Efficient Nearest Neighbor Classification Using a Cascade of Approximate Similarity Measures

Nearest neighbor classification using shape context can yield highly accurate results in a number of recognition problems. Unfortunately, the approach can be too slow for practical applications, and thus approximation strategies are needed to make shape context practical. This paper proposes a method for efficient and accurate nearest neighbor classification in non-Euclidean spaces, such as the space induced by the shape context measure. First, a method is introduced for constructing a Euclidean embedding that is optimized for nearest neighbor classification accuracy. Using that embedding, multiple approximations of the underlying non-Euclidean similarity measure are obtained, at different levels of accuracy and efficiency. The approximations are automatically combined to form a cascade classifier, which applies the slower approximations only to the hardest cases. Unlike typical cascade-of-classifiers approaches, that are applied to binary classification problems, our method constructs a cascade for a multiclass problem. Experiments with a standard shape data set indicate that a two-to-three order of magnitude speed up is gained over the standard shape context classifier, with minimal losses in classification accuracy.

Face Identification by a Cascade of Rejection Classifiers

Nearest neighbor search is commonly employed in face recognition but it does not scale well to large dataset sizes. A strategy to combine rejection classifiers into a cascade for face identification is proposed in this paper. A rejection classifier for a pair of classes is defined to reject at least one of the classes with high confidence. These rejection classifiers are able to share discriminants in feature space and at the same time have high confidence in the rejection decision. In the face identification problem, it is possible that a pair of known individual faces are very dissimilar. It is very unlikely that both of them are close to an unknown face in the feature space. Hence, only one of them needs to be considered. Using a cascade structure of rejection classifiers, the scope of nearest neighbor search can be reduced significantly. Experiments on Face Recognition Grand Challenge (FRGC) version 1 data demonstrate that the proposed method achieves significant speed up and an accuracy comparable with the brute force Nearest Neighbor method. In addition, a graph cut based clustering technique is employed to demonstrate that the pairwise separability of these rejection classifiers is capable of semantic grouping.

Learning Embeddings for Indexing, Retrieval, and Classification, with Applications to Object and Shape Recognition in Image Databases

Nearest neighbor retrieval is the task of identifying, given a database of objects and a query object, the objects in the database that are the most similar to the query. Retrieving nearest neighbors is a necessary component of many practical applications, in fields as diverse as computer vision, pattern recognition, multimedia databases, bioinformatics, and computer networks. At the same time, finding nearest neighbors accurately and efficiently can be challenging, especially when the database contains a large number of objects, and when the underlying distance measure is computationally expensive. This thesis proposes new methods for improving the efficiency and accuracy of nearest neighbor retrieval and classification in spaces with computationally expensive distance measures. The proposed methods are domain-independent, and can be applied in arbitrary spaces, including non-Euclidean and non-metric spaces. In this thesis particular emphasis is given to computer vision applications related to object and shape recognition, where expensive non-Euclidean distance measures are often needed to achieve high accuracy. The first contribution of this thesis is the BoostMap algorithm for embedding arbitrary spaces into a vector space with a computationally efficient distance measure. Using this approach, an approximate set of nearest neighbors can be retrieved efficiently - often orders of magnitude faster than retrieval using the exact distance measure in the original space. The BoostMap algorithm has two key distinguishing features with respect to existing embedding methods. First, embedding construction explicitly maximizes the amount of nearest neighbor information preserved by the embedding. Second, embedding construction is treated as a machine learning problem, in contrast to existing methods that are based on geometric considerations. The second contribution is a method for constructing query-sensitive distance measures for the purposes of nearest neighbor retrieval and classification. In high-dimensional spaces, query-sensitive distance measures allow for automatic selection of the dimensions that are the most informative for each specific query object. It is shown theoretically and experimentally that query-sensitivity increases the modeling power of embeddings, allowing embeddings to capture a larger amount of the nearest neighbor structure of the original space. The third contribution is a method for speeding up nearest neighbor classification by combining multiple embedding-based nearest neighbor classifiers in a cascade. In a cascade, computationally efficient classifiers are used to quickly classify easy cases, and classifiers that are more computationally expensive and also more accurate are only applied to objects that are harder to classify. An interesting property of the proposed cascade method is that, under certain conditions, classification time actually decreases as the size of the database increases, a behavior that is in stark contrast to the behavior of typical nearest neighbor classification systems. The proposed methods are evaluated experimentally in several different applications: hand shape recognition, off-line character recognition, online character recognition, and efficient retrieval of time series. In all datasets, the proposed methods lead to significant improvements in accuracy and efficiency compared to existing state-of-the-art methods. In some datasets, the general-purpose methods introduced in this thesis even outperform domain-specific methods that have been custom-designed for such datasets.

