Convolutional Neural Network-based place recognition

Recently Convolutional Neural Networks (CNNs) have been shown to achieve state-of-the-art performance on various classification tasks. In this paper, we present for the first time a place recognition technique based on CNN models, by combining the powerful features learnt by CNNs with a spatial and sequential filter. Applying the system to a 70 km benchmark place recognition dataset we achieve a 75% increase in recall at 100% precision, significantly outperforming all previous state of the art techniques. We also conduct a comprehensive performance comparison of the utility of features from all 21 layers for place recognition, both for the benchmark dataset and for a second dataset with more significant viewpoint changes.

Prediction of GMA welding characteristic parameter by artificial neural network system

Nowadays, demand for automated Gas metal arc welding (GMAW) is growing and consequently need for intelligent systems is increased to ensure the accuracy of the procedure. To date, welding pool geometry has been the most used factor in quality assessment of intelligent welding systems. But, it has recently been found that Mahalanobis Distance (MD) not only can be used for this purpose but also is more efficient. In the present paper, Artificial Neural Networks (ANN) has been used for prediction of MD parameter. However, advantages and disadvantages of other methods have been discussed. The Levenberg–Marquardt algorithm was found to be the most effective algorithm for GMAW process. It is known that the number of neurons plays an important role in optimal network design. In this work, using trial and error method, it has been found that 30 is the optimal number of neurons. The model has been investigated with different number of layers in Multilayer Perceptron (MLP) architecture and has been shown that for the aim of this work the optimal result is obtained when using MLP with one layer. Robustness of the system has been evaluated by adding noise into the input data and studying the effect of the noise in prediction capability of the network. The experiments for this study were conducted in an automated GMAW setup that was integrated with data acquisition system and prepared in a laboratory for welding of steel plate with 12 mm in thickness. The accuracy of the network was evaluated by Root Mean Squared (RMS) error between the measured and the estimated values. The low error value (about 0.008) reflects the good accuracy of the model. Also the comparison of the predicted results by ANN and the test data set showed very good agreement that reveals the predictive power of the model. Therefore, the ANN model offered in here for GMA welding process can be used effectively for prediction goals.

Autonomous helicopter hover using an artificial neural network

Details the developments to date of an unmanned air vehicle (UAV) based on a standard size 60 model helicopter. The design goal is to have the helicopter achieve stable hover with the aid of an INS and stereo vision. The focus of the paper is on the development of an artificial neural network (ANN) that makes use of only the INS data to generate hover commands, which are used to directly manipulate the flight servos. Current results show that networks incorporating some form of recurrency (state history) offer little advantage over those without. At this stage, the ANN has partially maintained periods of hover even with misaligned sensors.

Artificial Neural Network Based Automatic Emergency Landing Site Selection for UAVs and Highly Automated Aircraft

In this report an artificial neural network (ANN) based automated emergency landing site selection system for unmanned aerial vehicle (UAV) and general aviation (GA) is described. The system aims increase safety of UAV operation by emulating pilot decision making in emergency landing scenarios using an ANN to select a safe landing site from available candidates. The strength of an ANN to model complex input relationships makes it a perfect system to handle the multicriteria decision making (MCDM) process of emergency landing site selection. The ANN operates by identifying the more favorable of two landing sites when provided with an input vector derived from both landing site's parameters, the aircraft's current state and wind measurements. The system consists of a feed forward ANN, a pre-processor class which produces ANN input vectors and a class in charge of creating a ranking of landing site candidates using the ANN. The system was successfully implemented in C++ using the FANN C++ library and ROS. Results obtained from ANN training and simulations using randomly generated landing sites by a site detection simulator data verify the feasibility of an ANN based automated emergency landing site selection system.

Intelligent prognostics of machinery health utilising suspended condition monitoring data

The ability to forecast machinery failure is vital to reducing maintenance costs, operation downtime and safety hazards. Recent advances in condition monitoring technologies have given rise to a number of prognostic models for forecasting machinery health based on condition data. Although these models have aided the advancement of the discipline, they have made only a limited contribution to developing an effective machinery health prognostic system. The literature review indicates that there is not yet a prognostic model that directly models and fully utilises suspended condition histories (which are very common in practice since organisations rarely allow their assets to run to failure); that effectively integrates population characteristics into prognostics for longer-range prediction in a probabilistic sense; which deduces the non-linear relationship between measured condition data and actual asset health; and which involves minimal assumptions and requirements. This work presents a novel approach to addressing the above-mentioned challenges. The proposed model consists of a feed-forward neural network, the training targets of which are asset survival probabilities estimated using a variation of the Kaplan-Meier estimator and a degradation-based failure probability density estimator. The adapted Kaplan-Meier estimator is able to model the actual survival status of individual failed units and estimate the survival probability of individual suspended units. The degradation-based failure probability density estimator, on the other hand, extracts population characteristics and computes conditional reliability from available condition histories instead of from reliability data. The estimated survival probability and the relevant condition histories are respectively presented as “training target” and “training input” to the neural network. The trained network is capable of estimating the future survival curve of a unit when a series of condition indices are inputted. Although the concept proposed may be applied to the prognosis of various machine components, rolling element bearings were chosen as the research object because rolling element bearing failure is one of the foremost causes of machinery breakdowns. Computer simulated and industry case study data were used to compare the prognostic performance of the proposed model and four control models, namely: two feed-forward neural networks with the same training function and structure as the proposed model, but neglected suspended histories; a time series prediction recurrent neural network; and a traditional Weibull distribution model. The results support the assertion that the proposed model performs better than the other four models and that it produces adaptive prediction outputs with useful representation of survival probabilities. This work presents a compelling concept for non-parametric data-driven prognosis, and for utilising available asset condition information more fully and accurately. It demonstrates that machinery health can indeed be forecasted. The proposed prognostic technique, together with ongoing advances in sensors and data-fusion techniques, and increasingly comprehensive databases of asset condition data, holds the promise for increased asset availability, maintenance cost effectiveness, operational safety and – ultimately – organisation competitiveness.

