arXiv subjects

Matthew Cook

Publications and source records attributed to Matthew Cook.

32 records · Page 2Linked to original sources

arXiv2017

Deep Learning in the Automotive Industry: Applications and Tools

Deep Learning refers to a set of machine learning techniques that utilize neural networks with many hidden layers for tasks, such as image classification, speech recognition, language understanding. Deep learning has been proven to be very effective in these domains and is pervasively used by many Internet services. In this paper, we describe different automotive uses cases for deep learning in particular in the domain of computer vision. We surveys the current state-of-the-art in libraries, tools and infrastructures (e.\,g.\ GPUs and clouds) for implementing, training and deploying deep neural networks. We particularly focus on convolutional neural networks and computer vision use cases, such as the visual inspection process in manufacturing plants and the analysis of social media data. To train neural networks, curated and labeled datasets are essential. In particular, both the availability and scope of such datasets is typically very limited. A main contribution of this paper is the creation of an automotive dataset, that allows us to learn and automatically recognize different vehicle properties. We describe an end-to-end deep learning application utilizing a mobile app for data collection and process support, and an Amazon-based cloud backend for storage and training. For training we evaluate the use of cloud and on-premises infrastructures (including multiple GPUs) in conjunction with different neural network architectures and frameworks. We assess both the training times as well as the accuracy of the classifier. Finally, we demonstrate the effectiveness of the trained classifier in a real world setting during manufacturing process.

cs.LG↗

arXiv2017

A wake-sleep algorithm for recurrent, spiking neural networks

We investigate a recently proposed model for cortical computation which performs relational inference. It consists of several interconnected, structurally equivalent populations of leaky integrate-and-fire (LIF) neurons, which are trained in a self-organized fashion with spike-timing dependent plasticity (STDP). Despite its robust learning dynamics, the model is susceptible to a problem typical for recurrent networks which use a correlation based (Hebbian) learning rule: if trained with high learning rates, the recurrent connections can cause strong feedback loops in the network dynamics, which lead to the emergence of attractor states. This causes a strong reduction in the number of representable patterns and a decay in the inference ability of the network. As a solution, we introduce a conceptually very simple "wake-sleep" algorithm: during the wake phase, training is executed normally, while during the sleep phase, the network "dreams" samples from its generative model, which are induced by random input. This process allows us to activate the attractor states in the network, which can then be unlearned effectively by an anti-Hebbian mechanism. The algorithm allows us to increase learning rates up to a factor of ten while avoiding clustering, which allows the network to learn several times faster. Also for low learning rates, where clustering is not an issue, it improves convergence speed and reduces the final inference error.

cs.NE↗

arXiv2017

Optimised information gathering in smartphone users

Human activities from hunting to emailing are performed in a fractal-like scale invariant pattern. These patterns are considered efficient for hunting or foraging, but are they efficient for gathering information? Here we link the scale invariant pattern of inter-touch intervals on the smartphone to optimal strategies for information gathering. We recorded touchscreen touches in 65 individuals for a month and categorized the activity into checking for information vs. sharing content. For both categories, the inter-touch intervals were well described by power-law fits spanning 5 orders of magnitude, from 1 s to several hours. The power-law exponent typically found for checking was 1.5 and for generating it was 1.3. Next, by using computer simulations we addressed whether the checking pattern was efficient - in terms of minimizing futile attempts yielding no new information. We find that the best performing power law exponent depends on the duration of the assessment and the exponent of 1.5 was the most efficient in the short-term i.e. in the few minutes range. Finally, we addressed whether how people generated and shared content was in tune with the checking pattern. We assumed that the unchecked posts must be minimized for maximal efficiency and according to our analysis the most efficient temporal pattern to share content was the exponent of 1.3 - which was also the pattern displayed by the smartphone users. The behavioral organization for content generation is different from content consumption across time scales. We propose that this difference is a signature of optimal behavior and the short-term assessments used in modern human actions.

physics.soc-ph↗

arXiv2016

Room Temperature Formation of Carbon Onions via Ultrasonic Agitation of MoS2 in Isopropanol

Ultrasonic agitation is a proven method for breaking down layered materials such as MoS2 into single or few layer nanoparticles. In this experiment, MoS2 powder is sonicated in isopropanol for an extended period of time in an attempt to create particles of the smallest possible size. As expected, the process yielded a significant quantity of nanoscale MoS2 in the form of finite layer sheets with lateral dimensions as small as a few tens of nanometers. Although no evidence was found to indicate a larger the longer sonication times resulted in a significant increase in yield of single layer MoS2, the increased sonication did result in the formation of several types of carbon allotropes in addition to the sheets of MoS2. These carbon structures appear to originate from the breakdown of the isopropanol and consist of finite layer graphite platelets as well as a large number of multi-walled fullerenes, also known as carbon onions. Both the finite layer graphite and MoS2 nanoplatelets were both found to be heavily decorated with carbon onions. However, isolated clusters of carbon onions could also be found. Our results show that liquid exfoliation of MoS2 is not only useful for forming finite layer MoS2, but also creating carbon onions at room temperature as well.

