Abstract A problem with convolutional neural networks (CNNs) is that they require large datasets to obtain adequate robustness; on small datasets, they are prone to overfitting. Many methods have been proposed to overcome this shortcoming with CNNs. In cases where additional samples cannot easily be collected, a common approach is to generate more data points from existing data using an augmentation technique. In image classification, many augmentation approaches utilize simple image manipulation algorithms. In this work, we propose some new methods for data augmentation based on several image transformations: the Fourier transform (FT), the Radon transform (RT), and the discrete cosine transform (DCT). These and other data augmentation methods are considered in order to quantify their effectiveness in creating ensembles of neural networks. The novelty of this research is to consider different strategies for data augmentation to generate training sets from which to train several classifiers which are combined into an ensemble. Specifically, the idea is to create an ensemble based on a kind of bagging of the training set, where each model is trained on a different training set obtained by augmenting the original training set with different approaches. We build ensembles on the data level by adding images generated by combining fourteen augmentation approaches, with three based on FT, RT, and DCT, proposed here for the first time. Pretrained ResNet50 networks are finetuned on training sets that include images derived from each augmentation method. These networks and several fusions are evaluated and compared across eleven benchmarks. Results show that building ensembles on the data level by combining different data augmentation methods produce classifiers that not only compete competitively against the state-of-the-art but often surpass the best approaches reported in the literature.

MOTIVATION: Because DNA-binding proteins (DNA-BPs) play a vital role in all aspects of genetic activity, the development of reliable and efficient systems for automatic DNA-BP classification is becoming a crucial proteomic technology. Key to this technology is the discovery of powerful protein representations and feature extraction methods. The goal of this article is to develop experimentally a system for automatic DNA-BP classification by comparing and combining different descriptors taken from different types of protein representations. RESULTS: The descriptors we evaluate include those starting from the position-specific scoring matrix (PSSM) of proteins, those derived from the amino-acid sequence (AAS), various matrix representations of proteins and features taken from the three-dimensional tertiary structure of proteins. We also introduce some new variants of protein descriptors. Each descriptor is used to train a separate support vector machine (SVM), and results are combined by sum rule. Our final system obtains state-or-the-art results on three benchmark DNA-BP datasets. AVAILABILITY AND IMPLEMENTATION: The MATLAB code for replicating the experiments presented in this paper is available at https://github.com/LorisNanni.

Bioimage classification is increasingly becoming more important in many biological studies including those that require accurate cell phenotype recognition, subcellular localization, and histopathological classification. In this paper, we present a new General Purpose (GenP) bioimage classification method that can be applied to a large range of classification problems. The GenP system we propose is an ensemble that combines multiple texture features (both handcrafted and learned descriptors) for superior and generalizable discriminative power. Our ensemble obtains a boosting of performance by combining local features, dense sampling features, and deep learning features. Each descriptor is used to train a different Support Vector Machine that is then combined by sum rule. We evaluate our method on a diverse set of bioimage classification tasks each represented by a benchmark database, including some of those available in the IICBU 2008 database. Each bioimage classification task represents a typical subcellular, cellular, and tissue level classification problem. Our evaluation on these datasets demonstrates that the proposed GenP bioimage ensemble obtains state-of-the-art performance without any ad-hoc dataset tuning of the parameters (thereby avoiding any risk of overfitting/overtraining). To reproduce the experiments reported in this paper, the MATLAB code of all the descriptors is available at https://github.com/LorisNanni and https://www.dropbox.com/s/bguw035yrqz0pwp/ElencoCode.docx?dl=0.

Image identification of animals is mostly centred on identifying them based on their appearance, but there are other ways images can be used to identify animals, including by representing the sounds they make with images. In this study, the authors present a novel and effective approach for automated identification of birds and whales using some of the best texture descriptors in the computer vision literature. The visual features of sounds are built starting from the audio file and are taken from images constructed from different spectrograms and from harmonic and percussion images. These images are divided into sub‐windows from which sets of texture descriptors are extracted. The experiments reported in this study using a dataset of Bird vocalisations targeted for species recognition and a dataset of right whale calls targeted for whale detection (as well as three well‐known benchmarks for music genre classification) demonstrate that the fusion of different texture features enhances performance. The experiments also demonstrate that the fusion of different texture features with audio features is not only comparable with existing audio signal approaches but also statistically improves some of the stand‐alone audio features. The code for the experiments will be publicly available at https://www.dropbox.com/s/bguw035yrqz0pwp/ElencoCode.docx?dl=0 .

