Plausible Algorithm Research Articles

The Velocity Field Predicted by the Optical Redshift Survey

We have used the Optical Redshift Survey (ORS) of Santiago et al. to construct the gravity field due to fluctuations in the galaxy density field out to distances of 8000 km s-1. At large scales where linear theory applies, the comparison of this gravity field with the observed peculiar velocity field offers a powerful cosmological probe because the predicted flow field is proportional to the parameter Ω0.6/b, where Ω is the matter density and b is the bias of the galaxy distribution. The more densely sampled ORS gravity field, to excellent approximation, matches that of the earlier IRAS 1.2 Jy redshift survey of Fisher et al., provided β is reduced by a factor of bopt/bIRAS ≈ 1.4. Apart from this scaling, the most significant difference between the ORS and IRAS fields is induced by differing estimates of the overdensity of the Virgo cluster. Neither of these gravity fields is consistent with the peculiar velocity field constructed from the full Mark III sample. We find that a simple but plausible nonlinear bias algorithm for the galaxy distribution relative to the mass has a negligible effect on the derived fields. We conclude that the substitution of optical for IRAS catalogs cannot alone resolve the discrepancies between the IRAS gravity field and the Mark III peculiar velocity field.

Spatial and temporal pattern analysis via spiking neurons

Spiking neurons, receiving temporally encoded inputs, can compute radial basis functions (RBFs) by storing the relevant information in their delays. In this paper we show how these delays can be learned using exclusively locally available information (basically the time difference between the pre- and postsynaptic spikes). Our approach gives rise to a biologically plausible algorithm for finding clusters in a high-dimensional input space with networks of spiking neurons, even if the environment is changing dynamically. Furthermore, we show that our learning mechanism makes it possible that such RBF neurons can perform some kind of feature extraction where they recognize that only certain input coordinates carry relevant information. Finally we demonstrate that this model allows the recognition of temporal sequences even if they are distorted in various ways.

The complexity of understanding line drawings of origami scenes

This paper deals with the interpretation of line drawings of Origami scenes (Kanade, 1980), that is scenes obtained by assembling planar panels of negligible thickness, and it addresses the computational complexity of the problem of consistently assigning suitable labels to the segments describing 3D properties as convexity, concavity and occlusion (labeling problem). The main results of the paper are the following: (a) the labeling problem for line drawings of Origami scenes is NP, as for the case of trihedral scenes; (b) the problem remains NP even if the location of the vanishing points in the image plane is given, whereas for trihedral scenes the problem was polynomially solvable; (c) in case the vanishing points are known the labeling problem can be subdivided into two subproblems, the paneling problem and the labeling-a-paneled-line-drawing problem which are both polynomially solvable (there may be, however, exponentially many legal panelings to check against labelability). The approach provides geometrical constraints which help select ‘natural’ interpretations even in the presence of occlusions and in the absence of apparent 3D-symmetries, and which may help highlight the role of geometrical regularities in the design of biologically plausible algorithms for the interpretation of line drawings.

