-
Solving Boltzmann Optimization Problems with Deep Learning
Authors:
Fiona Knoll,
John T. Daly,
Jess J. Meyer
Abstract:
Decades of exponential scaling in high performance computing (HPC) efficiency is coming to an end. Transistor based logic in complementary metal-oxide semiconductor (CMOS) technology is approaching physical limits beyond which further miniaturization will be impossible. Future HPC efficiency gains will necessarily rely on new technologies and paradigms of compute. The Ising model shows particular…
▽ More
Decades of exponential scaling in high performance computing (HPC) efficiency is coming to an end. Transistor based logic in complementary metal-oxide semiconductor (CMOS) technology is approaching physical limits beyond which further miniaturization will be impossible. Future HPC efficiency gains will necessarily rely on new technologies and paradigms of compute. The Ising model shows particular promise as a future framework for highly energy efficient computation. Ising systems are able to operate at energies approaching thermodynamic limits for energy consumption of computation. Ising systems can function as both logic and memory. Thus, they have the potential to significantly reduce energy costs inherent to CMOS computing by eliminating costly data movement. The challenge in creating Ising-based hardware is in optimizing useful circuits that produce correct results on fundamentally nondeterministic hardware. The contribution of this paper is a novel machine learning approach, a combination of deep neural networks and random forests, for efficiently solving optimization problems that minimize sources of error in the Ising model. In addition, we provide a process to express a Boltzmann probability optimization problem as a supervised machine learning problem.
△ Less
Submitted 30 January, 2024;
originally announced January 2024.
-
A Comprehensive Taxonomy of Cellular Automata
Authors:
Michiel Rollier,
Kallil M. C. Zielinski,
Aisling J. Daly,
Odemir M. Bruno,
Jan M. Baetens
Abstract:
Cellular automata (CAs) are fully-discrete dynamical models that have received much attention due to the fact that their relatively simple setup can nonetheless express highly complex phenomena. Despite the model's theoretical maturity and abundant computational power, the current lack of a complete survey on the 'taxonomy' of various families of CAs impedes efficient and interdisciplinary progres…
▽ More
Cellular automata (CAs) are fully-discrete dynamical models that have received much attention due to the fact that their relatively simple setup can nonetheless express highly complex phenomena. Despite the model's theoretical maturity and abundant computational power, the current lack of a complete survey on the 'taxonomy' of various families of CAs impedes efficient and interdisciplinary progress. This review paper mitigates that deficiency; it provides a methodical overview of five important CA 'families': asynchronous, stochastic, multi-state, extended-neighbourhood, and non-uniform CAs. These five CA families are subsequently presented from four angles. First, a rigorous mathematical definition is given. Second, we map prominent variations within each CA family, as such highlighting mathematical equivalences with types from other families. Third, we discuss the genotype and phenotype of these CA types by means of mathematical tools, indicating when established tools break down. Fourth, we conclude each section with a brief overview of applications related to information theory and mathematical modelling.
△ Less
Submitted 16 January, 2024;
originally announced January 2024.
-
Information Retrieval and Classification of Real-Time Multi-Source Hurricane Evacuation Notices
Authors:
Tingting Zhao,
Shubo Tian,
Jordan Daly,
Melissa Geiger,
Minna Jia,
**feng Zhang
Abstract:
For an approaching disaster, the tracking of time-sensitive critical information such as hurricane evacuation notices is challenging in the United States. These notices are issued and distributed rapidly by numerous local authorities that may spread across multiple states. They often undergo frequent updates and are distributed through diverse online portals lacking standard formats. In this study…
▽ More
For an approaching disaster, the tracking of time-sensitive critical information such as hurricane evacuation notices is challenging in the United States. These notices are issued and distributed rapidly by numerous local authorities that may spread across multiple states. They often undergo frequent updates and are distributed through diverse online portals lacking standard formats. In this study, we developed an approach to timely detect and track the locally issued hurricane evacuation notices. The text data were collected mainly with a spatially targeted web scra** method. They were manually labeled and then classified using natural language processing techniques with deep learning models. The classification of mandatory evacuation notices achieved a high accuracy (recall = 96%). We used Hurricane Ian (2022) to illustrate how real-time evacuation notices extracted from local government sources could be redistributed with a Web GIS system. Our method applied to future hurricanes provides live data for situation awareness to higher-level government agencies and news media. The archived data helps scholars to study government responses toward weather warnings and individual behaviors influenced by evacuation history. The framework may be applied to other types of disasters for rapid and targeted retrieval, classification, redistribution, and archiving of real-time government orders and notifications.