CEDI: A Neural Model of Colour Vision, with Applications to Image Processing and Classification

Air Force Office of Scientific Research (F49620-01-1-0423); National Geospatial-Intelligence Agency (NMA 201-01-1-2016); National Science Foundation (SBE-035437, DEG-0221680); Office of Naval Research (N00014-01-1-0624)

ARTSCENE: A Neural System for Natural Scene Classification

How do humans rapidly recognize a scene? How can neural models capture this biological competence to achieve state-of-the-art scene classification? The ARTSCENE neural system classifies natural scene photographs by using multiple spatial scales to efficiently accumulate evidence for gist and texture. ARTSCENE embodies a coarse-to-fine Texture Size Ranking Principle whereby spatial attention processes multiple scales of scenic information, ranging from global gist to local properties of textures. The model can incrementally learn and predict scene identity by gist information alone and can improve performance through selective attention to scenic textures of progressively smaller size. ARTSCENE discriminates 4 landscape scene categories (coast, forest, mountain and countryside) with up to 91.58% correct on a test set, outperforms alternative models in the literature which use biologically implausible computations, and outperforms component systems that use either gist or texture information alone. Model simulations also show that adjacent textures form higher-order features that are also informative for scene recognition.

Fusion Artmap: A Neural Network Architecture for Multi-Channel Data Fusion and Classification

Fusion ARTMAP is a self-organizing neural network architecture for multi-channel, or multi-sensor, data fusion. Single-channel Fusion ARTMAP is functionally equivalent to Fuzzy ART during unsupervised learning and to Fuzzy ARTMAP during supervised learning. The network has a symmetric organization such that each channel can be dynamically configured to serve as either a data input or a teaching input to the system. An ART module forms a compressed recognition code within each channel. These codes, in turn, become inputs to a single ART system that organizes the global recognition code. When a predictive error occurs, a process called paraellel match tracking simultaneously raises vigilances in multiple ART modules until reset is triggered in one of them. Parallel match tracking hereby resets only that portion of the recognition code with the poorest match, or minimum predictive confidence. This internally controlled selective reset process is a type of credit assignment that creates a parsimoniously connected learned network. Fusion ARTMAP's multi-channel coding is illustrated by simulations of the Quadruped Mammal database.

Fusion ARTMAP: An Adaptive Fuzzy Network for Multi-Channel Classification

Fusion ARTMAP is a self-organizing neural network architecture for multi-channel, or multi-sensor, data fusion. Fusion ARTMAP generalizes the fuzzy ARTMAP architecture in order to adaptively classify multi-channel data. The network has a symmetric organization such that each channel can be dynamically configured to serve as either a data input or a teaching input to the system. An ART module forms a compressed recognition code within each channel. These codes, in turn, beco1ne inputs to a single ART system that organizes the global recognition code. When a predictive error occurs, a process called parallel match tracking simultaneously raises vigilances in multiple ART modules until reset is triggered in one of thmn. Parallel match tracking hereby resets only that portion of the recognition code with the poorest match, or minimum predictive confidence. This internally controlled selective reset process is a type of credit assignment that creates a parsimoniously connected learned network.

