Filters








13 Hits in 0.59 sec

Systemic Computation Using Graphics Processors [chapter]

Marjan Rouhipour, Peter J Bentley, Hooman Shayani
2010 Lecture Notes in Computer Science  
Previous work created the systemic computer -a model of computation designed to exploit many natural properties observed in biological systems, including parallelism. The approach has been proven through two existing implementations and many biological models and visualizations. However to date the systemic computer implementations have all been sequential simulations that do not exploit the true potential of the model. In this paper the first parallel implementation of systemic computation is
more » ... ntroduced. The GPU Systemic Computation Architecture is the first implementation that enables parallel systemic computation by exploiting multiple cores available in graphics processors. Comparisons with the serial implementation when running a genetic algorithm at different scales show that as the number of systems increases, the parallel architecture is several hundred times faster than the existing implementations, making it feasible to investigate systemic models of more complex biological systems.
doi:10.1007/978-3-642-15323-5_11 fatcat:tbn5orotbzdtzi3b2cvgn43upm

CAPRI-Net: Learning Compact CAD Shapes with Adaptive Primitive Assembly [article]

Fenggen Yu, Zhiqin Chen, Manyi Li, Aditya Sanghi, Hooman Shayani, Ali Mahdavi-Amiri, Hao Zhang
2021 arXiv   pre-print
We introduce CAPRI-Net, a neural network for learning compact and interpretable implicit representations of 3D computer-aided design (CAD) models, in the form of adaptive primitive assemblies. Our network takes an input 3D shape that can be provided as a point cloud or voxel grids, and reconstructs it by a compact assembly of quadric surface primitives via constructive solid geometry (CSG) operations. The network is self-supervised with a reconstruction loss, leading to faithful 3D
more » ... ns with sharp edges and plausible CSG trees, without any ground-truth shape assemblies. While the parametric nature of CAD models does make them more predictable locally, at the shape level, there is a great deal of structural and topological variations, which present a significant generalizability challenge to state-of-the-art neural models for 3D shapes. Our network addresses this challenge by adaptive training with respect to each test shape, with which we fine-tune the network that was pre-trained on a model collection. We evaluate our learning framework on both ShapeNet and ABC, the largest and most diverse CAD dataset to date, in terms of reconstruction quality, shape edges, compactness, and interpretability, to demonstrate superiority over current alternatives suitable for neural CAD reconstruction.
arXiv:2104.05652v1 fatcat:onnbfb7ikrdtnpxlyvuaoeukcq

The challenge of irrationality

Jean Krohn, Peter J. Bentley, Hooman Shayani
2009 Proceedings of the 11th Annual conference on Genetic and evolutionary computation - GECCO '09  
Computational development traditionally focuses on the use of an iterative, generative mapping process from genotype to phenotype in order to obtain complex phenotypes which comprise regularity, repetition and module reuse. This work examines whether an evolutionary computational developmental algorithm is capable of producing a phenotype with no known pattern at all: the irrational number PI. The paper summarizes the fractal protein algorithm, provides a new analysis of how fractals are
more » ... ed by the developmental process, then presents experiments, results and analysis showing that evolution is capable of producing an approximate algorithm for PI that goes beyond the limits of precision of the data types used.
doi:10.1145/1569901.1570000 dblp:conf/gecco/KrohnBS09 fatcat:mtyorouclvbvrn6pnww6yvimtq

Fast bio-inspired computation using a GPU-based systemic computer

Marjan Rouhipour, Peter J. Bentley, Hooman Shayani
2010 Parallel Computing  
Biology is inherently parallel. Models of biological systems and bio-inspired algorithms also share this parallelism, although most are simulated on serial computers. Previous work created the systemic computer -a new model of computation designed to exploit many natural properties observed in biological systems, including parallelism. The approach has been proven through two existing implementations and many biological models and visualizations. However to date the systemic computer
more » ... ions have all been sequential simulations that do not exploit the true potential of the model. In this paper the first ever parallel implementation of systemic computation is introduced. The GPU Systemic Computation Architecture is the first implementation that enables parallel systemic computation by exploiting the multiple cores available in graphics processors. Comparisons with the serial implementation when running two programs at different scales show that as the number of systems increases, the parallel architecture is several hundred times faster than the existing implementations, making it feasible to investigate systemic models of more complex biological systems.
doi:10.1016/j.parco.2010.07.004 fatcat:oyxyfaxntbebjmubeipdbuphma

UV-Net: Learning from Boundary Representations [article]