△ Less
Submitted 7 January, 2024;
originally announced January 2024.
-
Robustness for Free: Quality-Diversity Driven Discovery of Agile Soft Robotic Gaits
Authors:
John Daly,
Daniel Casper,
Muhammad Farooq,
Andrew James,
Ali Khan,
Phoenix Mulgrew,
Daniel Tyebkhan,
Bao Vo,
John Rieffel
Abstract:
Soft robotics aims to develop robots able to adapt their behavior across a wide range of unstructured and unknown environments. A critical challenge of soft robotic control is that nonlinear dynamics often result in complex behaviors hard to model and predict. Typically behaviors for mobile soft robots are discovered through empirical trial and error and hand-tuning. More recently, optimization al…
▽ More
Soft robotics aims to develop robots able to adapt their behavior across a wide range of unstructured and unknown environments. A critical challenge of soft robotic control is that nonlinear dynamics often result in complex behaviors hard to model and predict. Typically behaviors for mobile soft robots are discovered through empirical trial and error and hand-tuning. More recently, optimization algorithms such as Genetic Algorithms (GA) have been used to discover gaits, but these behaviors are often optimized for a single environment or terrain, and can be brittle to unplanned changes to terrain. In this paper we demonstrate how Quality Diversity Algorithms, which search of a range of high-performing behaviors, can produce repertoires of gaits that are robust to changing terrains. This robustness significantly out-performs that of gaits produced by a single objective optimization algorithm.
△ Less
Submitted 2 November, 2023;
originally announced November 2023.
-
Design of General Purpose Minimal-Auxiliary Ising Machines
Authors:
Isaac K. Martin,
Andrew G. Moore,
John T. Daly,
Jess J. Meyer,
Teresa M. Ranadive
Abstract:
Ising machines are a form of quantum-inspired processing-in-memory computer which has shown great promise for overcoming the limitations of traditional computing paradigms while operating at a fraction of the energy use. The process of designing Ising machines is known as the reverse Ising problem. Unfortunately, this problem is in general computationally intractable: it is a nonconvex mixed-integ…
▽ More
Ising machines are a form of quantum-inspired processing-in-memory computer which has shown great promise for overcoming the limitations of traditional computing paradigms while operating at a fraction of the energy use. The process of designing Ising machines is known as the reverse Ising problem. Unfortunately, this problem is in general computationally intractable: it is a nonconvex mixed-integer linear programming problem which cannot be naively brute-forced except in the simplest cases due to exponential scaling of runtime with number of spins. We prove new theoretical results which allow us to reduce the search space to one with quadratic scaling. We utilize this theory to develop general purpose algorithmic solutions to the reverse Ising problem. In particular, we demonstrate Ising formulations of 3-bit and 4-bit integer multiplication which use fewer total spins than previously known methods by a factor of more than three. Our results increase the practicality of implementing such circuits on modern Ising hardware, where spins are at a premium.
△ Less
Submitted 24 October, 2023;
originally announced October 2023.
