Search

Melanoma is the most deadly skin cancer. Early diagnosis is a challenge for clinicians. Current algorithms for skin lesions' classification focus mostly on segmentation and feature extraction. This article instead puts the emphasis on the learning process, testing the recognition performance of three different classifiers: support vector machine (SVM), artificial neural network and k-nearest neighbor. Extensive experiments were run on a database of more than 5000 dermoscopy images. The obtained results show that the SVM approach outperforms the other methods reaching an average recognition rate of 82.5% comparable with those obtained by skilled clinicians. If confirmed, our data suggest that this method may improve classification results of a computer-assisted diagnosis of melanoma. © 2010 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 20, 316-322, 2010 [ABSTRACT FROM AUTHOR]

Abstract: Classifying materials from their appearance is challenging. Impressive results have been obtained under varying illumination and pose conditions. Still, the effect of scale variations and the possibility to generalise across different material samples are still largely unexplored. This paper (A preliminary version of this work was presented in Hayman et al. [E. Hayman, B. Caputo, M.J. Fritz, J.-O. Eklundh, On the significance of real world conditions for material classification, in: Proceedings of the ECCV, Lecture Notes in Computer Science, vol. 4, Springer, Prague, 2004, pp. 253–266].) addresses these issues, proposing a pure learning approach based on support vector machines. We study the effect of scale variations first on the artificially scaled CUReT database, showing how performance depends on the amount of scale information available during training. Since the CUReT database contains little scale variation and only one sample per material, we introduce a new database containing 10 CUReT materials at different distances, pose and illumination. This database provides scale variations, while allowing to evaluate generalisation capabilities: does training on the CUReT database enable recognition of another piece of sandpaper? Our results demonstrate that this is not yet possible, and that material classification is far from being solved in scenarios of practical interest. [Copyright &y& Elsevier]

Multi-modal learning has gained significant attention due to its ability to enhance machine learning algorithms. However, it brings challenges related to modality heterogeneity and domain shift. In this work, we address these challenges by proposing a new approach called Relative Norm Alignment (RNA) loss. RNA loss exploits the observation that variations in marginal distributions between modalities manifest as discrepancies in their mean feature norms, and rebalances feature norms across domains, modalities, and classes. This rebalancing improves the accuracy of models on test data from unseen ("target") distributions. In the context of Unsupervised Domain Adaptation (UDA), we use unlabeled target data to enhance feature transferability. We achieve this by combining RNA loss with an adversarial domain loss and an Information Maximization term that regularizes predictions on target data. We present a comprehensive analysis and ablation of our method for both Domain Generalization and UDA settings, testing our approach on different modalities for tasks such as first and third person action recognition, object recognition, and fatigue detection. Experimental results show that our approach achieves competitive or state-of-the-art performance on the proposed benchmarks, showing the versatility and effectiveness of our method in a wide range of applications. [ABSTRACT FROM AUTHOR]

One of the main challenges for developing visual recognition systems working in the wild is to devise computational models immune from the domain shift problem, i.e., accurate when test data are drawn from a (slightly) different data distribution than training samples. In the last decade, several research efforts have been devoted to devise algorithmic solutions for this issue. Recent attempts to mitigate domain shift have resulted into deep learning models for domain adaptation which learn domain-invariant representations by introducing appropriate loss terms, by casting the problem within an adversarial learning framework or by embedding into deep network specific domain normalization layers. This paper describes a novel approach for unsupervised domain adaptation. Similarly to previous works we propose to align the learned representations by embedding them into appropriate network feature normalization layers. Opposite to previous works, our Domain Alignment Layers are designed not only to match the source and target feature distributions but also to automatically learn the degree of feature alignment required at different levels of the deep network. Differently from most previous deep domain adaptation methods, our approach is able to operate in a multi-source setting. Thorough experiments on four publicly available benchmarks confirm the effectiveness of our approach. [ABSTRACT FROM AUTHOR]