ARTMAP: Supervised Real-Time Learning and Classification of Nonstationary Data by a Self-Organizing Neural Network

This article introduces a new neural network architecture, called ARTMAP, that autonomously learns to classify arbitrarily many, arbitrarily ordered vectors into recognition categories based on predictive success. This supervised learning system is built up from a pair of Adaptive Resonance Theory modules (ARTa and ARTb) that are capable of self-organizing stable recognition categories in response to arbitrary sequences of input patterns. During training trials, the ARTa module receives a stream {a^(p)} of input patterns, and ARTb receives a stream {b^(p)} of input patterns, where b^(p) is the correct prediction given a^(p). These ART modules are linked by an associative learning network and an internal controller that ensures autonomous system operation in real time. During test trials, the remaining patterns a^(p) are presented without b^(p), and their predictions at ARTb are compared with b^(p). Tested on a benchmark machine learning database in both on-line and off-line simulations, the ARTMAP system learns orders of magnitude more quickly, efficiently, and accurately than alternative algorithms, and achieves 100% accuracy after training on less than half the input patterns in the database. It achieves these properties by using an internal controller that conjointly maximizes predictive generalization and minimizes predictive error by linking predictive success to category size on a trial-by-trial basis, using only local operations. This computation increases the vigilance parameter ρa of ARTa by the minimal amount needed to correct a predictive error at ARTb· Parameter ρa calibrates the minimum confidence that ARTa must have in a category, or hypothesis, activated by an input a^(p) in order for ARTa to accept that category, rather than search for a better one through an automatically controlled process of hypothesis testing. Parameter ρa is compared with the degree of match between a^(p) and the top-down learned expectation, or prototype, that is read-out subsequent to activation of an ARTa category. Search occurs if the degree of match is less than ρa. ARTMAP is hereby a type of self-organizing expert system that calibrates the selectivity of its hypotheses based upon predictive success. As a result, rare but important events can be quickly and sharply distinguished even if they are similar to frequent events with different consequences. Between input trials ρa relaxes to a baseline vigilance pa When ρa is large, the system runs in a conservative mode, wherein predictions are made only if the system is confident of the outcome. Very few false-alarm errors then occur at any stage of learning, yet the system reaches asymptote with no loss of speed. Because ARTMAP learning is self stabilizing, it can continue learning one or more databases, without degrading its corpus of memories, until its full memory capacity is utilized.

Fuzzy ART: An Adaptive Resonance Algorithm for Rapid, Stable Classification of Analog Patterns

The Fuzzy ART system introduced herein incorporates computations from fuzzy set theory into ART 1. For example, the intersection (n) operator used in ART 1 learning is replaced by the MIN operator (A) of fuzzy set theory. Fuzzy ART reduces to ART 1 in response to binary input vectors, but can also learn stable categories in response to analog input vectors. In particular, the MIN operator reduces to the intersection operator in the binary case. Learning is stable because all adaptive weights can only decrease in time. A preprocessing step, called complement coding, uses on-cell and off-cell responses to prevent category proliferation. Complement coding normalizes input vectors while preserving the amplitudes of individual feature activations.

Construction of Neural Network Classification Expert Systems Using Switching Theory Algorithms

A new family of neural network architectures is presented. This family of architectures solves the problem of constructing and training minimal neural network classification expert systems by using switching theory. The primary insight that leads to the use of switching theory is that the problem of minimizing the number of rules and the number of IF statements (antecedents) per rule in a neural network expert system can be recast into the problem of minimizing the number of digital gates and the number of connections between digital gates in a Very Large Scale Integrated (VLSI) circuit. The rules that the neural network generates to perform a task are readily extractable from the network's weights and topology. Analysis and simulations on the Mushroom database illustrate the system's performance.

Multi-sensor classification of tennis strokes

In this work, we investigate tennis stroke recognition using a single inertial measuring unit attached to a player’s forearm during a competitive match. This paper evaluates the best approach for stroke detection using either accelerometers, gyroscopes or magnetometers, which are embedded into the inertial measuring unit. This work concludes what is the optimal training data set for stroke classification and proves that classifiers can perform well when tested on players who were not used to train the classifier. This work provides a significant step forward for our overall goal, which is to develop next generation sports coaching tools using both inertial and visual sensors in an instrumented indoor sporting environment.