The sample complexity of pattern classification with neural networks: the size of the weights is more important than the size of the network

Sample complexity results from computational learning theory, when applied to neural network learning for pattern classification problems, suggest that for good generalization performance the number of training examples should grow at least linearly with the number of adjustable parameters in the network. Results in this paper show that if a large neural network is used for a pattern classification problem and the learning algorithm finds a network with small weights that has small squared error on the training patterns, then the generalization performance depends on the size of the weights rather than the number of weights. For example, consider a two-layer feedforward network of sigmoid units, in which the sum of the magnitudes of the weights associated with each unit is bounded by A and the input dimension is n. We show that the misclassification probability is no more than a certain error estimate (that is related to squared error on the training set) plus A3 √((log n)/m) (ignoring log A and log m factors), where m is the number of training patterns. This may explain the generalization performance of neural networks, particularly when the number of training examples is considerably smaller than the number of weights. It also supports heuristics (such as weight decay and early stopping) that attempt to keep the weights small during training. The proof techniques appear to be useful for the analysis of other pattern classifiers: when the input domain is a totally bounded metric space, we use the same approach to give upper bounds on misclassification probability for classifiers with decision boundaries that are far from the training examples.

Using strategic movement to calibrate a neural compass : a spiking network for tracking head direction in rats and robots

The head direction (HD) system in mammals contains neurons that fire to represent the direction the animal is facing in its environment. The ability of these cells to reliably track head direction even after the removal of external sensory cues implies that the HD system is calibrated to function effectively using just internal (proprioceptive and vestibular) inputs. Rat pups and other infant mammals display stereotypical warm-up movements prior to locomotion in novel environments, and similar warm-up movements are seen in adult mammals with certain brain lesion-induced motor impairments. In this study we propose that synaptic learning mechanisms, in conjunction with appropriate movement strategies based on warm-up movements, can calibrate the HD system so that it functions effectively even in darkness. To examine the link between physical embodiment and neural control, and to determine that the system is robust to real-world phenomena, we implemented the synaptic mechanisms in a spiking neural network and tested it on a mobile robot platform. Results show that the combination of the synaptic learning mechanisms and warm-up movements are able to reliably calibrate the HD system so that it accurately tracks real-world head direction, and that calibration breaks down in systematic ways if certain movements are omitted. This work confirms that targeted, embodied behaviour can be used to calibrate neural systems, demonstrates that ‘grounding’ of modeled biological processes in the real world can reveal underlying functional principles (supporting the importance of robotics to biology), and proposes a functional role for stereotypical behaviours seen in infant mammals and those animals with certain motor deficits. We conjecture that these calibration principles may extend to the calibration of other neural systems involved in motion tracking and the representation of space, such as grid cells in entorhinal cortex.

Predicting project performance through neural networks

Successful project delivery of construction projects depends on many factors. With regard to the construction of a facility, selecting a competent contractor for the job is paramount. As such, various approaches have been advanced to facilitate tender award decisions. Essentially, this type of decision involves the prediction of a bidderÕs performance based on information available at the tender stage. A neural network based prediction model was developed and presented in this paper. Project data for the study were obtained from the Hong Kong Housing Department. Information from the tender reports was used as input variables and performance records of the successful bidder during construction were used as output variables. It was found that the networks for the prediction of performance scores for Works gave the highest hit rate. In addition, the two most sensitive input variables toward such prediction are ‘‘Difference between Estimate’’ and ‘‘Difference between the next closest bid’’. Both input variables are price related, thus suggesting the importance of tender sufficiency for the assurance of quality production.