In this work, we present a new, algorithm for multi-domain learning. Given a pretrained architecture and a set of visual domains received sequentially, the goal of multi-domain learning is to produce a single model performing a task in all the domains together. Recent works showed how we can address this problem by masking the internal weights of a given original convnet through learned binary variables. In this work, we provide a general formulation of binary mask-based models for multi-domain learning by affine transformations of the original network parameters. Our formulation obtains significantly higher levels of adaptation to new domains, achieving performances comparable to domain-specific models while requiring slightly more than 1 bit per network parameter per additional domain. Experiments on two popular benchmarks showcase the power of our approach, achieving performances close to state-of-the-art methods on the Visual Decathlon Challenge. [ABSTRACT FROM AUTHOR]

Deep neural networks have enabled major progresses in semantic segmentation. However, even the most advanced neural architectures suffer from important limitations. First, they are vulnerable to catastrophic forgetting, i.e., they perform poorly when they are required to incrementally update their model as new classes are available. Second, they rely on large amount of pixel-level annotations to produce accurate segmentation maps. To tackle these issues, we introduce a novel incremental class learning approach for semantic segmentation taking into account a peculiar aspect of this task: since each training step provides annotation only for a subset of all possible classes, pixels of the background class exhibit a semantic shift. Therefore, we revisit the traditional distillation paradigm by designing novel loss terms which explicitly account for the background shift. Additionally, we introduce a novel strategy to initialize classifier's parameters at each step in order to prevent biased predictions toward the background class. Finally, we demonstrate that our approach can be extended to point- and scribble-based weakly supervised segmentation, modeling the partial annotations to create priors for unlabeled pixels. We demonstrate the effectiveness of our approach with an extensive evaluation on the Pascal-VOC, ADE20K, and Cityscapes datasets, significantly outperforming state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Deep neural networks have enabled major progresses in semantic segmentation. However, even the most advanced neural architectures suffer from important limitations. First, they are vulnerable to catastrophic forgetting, i.e., they perform poorly when they are required to incrementally update their model as new classes are available. Second, they rely on large amount of pixel-level annotations to produce accurate segmentation maps. To tackle these issues, we introduce a novel incremental class learning approach for semantic segmentation taking into account a peculiar aspect of this task: since each training step provides annotation only for a subset of all possible classes, pixels of the background class exhibit a semantic shift. Therefore, we revisit the traditional distillation paradigm by designing novel loss terms which explicitly account for the background shift. Additionally, we introduce a novel strategy to initialize classifier's parameters at each step in order to prevent biased predictions toward the background class. Finally, we demonstrate that our approach can be extended to point- and scribble-based weakly supervised segmentation, modeling the partial annotations to create priors for unlabeled pixels. We demonstrate the effectiveness of our approach with an extensive evaluation on the Pascal-VOC, ADE20K, and Cityscapes datasets, significantly outperforming state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Human adaptability relies crucially on learning and merging knowledge from both supervised and unsupervised tasks: the parents point out few important concepts, but then the children fill in the gaps on their own. This is particularly effective, because supervised learning can never be exhaustive and thus learning autonomously allows to discover invariances and regularities that help to generalize. In this paper we propose to apply a similar approach to the problem of object recognition across domains: our model learns the semantic labels in a supervised fashion, and broadens its understanding of the data by learning from self-supervised signals on the same images. This secondary task helps the network to focus on object shapes, learning concepts like spatial orientation and part correlation, while acting as a regularizer for the classification task over multiple visual domains. Extensive experiments confirm our intuition and show that our multi-task method, combining supervised and self-supervised knowledge, provides competitive results with respect to more complex domain generalization and adaptation solutions. It also proves its potential in the novel and challenging predictive and partial domain adaptation scenarios. [ABSTRACT FROM AUTHOR]

