O(n^5+n^3 (\log \alpha)^2 + \log (\|q x\|^2))} many bits where q is the common denominator for x.
We consider the problem of unifying a set of equations between second-order terms. Terms are constructed from function symbols, constant symbols and variables, and furthermore using monadic second-order variables that may stand for a term with one hole, and parametric terms. We consider stratified systems, where for every first-order and second-order variable, the string of second-order variables on the path from the root of a term to every occurrence of this variable is always the same. It is shown that unification of stratified second-order terms is decidable by describing a nondeterministic decision algorithm that eventually uses Makanin's algorithm for deciding the unifiability of word equations. As a generalization, we show that the method can be used as a unification procedure for non-stratified second-order systems, and describe conditions for termination in the general case.