-
Generating GPU Compiler Heuristics using Reinforcement Learning
Authors:
Ian Colbert,
Jake Daly,
Norm Rubin
Abstract:
GPU compilers are complex software programs with many optimizations specific to target hardware. These optimizations are often controlled by heuristics hand-designed by compiler experts using time- and resource-intensive processes. In this paper, we developed a GPU compiler autotuning framework that uses off-policy deep reinforcement learning to generate heuristics that improve the frame rates of…
▽ More
GPU compilers are complex software programs with many optimizations specific to target hardware. These optimizations are often controlled by heuristics hand-designed by compiler experts using time- and resource-intensive processes. In this paper, we developed a GPU compiler autotuning framework that uses off-policy deep reinforcement learning to generate heuristics that improve the frame rates of graphics applications. Furthermore, we demonstrate the resilience of these learned heuristics to frequent compiler updates by analyzing their stability across a year of code check-ins without retraining. We show that our machine learning-based compiler autotuning framework matches or surpasses the frame rates for 98% of graphics benchmarks with an average uplift of 1.6% up to 15.8%.
△ Less
Submitted 23 November, 2021;
originally announced November 2021.
-
UNICEF Guidance on AI for Children: Application to the Design of a Social Robot For and With Autistic Children
Authors:
Séverin Lemaignan,
Nigel Newbutt,
Louis Rice,
Joe Daly,
Vicky Charisi
Abstract:
For a period of three weeks in June 2021, we embedded a social robot (Softbank Pepper) in a Special Educational Needs (SEN) school, with a focus on supporting the well-being of autistic children. Our methodology to design and embed the robot among this vulnerable population follows a comprehensive participatory approach. We used the research project as a test-bed to demonstrate in a complex real-w…
▽ More
For a period of three weeks in June 2021, we embedded a social robot (Softbank Pepper) in a Special Educational Needs (SEN) school, with a focus on supporting the well-being of autistic children. Our methodology to design and embed the robot among this vulnerable population follows a comprehensive participatory approach. We used the research project as a test-bed to demonstrate in a complex real-world environment the importance and suitability of the nine UNICEF guidelines on AI for Children. The UNICEF guidelines on AI for Children closely align with several of the UN goals for sustainable development, and, as such, we report here our contribution to these goals.
△ Less
Submitted 27 August, 2021;
originally announced August 2021.
-
A Competitive Edge: Can FPGAs Beat GPUs at DCNN Inference Acceleration in Resource-Limited Edge Computing Applications?
Authors:
Ian Colbert,
Jake Daly,
Ken Kreutz-Delgado,
Srinjoy Das
Abstract:
When trained as generative models, Deep Learning algorithms have shown exceptional performance on tasks involving high dimensional data such as image denoising and super-resolution. In an increasingly connected world dominated by mobile and edge devices, there is surging demand for these algorithms to run locally on embedded platforms. FPGAs, by virtue of their reprogrammability and low-power char…
▽ More
When trained as generative models, Deep Learning algorithms have shown exceptional performance on tasks involving high dimensional data such as image denoising and super-resolution. In an increasingly connected world dominated by mobile and edge devices, there is surging demand for these algorithms to run locally on embedded platforms. FPGAs, by virtue of their reprogrammability and low-power characteristics, are ideal candidates for these edge computing applications. As such, we design a spatio-temporally parallelized hardware architecture capable of accelerating a deconvolution algorithm optimized for power-efficient inference on a resource-limited FPGA. We propose this FPGA-based accelerator to be used for Deconvolutional Neural Network (DCNN) inference in low-power edge computing applications. To this end, we develop methods that systematically exploit micro-architectural innovations, design space exploration, and statistical analysis. Using a Xilinx PYNQ-Z2 FPGA, we leverage our architecture to accelerate inference for two DCNNs trained on the MNIST and CelebA datasets using the Wasserstein GAN framework. On these networks, our FPGA design achieves a higher throughput to power ratio with lower run-to-run variation when compared to the NVIDIA Jetson TX1 edge computing GPU.
△ Less
Submitted 9 March, 2021; v1 submitted 30 January, 2021;
originally announced February 2021.