Learning a visual object category from few samples is a compelling and challenging problem. In several real-world applications collecting many annotated data is costly and not always possible. However, a small training set does not allow to cover the high intraclass variability typical of visual objects. In this condition, machine learning methods provide very few guarantees. This paper presents a discriminative model adaptation algorithm able to proficiently learn a target object with few examples by relying on other previously learned source categories. The proposed method autonomously chooses from where and how much to transfer information by solving a convex optimization problem which ensures to have the minimal leave-one-out error on the available training set. We analyze several properties of the described approach and perform an extensive experimental comparison with other existing transfer solutions, consistently showing the value of our algorithm. [ABSTRACT FROM PUBLISHER]

In recent years there has been a lot of interest in designing principled classification algorithms over multiple cues, based on the intuitive notion that using more features should lead to better performance. In the domain of kernel methods, a principled way to use multiple features is the Multi Kernel Learning (MKL) approach. Here we present a MKL optimization algorithm based on stochastic gradient descent that has a guaranteed convergence rate. We directly solve the MKL problem in the primal formulation. By having a p-norm formulation of MKL, we introduce a parameter that controls the level of sparsity of the solution, while leading to an easier optimization problem. We prove theoretically and experimentally that 1) our algorithm has a faster convergence rate as the number of kernels grows; 2) the training complexity is linear in the number of training examples; 3) very few iterations are sufficient to reach good solutions. Experiments on standard benchmark databases support our claims. [ABSTRACT FROM AUTHOR]

Abstract: The capability to learn from experience is a key property for autonomous cognitive systems working in realistic settings. To this end, this paper presents an SVM-based algorithm, capable of learning model representations incrementally while keeping under control memory requirements. We combine an incremental extension of SVMs with a method reducing the number of support vectors needed to build the decision function without any loss in performance introducing a parameter which permits a user-set trade-off between performance and memory. The resulting algorithm is able to achieve the same recognition results as the original incremental method while reducing the memory growth. Our method is especially suited to work for autonomous systems in realistic settings. We present experiments on two common scenarios in this domain: adaptation in presence of dynamic changes and transfer of knowledge between two different autonomous agents, focusing in both cases on the problem of visual place recognition applied to mobile robot topological localization. Experiments in both scenarios clearly show the power of our approach. [Copyright &y& Elsevier]

Abstract: Support vector machines (SVMs) are one of the most successful algorithms for classification. However, due to their space and time requirements, they are not suitable for on-line learning, that is, when presented with an endless stream of training observations. In this paper we propose a new on-line algorithm, called on-line independent support vector machines (OISVMs), which approximately converges to the standard SVM solution each time new observations are added; the approximation is controlled via a user-defined parameter. The method employs a set of linearly independent observations and tries to project every new observation onto the set obtained so far, dramatically reducing time and space requirements at the price of a negligible loss in accuracy. As opposed to similar algorithms, the size of the solution obtained by OISVMs is always bounded, implying a bounded testing time. These statements are supported by extensive experiments on standard benchmark databases as well as on two real-world applications, namely place recognition by a mobile robot in an indoor environment and human grasping posture classification. [Copyright &y& Elsevier]

A common problem of kernel-based online algorithms, such as the kernel-based Perceptron algorithm, is the amount of memory required to store the online hypothesis, which may increase without bound as the algorithm progresses. Furthermore, the computational load of such algorithms grows linearly with the amount of memory used to store the hypothesis. To attack these problems, most previous work has focused on discarding some of the instances, in order to keep the memory bounded. In this paper we present a new algorithm, in which the instances are not discarded, but are instead projected onto the space spanned by the previous online hypothesis. We call this algorithm Projectron. While the memory size of the Projectron solution cannot be predicted before training, we prove that its solution is guaranteed to be bounded. We derive a relative mistake bound for the proposed algorithm, and deduce from it a slightly different algorithm which outperforms the Perceptron. We call this second algorithm Projectron++. We show that this algorithm can be extended to handle the multiclass and the structured output settings, resulting, as far as we know, in the first online bounded algorithm that can learn complex classification tasks. The method of bounding the hypothesis representation can be applied to any conservative online algorithm and to other online algorithms, as it is demonstrated for ALMA2. Experimental results on various data sets show the empirical advantage of our technique compared to various bounded online algorithms, both in terms of memory and accuracy. [ABSTRACT FROM AUTHOR]