Pradeep Kumar Jayaraman, Aditya Sanghi, Joseph G. Lambourne, Karl D.D. Willis, Thomas Davies, Hooman Shayani, Nigel Morris
2021 arXiv   pre-print
We introduce UV-Net, a novel neural network architecture and representation designed to operate directly on Boundary representation (B-rep) data from 3D CAD models. The B-rep format is widely used in the design, simulation and manufacturing industries to enable sophisticated and precise CAD modeling operations. However, B-rep data presents some unique challenges when used with modern machine learning due to the complexity of the data structure and its support for both continuous non-Euclidean
more » ... ometric entities and discrete topological entities. In this paper, we propose a unified representation for B-rep data that exploits the U and V parameter domain of curves and surfaces to model geometry, and an adjacency graph to explicitly model topology. This leads to a unique and efficient network architecture, UV-Net, that couples image and graph convolutional neural networks in a compute and memory-efficient manner. To aid in future research we present a synthetic labelled B-rep dataset, SolidLetters, derived from human designed fonts with variations in both geometry and topology. Finally we demonstrate that UV-Net can generalize to supervised and unsupervised tasks on five datasets, while outperforming alternate 3D shape representations such as point clouds, voxels, and meshes.
arXiv:2006.10211v2 fatcat:vccqocfodvcihke2kk4lx6mvli

BRepNet: A topological message passing system for solid models [article]

Joseph G. Lambourne, Karl D.D. Willis, Pradeep Kumar Jayaraman, Aditya Sanghi, Peter Meltzer, Hooman Shayani
2021 arXiv   pre-print
Boundary representation (B-rep) models are the standard way 3D shapes are described in Computer-Aided Design (CAD) applications. They combine lightweight parametric curves and surfaces with topological information which connects the geometric entities to describe manifolds. In this paper we introduce BRepNet, a neural network architecture designed to operate directly on B-rep data structures, avoiding the need to approximate the model as meshes or point clouds. BRepNet defines convolutional
more » ... els with respect to oriented coedges in the data structure. In the neighborhood of each coedge, a small collection of faces, edges and coedges can be identified and patterns in the feature vectors from these entities detected by specific learnable parameters. In addition, to encourage further deep learning research with B-reps, we publish the Fusion 360 Gallery segmentation dataset. A collection of over 35,000 B-rep models annotated with information about the modeling operations which created each face. We demonstrate that BRepNet can segment these models with higher accuracy than methods working on meshes, and point clouds.
arXiv:2104.00706v2 fatcat:q2em4c5mafcjnal3d72wqugmru

A more bio-plausible approach to the evolutionary inference of finite state machines

Hooman Shayani, Peter J. Bentley
2007 Proceedings of the 2007 GECCO conference companion on Genetic and evolutionary computation - GECCO '07  
With resemblance of finite-state machines to some biological mechanisms in cells and numerous applications of finite automata in different fields, this paper uses analogies and metaphors to introduce an element of bio-plausibility to evolutionary grammatical inference. Inference of a finite-state machine that generalizes well over unseen input-output examples is an NP-complete problem. Heuristic algorithms exist to minimize the size of an FSM keeping it consistent with all the input-output
more » ... nces. However, their performance dramatically degrades in presence of noise in the training set. Evolutionary algorithms perform better for noisy data sets but they do not scale well and their performance drops as size or complexity of the target machine grows. Here, inspired by a biological perspective, an evolutionary algorithm with a novel representation and a new fitness function for inference of Moore finite-state machines of limited size is proposed and compared with one of the latest evolutionary techniques.
doi:10.1145/1274000.1274039 dblp:conf/gecco/ShayaniB07 fatcat:bdjhl63vtfddbozjvadt3wrnlm

A Multi-cellular Developmental Representation for Evolution of Adaptive Spiking Neural Microcircuits in an FPGA

Hooman Shayani, Peter J. Bentley, Andy M. Tyrrell
2009 2009 NASA/ESA Conference on Adaptive Hardware and Systems  
It has been shown that evolutionary and developmental processes can be used for emergence of scalability, robustness and fault-tolerance in hardware. However, designing a suitable representation for such processes is far from straightforward. Here, a bio-inspired developmental genotype-phenotype mapping for evolution of spiking neural microcircuits in an FPGA is introduced, based on a digital neuron model and cortex structure suggested and verified previously by the authors. The new
more » ... l process is based on complex multi-cellular proteinprotein and gene-protein interactions and signaling. Suitability of the representation for evolution of useful architectures and its adaptability is shown through statistical analysis and examples of scalability, modularity and fault-tolerance.
doi:10.1109/ahs.2009.39 dblp:conf/ahs/ShayaniBT09 fatcat:hy7bw7i6avevjjadkwtpyamgba

Hardware Implementation of a Bio-plausible Neuron Model for Evolution and Growth of Spiking Neural Networks on FPGA