cond-mat.mes-hall↗

arXiv2016

Learning and Inferring Relations in Cortical Networks

A pressing scientific challenge is to understand how brains work. Of particular interest is the neocortex,the part of the brain that is especially large in humans, capable of handling a wide variety of tasks including visual, auditory, language, motor, and abstract processing. These functionalities are processed in different self-organized regions of the neocortical sheet, and yet the anatomical structure carrying out the processing is relatively uniform across the sheet. We are at a loss to explain, simulate, or understand such a multi-functional homogeneous sheet-like computational structure - we do not have computational models which work in this way. Here we present an important step towards developing such models: we show how uniform modules of excitatory and inhibitory neurons can be connected bidirectionally in a network that, when exposed to input in the form of population codes, learns the input encodings as well as the relationships between the inputs. STDP learning rules lead the modules to self-organize into a relational network, which is able to infer missing inputs,restore noisy signals, decide between conflicting inputs, and combine cues to improve estimates. These networks show that it is possible for a homogeneous network of spiking units to self-organize so as to provide meaningful processing of its inputs. If such networks can be scaled up, they could provide an initial computational model relevant to the large scale anatomy of the neocortex.

cs.NE↗

arXiv2016

Adaptive coherence estimator (ACE) for explosive hazard detection using wideband electromagnetic induction (WEMI)

The adaptive coherence estimator (ACE) estimates the squared cosine of the angle between a known target vector and a sample vector in a whitened coordinate space. The space is whitened according to an estimation of the background statistics, which directly effects the performance of the statistic as a target detector. In this paper, the ACE detection statistic is used to detect buried explosive hazards with data from a Wideband Electromagnetic Induction (WEMI) sensor. Target signatures are based on a dictionary defined using a Discrete Spectrum of Relaxation Frequencies (DSRF) model. Results are summarized as a receiver operator curve (ROC) and compared to other leading methods.

cs.CV↗

arXiv2016

Buried object detection using handheld WEMI with task-driven extended functions of multiple instances

Many effective supervised discriminative dictionary learning methods have been developed in the literature. However, when training these algorithms, precise ground-truth of the training data is required to provide very accurate point-wise labels. Yet, in many applications, accurate labels are not always feasible. This is especially true in the case of buried object detection in which the size of the objects are not consistent. In this paper, a new multiple instance dictionary learning algorithm for detecting buried objects using a handheld WEMI sensor is detailed. The new algorithm, Task Driven Extended Functions of Multiple Instances, can overcome data that does not have very precise point-wise labels and still learn a highly discriminative dictionary. Results are presented and discussed on measured WEMI data.

cs.CV↗

arXiv2016

TED: A Tolerant Edit Distance for Segmentation Evaluation

In this paper, we present a novel error measure to compare a segmentation against ground truth. This measure, which we call Tolerant Edit Distance (TED), is motivated by two observations: (1) Some errors, like small boundary shifts, are tolerable in practice. Which errors are tolerable is application dependent and should be a parameter of the measure. (2) Non-tolerable errors have to be corrected manually. The time needed to do so should be reflected by the error measure. Using integer linear programming, the TED finds the minimal weighted sum of split and merge errors exceeding a given tolerance criterion, and thus provides a time-to-fix estimate. In contrast to commonly used measures like Rand index or variation of information, the TED (1) does not count small, but tolerable, differences, (2) provides intuitive numbers, (3) gives a time-to-fix estimate, and (4) can localize and classify the type of errors. By supporting both isotropic and anisotropic volumes and having a flexible tolerance criterion, the TED can be adapted to different requirements. On example segmentations for 3D neuron segmentation, we demonstrate that the TED is capable of counting topological errors, while ignoring small boundary shifts.

cs.CV↗

arXiv2015

A Cellular Automaton for Blocking Queen Games

We show that the winning positions of a certain type of two-player game form interesting patterns which often defy analysis, yet can be computed by a cellular automaton. The game, known as {\em Blocking Wythoff Nim}, consists of moving a queen as in chess, but always towards (0,0), and it may not be moved to any of $k-1$ temporarily "blocked" positions specified on the previous turn by the other player. The game ends when a player wins by blocking all possible moves of the other player. The value of $k$ is a parameter that defines the game, and the pattern of winning positions can be very sensitive to $k$. As $k$ becomes large, parts of the pattern of winning positions converge to recurring chaotic patterns that are independent of $k$. The patterns for large $k$ display an unprecedented amount of self-organization at many scales, and here we attempt to describe the self-organized structure that appears.