GYAN: A methodology for rule extraction from artificial neural networks

Artificial neural network (ANN) learning methods provide a robust and non-linear approach to approximating the target function for many classification, regression and clustering problems. ANNs have demonstrated good predictive performance in a wide variety of practical problems. However, there are strong arguments as to why ANNs are not sufficient for the general representation of knowledge. The arguments are the poor comprehensibility of the learned ANN, and the inability to represent explanation structures. The overall objective of this thesis is to address these issues by: (1) explanation of the decision process in ANNs in the form of symbolic rules (predicate rules with variables); and (2) provision of explanatory capability by mapping the general conceptual knowledge that is learned by the neural networks into a knowledge base to be used in a rule-based reasoning system. A multi-stage methodology GYAN is developed and evaluated for the task of extracting knowledge from the trained ANNs. The extracted knowledge is represented in the form of restricted first-order logic rules, and subsequently allows user interaction by interfacing with a knowledge based reasoner. The performance of GYAN is demonstrated using a number of real world and artificial data sets. The empirical results demonstrate that: (1) an equivalent symbolic interpretation is derived describing the overall behaviour of the ANN with high accuracy and fidelity, and (2) a concise explanation is given (in terms of rules, facts and predicates activated in a reasoning episode) as to why a particular instance is being classified into a certain category.

Aerial image analysis using spiking neural networks with application to power line corridor monitoring

Trees, shrubs and other vegetation are of continued importance to the environment and our daily life. They provide shade around our roads and houses, offer a habitat for birds and wildlife, and absorb air pollutants. However, vegetation touching power lines is a risk to public safety and the environment, and one of the main causes of power supply problems. Vegetation management, which includes tree trimming and vegetation control, is a significant cost component of the maintenance of electrical infrastructure. For example, Ergon Energy, the Australia’s largest geographic footprint energy distributor, currently spends over $80 million a year inspecting and managing vegetation that encroach on power line assets. Currently, most vegetation management programs for distribution systems are calendar-based ground patrol. However, calendar-based inspection by linesman is labour-intensive, time consuming and expensive. It also results in some zones being trimmed more frequently than needed and others not cut often enough. Moreover, it’s seldom practicable to measure all the plants around power line corridors by field methods. Remote sensing data captured from airborne sensors has great potential in assisting vegetation management in power line corridors. This thesis presented a comprehensive study on using spiking neural networks in a specific image analysis application: power line corridor monitoring. Theoretically, the thesis focuses on a biologically inspired spiking cortical model: pulse coupled neural network (PCNN). The original PCNN model was simplified in order to better analyze the pulse dynamics and control the performance. Some new and effective algorithms were developed based on the proposed spiking cortical model for object detection, image segmentation and invariant feature extraction. The developed algorithms were evaluated in a number of experiments using real image data collected from our flight trails. The experimental results demonstrated the effectiveness and advantages of spiking neural networks in image processing tasks. Operationally, the knowledge gained from this research project offers a good reference to our industry partner (i.e. Ergon Energy) and other energy utilities who wants to improve their vegetation management activities. The novel approaches described in this thesis showed the potential of using the cutting edge sensor technologies and intelligent computing techniques in improve power line corridor monitoring. The lessons learnt from this project are also expected to increase the confidence of energy companies to move from traditional vegetation management strategy to a more automated, accurate and cost-effective solution using aerial remote sensing techniques.

Frequency response function based structural damage detection using artificial neural networks

Damage detection in structures has become increasingly important in recent years. While a number of damage detection and localization methods have been proposed, few attempts have been made to explore the structure damage with frequency response functions (FRFs). This paper illustrates the damage identification and condition assessment of a beam structure using a new frequency response functions (FRFs) based damage index and Artificial Neural Networks (ANNs). In practice, usage of all available FRF data as an input to artificial neural networks makes the training and convergence impossible. Therefore one of the data reduction techniques Principal Component Analysis (PCA) is introduced in the algorithm. In the proposed procedure, a large set of FRFs are divided into sub-sets in order to find the damage indices for different frequency points of different damage scenarios. The basic idea of this method is to establish features of damaged structure using FRFs from different measurement points of different sub-sets of intact structure. Then using these features, damage indices of different damage cases of the structure are identified after reconstructing of available FRF data using PCA. The obtained damage indices corresponding to different damage locations and severities are introduced as input variable to developed artificial neural networks. Finally, the effectiveness of the proposed method is illustrated and validated by using the finite element modal of a beam structure. The illustrated results show that the PCA based damage index is suitable and effective for structural damage detection and condition assessment of building structures.

Quo Vadis? Reliable and Practical Rule Extraction from Neural Networks

Rule extraction from neural network algorithms have been investigated for two decades and there have been significant applications. Despite this level of success, rule extraction from neural network methods are generally not part of data mining tools, and a significant commercial breakthrough may still be some time away. This paper briefly reviews the state-of-the-art and points to some of the obstacles, namely a lack of evaluation techniques in experiments and larger benchmark data sets. A significant new development is the view that rule extraction from neural networks is an interactive process which actively involves the user. This leads to the application of assessment and evaluation techniques from information retrieval which may lead to a range of new methods.