Combining and choosing case base maintenance algorithms

Case-Based Reasoning (CBR) uses past experiences to solve new problems. The quality of the past experiences, which are stored as cases in a case base, is a big factor in the performance of a CBR system. The system's competence may be improved by adding problems to the case base after they have been solved and their solutions verified to be correct. However, from time to time, the case base may have to be refined to reduce redundancy and to get rid of any noisy cases that may have been introduced. Many case base maintenance algorithms have been developed to delete noisy and redundant cases. However, different algorithms work well in different situations and it may be difficult for a knowledge engineer to know which one is the best to use for a particular case base. In this thesis, we investigate ways to combine algorithms to produce better deletion decisions than the decisions made by individual algorithms, and ways to choose which algorithm is best for a given case base at a given time. We analyse five of the most commonly-used maintenance algorithms in detail and show how the different algorithms perform better on different datasets. This motivates us to develop a new approach: maintenance by a committee of experts (MACE). MACE allows us to combine maintenance algorithms to produce a composite algorithm which exploits the merits of each of the algorithms that it contains. By combining different algorithms in different ways we can also define algorithms that have different trade-offs between accuracy and deletion. While MACE allows us to define an infinite number of new composite algorithms, we still face the problem of choosing which algorithm to use. To make this choice, we need to be able to identify properties of a case base that are predictive of which maintenance algorithm is best. We examine a number of measures of dataset complexity for this purpose. These provide a numerical way to describe a case base at a given time. We use the numerical description to develop a meta-case-based classification system. This system uses previous experience about which maintenance algorithm was best to use for other case bases to predict which algorithm to use for a new case base. Finally, we give the knowledge engineer more control over the deletion process by creating incremental versions of the maintenance algorithms. These incremental algorithms suggest one case at a time for deletion rather than a group of cases, which allows the knowledge engineer to decide whether or not each case in turn should be deleted or kept. We also develop incremental versions of the complexity measures, allowing us to create an incremental version of our meta-case-based classification system. Since the case base changes after each deletion, the best algorithm to use may also change. The incremental system allows us to choose which algorithm is the best to use at each point in the deletion process.

Digital signal processing and artificial intelligence for the automated classification of food allergy

As a by-product of the ‘information revolution’ which is currently unfolding, lifetimes of man (and indeed computer) hours are being allocated for the automated and intelligent interpretation of data. This is particularly true in medical and clinical settings, where research into machine-assisted diagnosis of physiological conditions gains momentum daily. Of the conditions which have been addressed, however, automated classification of allergy has not been investigated, even though the numbers of allergic persons are rising, and undiagnosed allergies are most likely to elicit fatal consequences. On the basis of the observations of allergists who conduct oral food challenges (OFCs), activity-based analyses of allergy tests were performed. Algorithms were investigated and validated by a pilot study which verified that accelerometer-based inquiry of human movements is particularly well-suited for objective appraisal of activity. However, when these analyses were applied to OFCs, accelerometer-based investigations were found to provide very poor separation between allergic and non-allergic persons, and it was concluded that the avenues explored in this thesis are inadequate for the classification of allergy. Heart rate variability (HRV) analysis is known to provide very significant diagnostic information for many conditions. Owing to this, electrocardiograms (ECGs) were recorded during OFCs for the purpose of assessing the effect that allergy induces on HRV features. It was found that with appropriate analysis, excellent separation between allergic and nonallergic subjects can be obtained. These results were, however, obtained with manual QRS annotations, and these are not a viable methodology for real-time diagnostic applications. Even so, this was the first work which has categorically correlated changes in HRV features to the onset of allergic events, and manual annotations yield undeniable affirmation of this. Fostered by the successful results which were obtained with manual classifications, automatic QRS detection algorithms were investigated to facilitate the fully automated classification of allergy. The results which were obtained by this process are very promising. Most importantly, the work that is presented in this thesis did not obtain any false positive classifications. This is a most desirable result for OFC classification, as it allows complete confidence to be attributed to classifications of allergy. Furthermore, these results could be particularly advantageous in clinical settings, as machine-based classification can detect the onset of allergy which can allow for early termination of OFCs. Consequently, machine-based monitoring of OFCs has in this work been shown to possess the capacity to significantly and safely advance the current state of clinical art of allergy diagnosis