Hooman Shayani, Peter J. Bentley, Andy M. Tyrrell
2008 2008 NASA/ESA Conference on Adaptive Hardware and Systems  
We propose a digital neuron model suitable for evolving and growing heterogeneous spiking neural networks on FPGAs by introducing a novel flexible dendrite architecture and the new PLAQIF (Piecewise-Linear Approximation of Quadratic Integrate and Fire) soma model. A network of 161 neurons and 1610 synapses was simulated, implemented, and verified on a Virtex-5 chip with 4210 times real-time speed with 1 ms resolution. The parametric flexibility of the soma model was shown through a set of
more » ... ments. NASA/ESA Conference on Adaptive Hardware and Systems 978-0-7695-3166-3/08 $25.00
doi:10.1109/ahs.2008.13 dblp:conf/ahs/ShayaniBT08 fatcat:zb33pcttgzgmfjklxpxexirzfi

UVStyle-Net: Unsupervised Few-shot Learning of 3D Style Similarity Measure for B-Reps [article]

Peter Meltzer, Hooman Shayani, Amir Khasahmadi, Pradeep Kumar Jayaraman, Aditya Sanghi, Joseph Lambourne
2021 arXiv   pre-print
Boundary Representations (B-Reps) are the industry standard in 3D Computer Aided Design/Manufacturing (CAD/CAM) and industrial design due to their fidelity in representing stylistic details. However, they have been ignored in the 3D style research. Existing 3D style metrics typically operate on meshes or pointclouds, and fail to account for end-user subjectivity by adopting fixed definitions of style, either through crowd-sourcing for style labels or hand-crafted features. We propose
more » ... , a style similarity measure for B-Reps that leverages the style signals in the second order statistics of the activations in a pre-trained (unsupervised) 3D encoder, and learns their relative importance to a subjective end-user through few-shot learning. Our approach differs from all existing data-driven 3D style methods since it may be used in completely unsupervised settings, which is desirable given the lack of publicly available labelled B-Rep datasets. More importantly, the few-shot learning accounts for the inherent subjectivity associated with style. We show quantitatively that our proposed method with B-Reps is able to capture stronger style signals than alternative methods on meshes and pointclouds despite its significantly greater computational efficiency. We also show it is able to generate meaningful style gradients with respect to the input shape, and that few-shot learning with as few as two positive examples selected by an end-user is sufficient to significantly improve the style measure. Finally, we demonstrate its efficacy on a large unlabeled public dataset of CAD models. Source code and data will be released in the future.
arXiv:2105.02961v3 fatcat:sxdznzourzddnbtvaz3g4yy2kq

UNIST: Unpaired Neural Implicit Shape Translation Network [article]

Qimin Chen, Johannes Merz, Aditya Sanghi, Hooman Shayani, Ali Mahdavi-Amiri, Hao Zhang
2021
We introduce UNIST, the first deep neural implicit model for general-purpose, unpaired shape-to-shape translation, in both 2D and 3D domains. Our model is built on autoencoding implicit fields, rather than point clouds which represents the state of the art. Furthermore, our translation network is trained to perform the task over a latent grid representation which combines the merits of both latent-space processing and position awareness, to not only enable drastic shape transforms but also well
more » ... preserve spatial features and fine local details for natural shape translations. With the same network architecture and only dictated by the input domain pairs, our model can learn both style-preserving content alteration and content-preserving style transfer. We demonstrate the generality and quality of the translation results, and compare them to well-known baselines.
doi:10.48550/arxiv.2112.05381 fatcat:6cinlgdwqbadhfubjlqhwefoay

A Cellular Structure for Online Routing of Digital Spiking Neuron Axons and Dendrites on FPGAs [chapter]

Hooman Shayani, Peter Bentley, Andy M. Tyrrell
Lecture Notes in Computer Science  
As a step towards creating evolutionary developmental neural networks on FPGAs, a bio-inspired cellular structure suitable for online routing of axons and dendrites on FPGAs based on a new digital spiking neuron model (introduced previously by the authors) is proposed here. This structure is designed to allow changing the routing of the dendrites and axons and formation/elimination of synapses on the fly by dynamic partial reconfiguration of the LUTs. The feasibility and techniques for
more » ... ing this structure on a Xilinx Virtex-5 FPGA are also studied.
doi:10.1007/978-3-540-85857-7_24 fatcat:lwb7l3owtnchjbus5sqt3kkn2m

Group-disentangled Representation Learning with Weakly-Supervised Regularization [article]

Linh Tran, Amir Hosein Khasahmadi, Aditya Sanghi, Saeid Asgari
2021 arXiv   pre-print
ACKNOWLEDGMENTS We thank Hooman Shayani and Tonya Custis for useful discussions and comments on the paper.  ... 
arXiv:2110.12185v1 fatcat:ywcwv65qbrbk7b57mrzj27x2ui