-
Multiplication with Fourier Optics Simulating 16-bit Modular Multiplication
Authors:
Abigail Timmel,
John Daly
Abstract:
This paper will describe a simulator developed by the authors to explore the design of Fourier transform based multiplication using optics. Then it will demonstrate an application to the problem of constructing an all-optical modular multiplication circuit. That circuit implements a novel approximate version of the Montgomery multiplication algorithm that enables the calculation to be performed en…
▽ More
This paper will describe a simulator developed by the authors to explore the design of Fourier transform based multiplication using optics. Then it will demonstrate an application to the problem of constructing an all-optical modular multiplication circuit. That circuit implements a novel approximate version of the Montgomery multiplication algorithm that enables the calculation to be performed entirely in the analog domain. The results will be used to corroborate the feasibility of scaling the design up to 16-bits without the need for analog to digital conversions at intermediate steps.
△ Less
Submitted 19 October, 2018; v1 submitted 3 January, 2018;
originally announced January 2018.
-
The neutron long counter NERO for studies of beta-delayed neutron emission in the r-process
Authors:
J. Pereira,
P. Hosmer,
G. Lorusso,
P. Santi,
A. Couture,
J. Daly,
M. Del Santo,
T. Elliot,
J. Goerres,
C. Herlitzius,
K. -L. Kratz,
L. O. Lamm,
H. Y. Lee,
F. Montes,
M. Ouellette,
E. Pellegrini,
P. Reeder,
H. Schatz,
F. Schertz,
L. Schnorrenberger,
K. Smith,
E. Stech,
E. Strandberg,
C. Ugalde,
M. Wiescher
, et al. (1 additional authors not shown)
Abstract:
The neutron long counter NERO was built at the National Superconducting Cyclotron Laboratory (NSCL), Michigan State University, for measuring beta-delayed neutron-emission probabilities. The detector was designed to work in conjunction with a beta-decay implantation station, so that beta decays and beta-delayed neutrons emitted from implanted nuclei can be measured simultaneously. The high efficie…
▽ More
The neutron long counter NERO was built at the National Superconducting Cyclotron Laboratory (NSCL), Michigan State University, for measuring beta-delayed neutron-emission probabilities. The detector was designed to work in conjunction with a beta-decay implantation station, so that beta decays and beta-delayed neutrons emitted from implanted nuclei can be measured simultaneously. The high efficiency of about 40%, for the range of energies of interest, along with the small background, are crucial for measuring beta-delayed neutron emission branchings for neutron-rich r-process nuclei produced as low intensity fragmentation beams in in-flight separator facilities.
△ Less
Submitted 28 July, 2010;
originally announced July 2010.
-
Age, Sex, and Genetic Architecture of Human Gene Expression in EBV Transformed Cell Lines
Authors:
Manuel A. Rivas,
Mark J. Daly,
Itsik Pe'er
Abstract:
Individual expression profiles from EBV transformed cell lines are an emerging resource for genomic investigation. In this study we characterize the effects of age, sex, and genetic variation on gene expression by surveying public datasets of such profiles. We establish that the expression space of cell lines maintains genetic as well as non-germline information, in an individual-specific and cr…
▽ More
Individual expression profiles from EBV transformed cell lines are an emerging resource for genomic investigation. In this study we characterize the effects of age, sex, and genetic variation on gene expression by surveying public datasets of such profiles. We establish that the expression space of cell lines maintains genetic as well as non-germline information, in an individual-specific and cross-tissue manner. Age of donor is associated with the expression of 949 genes in the derived cell line. Age-associated genes include over-representation of immune-related genes, specifically MHC Class I genes, a phenomenon that replicates across tissues and organisms. Sex associated genes in these cell lines include likely candidates, such as genes that escape X-inactivation,testis specific expressed genes, androgen and estrogen specific genes, but also gene families previously unknown to be sex associated such as common microRNA targets (MIR-490, V_ARP1_01, MIR-489). Finally, we report 494 transcripts whose expression levels are associated with a genetic variant in cis, overlap** and validating previous reports. Incorporating age in analysis of association facilitates additional discovery of trans-acting regulatory genetic variants. Our findings promote expression profiling of transformed cell lines as a vehicle for understanding cellular systems beyond the specific lines.
△ Less
Submitted 1 November, 2009;
originally announced November 2009.