Skin and bone: the face in the archaeological imagination

Thirteen unique archaeological countenances from Ireland were produced through the Manchester method of facial reconstruction. Their gaze prompts a space for a broad discourse regarding the face found within human and artefactual remains of Ireland. These faces are reminders of the human element which is at the core of the discipline of archaeology. These re-constructions create a voyeuristic relationship with the past. At once sating a curiosity about the past, facial reconstructions also provide a catharsis to our presently situated selves. As powerful visual documents, archaeological facial reconstructions illustrate re-presentations of the past as well as how the present can be connected to the past. Through engagment with Emmanuel Levinas’s (1906- 1995) main philosophical themes, the presence of the face is examined in a diachronic structure. The ‘starting point’ is the Neolithic period which has been associated with the notion of visuality with a reconstruction from the early Neolithic site of Annagh, Co. Limerick. The following layer of analysis appears with attention to intersubjectivity in the early medieval period with facial reconstructions from Dooey, Co. Donegal and Owenbristy, Co. Galway. Building upon the past concepts, the late medieval period is associated with the notion of alterity and paired with faces from Ballinderry, Co. Kildare and a sample of males from Gallen Priory, Co. Offaly. The final layer of examination culminates with the application of response and respons-ibility to the post-medieval Irish landscape with facial reconstructions from the prison on Spike Island, Co. Cork. These layers of investigation are similar to the stratigraphical composition of both the archaeological landscape and the skeletal/soft tissue landscape of the face. The separation of the neglected phenomenon of the face from the overwhelming embrace of the field of craniometrics is necessary. Through this detachment a new manner in which to discuss the face and its place within the (bio)archaeological record is possible. Encountering the faces seen in mortuary contexts, material culture, and archaeological facial reconstructions, inform and shape the archaeological imagination.

Neonatal EEG classification using atomic decomposition

The electroencephalogram (EEG) is an important noninvasive tool used in the neonatal intensive care unit (NICU) for the neurologic evaluation of the sick newborn infant. It provides an excellent assessment of at-risk newborns and formulates a prognosis for long-term neurologic outcome.The automated analysis of neonatal EEG data in the NICU can provide valuable information to the clinician facilitating medical intervention. The aim of this thesis is to develop a system for automatic classification of neonatal EEG which can be mainly divided into two parts: (1) classification of neonatal EEG seizure from nonseizure, and (2) classifying neonatal background EEG into several grades based on the severity of the injury using atomic decomposition. Atomic decomposition techniques use redundant time-frequency dictionaries for sparse signal representations or approximations. The first novel contribution of this thesis is the development of a novel time-frequency dictionary coherent with the neonatal EEG seizure states. This dictionary was able to track the time-varying nature of the EEG signal. It was shown that by using atomic decomposition and the proposed novel dictionary, the neonatal EEG transition from nonseizure to seizure states could be detected efficiently. The second novel contribution of this thesis is the development of a neonatal seizure detection algorithm using several time-frequency features from the proposed novel dictionary. It was shown that the time-frequency features obtained from the atoms in the novel dictionary improved the seizure detection accuracy when compared to that obtained from the raw EEG signal. With the assistance of a supervised multiclass SVM classifier and several timefrequency features, several methods to automatically grade EEG were explored. In summary, the novel techniques proposed in this thesis contribute to the application of advanced signal processing techniques for automatic assessment of neonatal EEG recordings.