Abstract: Automatic annotation of medical images is an increasingly important tool for physicians in their daily activity. Hospitals nowadays produce an increasing amount of data. Manual annotation is very costly and prone to human mistakes. This paper proposes a multi-cue approach to automatic medical image annotation. We represent images using global and local features. These cues are then combined using three alternative approaches, all based on the support vector machine algorithm. We tested our methods on the IRMA database, and with two of the three approaches proposed here we participated in the 2007 ImageCLEFmed benchmark evaluation, in the medical image annotation track. These algorithms ranked first and fifth, respectively among all submission. Experiments using the third approach also confirm the power of cue integration for this task. [Copyright &y& Elsevier]

Unsupervised Domain Adaptation (UDA) refers to the problem of learning a model in a target domain where labeled data are not available by leveraging information from annotated data in a source domain. Most deep UDA approaches operate in a single-source, single-target scenario, i.e., they assume that the source and the target samples arise from a single distribution. However, in practice most datasets can be regarded as mixtures of multiple domains. In these cases, exploiting traditional single-source, single-target methods for learning classification models may lead to poor results. Furthermore, it is often difficult to provide the domain labels for all data points, i.e. latent domains should be automatically discovered. This paper introduces a novel deep architecture which addresses the problem of UDA by automatically discovering latent domains in visual datasets and exploiting this information to learn robust target classifiers. Specifically, our architecture is based on two main components, i.e. a side branch that automatically computes the assignment of each sample to its latent domain and novel layers that exploit domain membership information to appropriately align the distribution of the CNN internal feature representations to a reference distribution. We evaluate our approach on publicly available benchmarks, showing that it outperforms state-of-the-art domain adaptation methods. [ABSTRACT FROM AUTHOR]

Improving the functionality of prosthetic hands with noninvasive techniques is still a challenge. Surface electromyography (sEMG) currently gives limited control capabilities; however, the application of machine learning to the analysis of sEMG signals is promising and has recently been applied in practice, but many questions still remain. In this study, we recorded the sEMG activity of the forearm of 11 male subjects with transradial amputation who were mentally performing 40 hand and wrist movements. The classification performance and the number of independent movements (defined as the subset of movements that could be distinguished with >90% accuracy) were studied in relationship to clinical parameters related to the amputation. The analysis showed that classification accuracy and the number of independent movements increased significantly with phantom limb sensation intensity, remaining forearm percentage, and time since amputation. The classification results suggest the possibility of naturally controlling up to 11 movements of a robotic prosthetic hand with almost no training. Knowledge of the relationship between classification accuracy and clinical parameters adds new information regarding the nature of phantom limb pain as well as other clinical parameters, and it can lay the foundations for future "functional amputation" procedures in surgery. [ABSTRACT FROM AUTHOR]

At the time of this writing, the main means of control for polyarticulated self-powered hand prostheses is surface electromyography (sEMG). In the clinical setting, data collected from two electrodes are used to guide the hand movements selecting among a finite number of postures. Machine learning has been applied in the past to the sEMG signal (not in the clinical setting) with interesting results, which provide more insight on how these data could be used to improve prosthetic functionality. Researchers have mainly concentrated so far on increasing the accuracy of sEMG classification and/or regression, but, in general, a finer control implies a longer training period. A desirable characteristic would be to shorten the time needed by a patient to learn how to use the prosthesis. To this aim, we propose here a general method to reuse past experience, in the form of models synthesized from previous subjects, to boost the adaptivity of the prosthesis. Extensive tests on databases recorded from healthy subjects in controlled and noncontrolled conditions reveal that the method significantly improves the results over the baseline nonadaptive case. This promising approach might be employed to pretrain a prosthesis before shipping it to a patient, leading to a shorter training phase. [ABSTRACT FROM AUTHOR]