Danijel Skočaj

Combining reconstructive and discriminative subspace methods for robust classification and regression by subsampling

Linear subspace methods that provide sufficient reconstruction of the data, such as PCA, offer an efficient way of dealing with missing pixels, outliers, and occlusions that often appear in the visual data. Discriminative methods, such as LDA, which, on the other hand, are better suited for classification tasks, are highly sensitive to corrupted data. We present a theoretical framework for achieving the best of both types of methods: an approach that combines the discrimination power of discriminative methods with the reconstruction property of reconstructive methods which enables one to work on subsets of pixels in images to efficiently detect and reject the outliers. The proposed approach is therefore capable of robust classification with a high-breakdown point. We also show that subspace methods, such as CCA, which are used for solving regression tasks, can be treated in a similar manner. The theoretical results are demonstrated on several computer vision tasks showing that the proposed approach significantly outperforms the standard discriminative methods in the case of missing pixels and images containing occlusions and outliers.

Weighted and robust learning of subspace representations

A reliable system for visual learning and recognition should enable a selective treatment of individual parts of input data and should successfully deal with noise and occlusions. These requirements are not satisfactorily met when visual learning is approached by appearance-based modeling of objects and scenes using the traditional PCA approach. In this paper we extend standard PCA approach to overcome these shortcomings. We first present a weighted version of PCA, which, unlike the standard approach, considers individual pixels and images selectively, depending on the corresponding weights. Then we propose a robust PCA method for obtaining a consistent subspace representation in the presence of outlying pixels in the training images. The method is based on the EM algorithm for estimation of principal subspaces in the presence of missing data. We demonstrate the efficiency of the proposed methods in a number of experiments.

A Robust PCA Algorithm for Building Representations from Panoramic Images

Appearance-based modeling of objects and scenes using PCA has been successfully applied in many recognition tasks. Robust methods which have made the recognition stage less susceptible to outliers, occlusions, and varying illumination have further enlarged the domain of applicability. However, much less research has been done in achieving robustness in the learning stage. In this paper, we propose a novel robust PCA method for obtaining a consistent subspace representation in the presence of outlying pixels in the training images. The method is based on the EM algorithm for estimation of principal subspaces in the presence of missing data. By treating the outlying points as missing pixels, we arrive at a robust PCA representation. We demonstrate experimentally that the proposed method is efficient. In addition, we apply the method to a set of panoramic images to build a representation that enables surveillance and view-based mobile robot localization.

Range image acquisition of objects with non-uniform albedo using structured light range sensor

We present an approach to acquisition of range images of objects with non-uniform albedo using a structured light sensor. The main idea is to systematically vary the intensity of the light projector and to form high dynamic scale radiance maps. The range images are then formed from these radiance maps. We tested the method on the objects which have surfaces with very different reflectance properties. We demonstrate that the range images obtained from the high dynamic scale radiance maps are of much better quality, than those obtained directly from the original images of a limited dynamic scale.

Self-supervised cross-modal online learning of basic object affordances for developmental robotic systems

For a developmental robotic system to function successfully in the real world, it is important that it be able to form its own internal representations of affordance classes based on observable regularities in sensory data. Usually successful classifiers are built using labeled training data, but it is not always realistic to assume that labels are available in a developmental robotics setting. There does, however, exist an advantage in this setting that can help circumvent the absence of labels: co-occurrence of correlated data across separate sensory modalities over time. The main contribution of this paper is an online classifier training algorithm based on Kohonens learning vector quantization (LVQ) that, by taking advantage of this co-occurrence information, does not require labels during training, either dynamically generated or otherwise. We evaluate the algorithm in experiments involving a robotic arm that interacts with various household objects on a table surface where camera systems extract features for two separate visual modalities. It is shown to improve its ability to classify the affordances of novel objects over time, coming close to the performance of equivalent fully-supervised algorithms.

Self-Understanding and Self-Extension: A Systems and Representational Approach

There are many different approaches to building a system that can engage in autonomous mental development. In this paper, we present an approach based on what we term self-understanding, by which we mean the explicit representation of and reasoning about what a system does and does not know, and how that knowledge changes under action. We present an architecture and a set of representations used in two robot systems that exhibit a limited degree of autonomous mental development, which we term self-extension. The contributions include: representations of gaps and uncertainty for specific kinds of knowledge, and a goal management and planning system for setting and achieving learning goals.

Incremental LDA Learning by Combining Reconstructive and Discriminative Approaches

Incremental subspace methods have proven to enable efficient training if large amounts of training data have to be processed or if not all data is available in advance. In this paper we focus on incremental LDA learning which provides good classification results while it assures a compact data representation. In contrast to existing incremental LDA methods we additionally consider reconstructive information when incrementally building the LDA subspace. Hence, we get a more flexible representation that is capable to adapt to new data. Moreover, this allows to add new instances to existing classes as well as to add new classes. The experimental results show that the proposed approach outperforms other incremental LDA methods even approaching classification results obtained by batch learning.

Incremental and robust learning of subspace representations

Learning is a fundamental capability of any cognitive system. To enable efficient operation of a cognitive agent in a real-world environment, visual learning has to be a continuous and robust process. In this article, we present a method for subspace learning, which takes these considerations into account. We present an incremental method, which sequentially updates the principal subspace considering weighted influence of individual images as well as individual pixels within an image. We further extend this approach to enable determination of consistencies in the input data and imputation of the inconsistent values using the previously acquired knowledge, resulting in a novel method for incremental, weighted, and robust subspace learning. We demonstrate the effectiveness of the proposed concept in several experiments on learning of object and background representations.

Online kernel density estimation for interactive learning

In this paper we propose a Gaussian-kernel-based online kernel density estimation which can be used for applications of online probability density estimation and online learning. Our approach generates a Gaussian mixture model of the observed data and allows online adaptation from positive examples as well as from the negative examples. The adaptation from the negative examples is realized by a novel concept of unlearning in mixture models. Low complexity of the mixtures is maintained through a novel compression algorithm. In contrast to the existing approaches, our approach does not require fine-tuning parameters for a specific application, we do not assume specific forms of the target distributions and temporal constraints are not assumed on the observed data. The strength of the proposed approach is demonstrated with examples of online estimation of complex distributions, an example of unlearning, and with an interactive learning of basic visual concepts.

A system for interactive learning in dialogue with a tutor

In this paper we present representations and mechanisms that facilitate continuous learning of visual concepts in dialogue with a tutor and show the implemented robot system. We present how beliefs about the world are created by processing visual and linguistic information and show how they are used for planning system behaviour with the aim at satisfying its internal drive - to extend its knowledge. The system facilitates different kinds of learning initiated by the human tutor or by the system itself. We demonstrate these principles in the case of learning about object colours and basic shapes.