math.CO↗

arXiv2013

Proceedings Machines, Computations and Universality 2013

This volume contains the papers presented at the 6th conference on Machines, Computations and Universality (MCU 2013). MCU 2013 was held in Zurich, Switzerland, September 9-11, 2013. The MCU series began in Paris in 1995 and has since been concerned with gaining a deeper understanding of computation through the study of models of general purpose computation. This volume continues in this tradition and includes new simple universal models of computation, and other results that clarify the relationships between models.

cs.FL↗

arXiv2011

Multi-Hypothesis CRF-Segmentation of Neural Tissue in Anisotropic EM Volumes

We present an approach for the joint segmentation and grouping of similar components in anisotropic 3D image data and use it to segment neural tissue in serial sections electron microscopy (EM) images. We first construct a nested set of neuron segmentation hypotheses for each slice. A conditional random field (CRF) then allows us to evaluate both the compatibility of a specific segmentation and a specific inter-slice assignment of neuron candidates with the underlying observations. The model is solved optimally for an entire image stack simultaneously using integer linear programming (ILP), which yields the maximum a posteriori solution in amortized linear time in the number of slices. We evaluate the performance of our approach on an annotated sample of the Drosophila larva neuropil and show that the consideration of different segmentation hypotheses in each slice leads to a significant improvement in the segmentation and assignment accuracy.

cs.CV↗

arXiv2010

Temperature 1 Self-Assembly: Deterministic Assembly in 3D and Probabilistic Assembly in 2D

We investigate the power of the Wang tile self-assembly model at temperature 1, a threshold value that permits attachment between any two tiles that share even a single bond. When restricted to deterministic assembly in the plane, no temperature 1 assembly system has been shown to build a shape with a tile complexity smaller than the diameter of the shape. In contrast, we show that temperature 1 self-assembly in 3 dimensions, even when growth is restricted to at most 1 step into the third dimension, is capable of simulating a large class of temperature 2 systems, in turn permitting the simulation of arbitrary Turing machines and the assembly of $n\times n$ squares in near optimal $O(\log n)$ tile complexity. Further, we consider temperature 1 probabilistic assembly in 2D, and show that with a logarithmic scale up of tile complexity and shape scale, the same general class of temperature $τ=2$ systems can be simulated with high probability, yielding Turing machine simulation and $O(\log^2 n)$ assembly of $n\times n$ squares with high probability. Our results show a sharp contrast in achievable tile complexity at temperature 1 if either growth into the third dimension or a small probability of error are permitted. Motivated by applications in nanotechnology and molecular computing, and the plausibility of implementing 3 dimensional self-assembly systems, our techniques may provide the needed power of temperature 2 systems, while at the same time avoiding the experimental challenges faced by those systems.

cs.CC↗

arXiv2009

A Concrete View of Rule 110 Computation

Rule 110 is a cellular automaton that performs repeated simultaneous updates of an infinite row of binary values. The values are updated in the following way: 0s are changed to 1s at all positions where the value to the right is a 1, while 1s are changed to 0s at all positions where the values to the left and right are both 1. Though trivial to define, the behavior exhibited by Rule 110 is surprisingly intricate, and in (Cook, 2004) we showed that it is capable of emulating the activity of a Turing machine by encoding the Turing machine and its tape into a repeating left pattern, a central pattern, and a repeating right pattern, which Rule 110 then acts on. In this paper we provide an explicit compiler for converting a Turing machine into a Rule 110 initial state, and we present a general approach for proving that such constructions will work as intended. The simulation was originally assumed to require exponential time, but surprising results of Neary and Woods (2006) have shown that in fact, only polynomial time is required. We use the methods of Neary and Woods to exhibit a direct simulation of a Turing machine by a tag system in polynomial time.

cs.CC↗

arXiv2000

Tournament Sequences and Meeussen Sequences

A "tournament sequence" is an increasing sequence of positive integers (t_1,t_2,...) such that t_1=1 and t_{i+1} <= 2 t_i. A "Meeussen sequence" is an increasing sequence of positive integers (m_1,m_2,...) such that m_1=1, every nonnegative integer is the sum of a subset of the {m_i}, and each integer m_i-1 is the sum of a unique such subset. We show that these two properties are isomorphic. That is, we present a bijection between tournament and Meeussen sequences which respects the natural tree structure on each set. We also present an efficient technique for counting the number of tournament sequences of length n, and discuss the asymptotic growth of this number. The counting technique we introduce is suitable for application to other well-behaved counting problems of the same sort where a closed form or generating function cannot be found.

math.CO↗