Large-scale structure in a low-bias universe

view Abstract Citations (95) References (31) Co-Reads Similar Papers Volume Content Graphics Metrics Export Citation NASA/ADS Large-Scale Structure in a Low-Bias Universe Couchman, H. M. P. ; Carlberg, R. G. Abstract We show that a highly evolved, or low bias, {OMEGA} = 1 cold dark matter model is compatible with current observational data. Nonlinear evolution is sufficient to account for recent observations of structure on scales ~20-40 h^-1^ Mpc. Within dissipationless simulations we adopt a physically plausible algorithm to select galaxies which follows their merging. Galaxy tracers are selected at redshifts ~3 and then assembled into galaxies at the current epoch. The resulting distribution is an acceptable fit to the data at a linear bias, b_l_, as low as 0.8 in a CDM model: the current microwave background limit. A simple technique is used to derive statistically correct infinite volume large-scale correlations from the finite simulation cubes for both galaxies and the density; ξ_gg_ and ξ_pp_. For low bias values, b_l_~ 1, nonlinear evolution on large scales causes a modest increase in the zero crossing scale of ξ_pp_ and a substantial increase in the amplitude where it falls steeply through zero. On large scales ξ_gg_ is enhanced beyond ξ_pp_ reflecting a significant contribution from galaxies dynamically clustering into groups. On small scales ξ_pp_ steepens with evolution and, for b_l_ <~ 1.5, is not well approximated by a single power law. For b_l_ = 0.8 the galaxy correlation is a power law on small scales, ξ_gg_ is proportional to r^-1.7^, and is strongly antibiased relative to the dark matter by a factor of ~2.5 - 3 in relative correlation amplitude at 1 h^-1^ Mpc. This antibias is a consequence of increased merging in cluster halos. Within clusters, galaxies are a cooler and hence more centrally condensed population relative to the dark matter. The velocity dispersion of rich cluster galaxies is ~70% that of the dark matter. Virial analysis of a large cluster underestimates the cluster mass by a factor ~4, and gives a local estimate of {OMEGA}~ 0.27. The one-dimensional pairwise dispersion of all galaxies found in the model is ~490 km s^-1^ at 0.5 h^-1^ Mpc. A large pairwise velocity bias exists between galaxies and density, b_v_~ 0.37. Roughly half of the velocity bias is due to the cooler galaxy population within individual cluster halos. The remainder arises because relatively fewer galaxies occur within massive high dispersion regions. Straightforward application of the cosmic virial theorem implies a {OMEGA} ~0.2. Publication: The Astrophysical Journal Pub Date: April 1992 DOI: 10.1086/171222 Bibcode: 1992ApJ...389..453C Keywords: Cosmology; Dark Matter; Galactic Clusters; Galactic Evolution; Red Shift; Virial Theorem; Astronomical Models; Computerized Simulation; Infrared Astronomy Satellite; Universe; Astrophysics; COSMOLOGY: DARK MATTER; COSMOLOGY: THEORY; GALAXIES: CLUSTERING; COSMOLOGY: LARGE-SCALE STRUCTURE OF UNIVERSE full text sources ADS |

On Listing's law

Listing's law states that visual directions of sight are related to rotations of the eye so that all rotation axes lie in a plane. The geometry ofSO(3) indicates several plausible algorithms how the human brain could relate vision to eye movements satisfying Listing's law, and suggests crucial experiments which we have carried out.

A network of two drive-reinforcement neurons that learns a solution to a realtime dynamic control problem: [formula omitted]. Avionics Laboratory, Air Force Wright Aeronautical Laboratories, AFWAL/AAAT-3, Wright-Patterson Air Force Base, Ohio 45433

Pattern recognition (recognizing a pattern from inputs) and recall (describing or predicting the inputs associated with a recognizable pattern) are essential for neural-symbolic processing and cognitive capacities. Without them the brain cannot interact with the world e.g.: form internal representations and recall memory upon which it can perform logic and reason. Neural networks are efficient, biologically plausible algorithms that can perform large scale recognition. However, most neural network models of recognition perform recognition but not recall: they are sub-symbolic. It remains difficult to connect models of recognition with models of logic and emulate fundamental brain functions, because of the symbolic recall limitation.I introduce a completely symbolic neural network that is similar in function to standard feedforward neural networks but uses feedforward-feedback connections similar to auto-associative networks. However, unlike auto-associative networks, the symmetrical feedback connections are inhibitory not excitatory. Initially it may seem counterintuitive that recognition can even occur because the top-down connections are self-inhibitory. The self-inhibitory configuration is used to implement a gradient descent mechanism that functions during recognition not learning. The purpose of the gradient-descent is not to learn weights (weights are still learned during learning) but to find neuron activation. The advantage of this approach is the weights can now be symbolic (representing prototypes of expected patterns) allowing recall within neural networks. Moreover, considering the costs of both learning and recognition, this approach may be more efficient than feedforward recognition. I show that this model mathematically emulates standard feedforward model equations in single layer networks without hidden units. Comparisons that include more layers are planned in the future.

Induction of Horn clauses: methods and the plausible generalization algorithm

We are considering the problem of induction of uncertainty-free descriptions for concepts when arbitrary background knowledge is available, to perform constructive induction, for instance. As an idealised context, we consider that descriptions and rules in the knowledge-base are in the form of definite (Horn) clauses. Using a recently developed model of generality for definite clauses, we argue that some induction techniques are inadequate for the problem. We propose a framework where induction is viewed as a process of model-directed discovery of consistent patterns (constraints and rules) in data, and describe a new algorithm, the Plausible Generalization Algorithm, that has been used to investigate the sub-problem of discovering rules. The algorithm raises a number of interesting questions: How can we identify irrelevance during the generalization process? How can our knowledgebase answer queries of the form “What do (objects) X and Y have in common that is relevant to (situation) S?”