MOTIVATION: Given an unknown compound, is it possible to predict its Anatomical Therapeutic Chemical class/classes? This is a challenging yet important problem since such a prediction could be used to deduce not only a compound's possible active ingredients but also its therapeutic, pharmacological and chemical properties, thereby substantially expediting the pace of drug development. The problem is challenging because some drugs and compounds belong to two or more ATC classes, making machine learning extremely difficult. RESULTS: In this article a multi-label classifier system is proposed that incorporates information about a compound's chemical-chemical interaction and its structural and fingerprint similarities to other compounds belonging to the different ATC classes. The proposed system reshapes a 1D feature vector to obtain a 2D matrix representation of the compound. This matrix is then described by a histogram of gradients that is fed into a Multi-Label Learning with Label-Specific Features classifier. Rigorous cross-validations demonstrate the superior prediction quality of this method compared with other state-of-the-art approaches developed for this problem, a superiority that is reflected particularly in the absolute true rate, the most important and harshest metric for assessing multi-label systems. AVAILABILITY AND IMPLEMENTATION: The MATLAB code for replicating the experiments presented in this article is available at https://www.dropbox.com/s/7v1mey48tl9bfgz/ToolPaperATC.rar?dl=0 . CONTACT: [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

In this paper a novel approach for automatic bird species classification is described. The proposed strategy is based on features taken from the textural content of spectrogram images of bird vocalizations. We show how several texture descriptors can be used for representing the spectrograms. The following approaches are tested here with spectrograms for the first time: Local Ternary Phase Quantization, Heterogeneous Auto-Similarities of Characteristics, and an ensemble of variants of Local Binary Pattern Histogram Fourier. Combining this set of descriptors greatly increases classification performance and markedly improves previous ensembles of texture descriptors used for describing a spectrogram. Moreover, a further improvement is obtained when the texture descriptors are combined with the acoustic features. SVM classifiers are used in the classification step, with final results computed using 10-fold cross-validation. For a fair comparison with other methods in the literature, the experiments are performed on a benchmark database composed of 46 bird species used for this classification task. The best accuracy rate obtained is about 94.5%. The MATLAB code we used is publicly available to other researchers for future comparisons, as well as the database used in the experiments.

MOTIVATION: The microarray report measures the expressions of tens of thousands of genes, producing a feature vector that is high in dimensionality and that contains much irrelevant information. This dimensionality degrades classification performance. Moreover, datasets typically contain few samples for training, leading to the 'curse of dimensionality' problem. It is essential, therefore, to find good methods for reducing the size of the feature set. RESULTS: In this article, we propose a method for gene microarray classification that combines different feature reduction approaches for improving classification performance. Using a support vector machine (SVM) as our classifier, we examine an SVM trained using a set of selected genes; an SVM trained using the feature set obtained by Neighborhood Preserving Embedding feature transform; a set of SVMs trained using a set of orthogonal wavelet coefficients of different wavelet mothers; a set of SVMs trained using texture descriptors extracted from the microarray, considering it as an image; and an ensemble that combines the best feature extraction methods listed above. The positive results reported offer confirmation that combining different features extraction methods greatly enhances system performance. The experiments were performed using several different datasets, and our results [expressed as both accuracy and area under the receiver operating characteristic (ROC) curve] show the goodness of the proposed approach with respect to the state of the art. AVAILABILITY: The MATHLAB code of the proposed approach is publicly available at bias.csr.unibo.it/nanni/micro.rar.

Conversational agents are attributed humanlike characteristics; in particular, they are often assumed to have a gender. There is evidence that gender sets up expectations that have an impact on user experiences with agents. The objective of this paper is to explore gender affordances of conversational agents. Our examination takes a holistic approach to the analysis of the application of gender stereotypes to nine chatterbots: six embodied (three male and three female), two disembodied (male and female), and a robot embodiment. Building on social psychology research, we test the persistence of gender stereotypes in the selection of conversation topics and in the elicitation of disinhibition and verbal abuse. Our study is based on quantitative textual analysis of interaction logs. A dictionary of English sexual slang and derogatory terms was developed for this study. Results show that gender stereotypes tend to affect interaction more at the relational (style) level then at the referential (content) level of conversation. People attribute negative stereotypes to female-presenting chatterbots more often than they do to male-presenting chatterbots, and female-presenting chatterbots are more often the objects of implicit and explicit sexual attention and swear words. We conclude by calling for a more informed analysis of user interactions that considers the full range of user interactions.

Two fundamental (and oftentimes opposing) metaphors have directed much of HCI design: HCI is communication and HCI is direct manipulation. Beneath these HCI metaphors, however, is the unspoken metaphor of computer is woman. In this paper we expose this foundational metaphor. We begin by identifying the origin of computer is woman in the early history of computing. Drawing upon postmodern feminist theory, we then explore how this metaphor has resulted in the feminization of HCI is communication and second person interfaces. We show how images of femininity proliferate, becoming the projected images of male fantasies and ideals of womanhood. In becoming these idealized images, the interface is revealed as man in female drag. Finally, not only do we undress the interface to uncover how HCI is communication wraps the computer’s difference from human being within the more basic metaphor of computer is woman, but we also disclose dangers that can arise when this metaphor goes unacknowledged and unexamined.

This paper presents a descriptive lexical analysis of spontaneous conversations between users and the 2005 Loebner prize winning chatterbot, Jabberwacky. The study was motivated in part by the suspicion that evidence in support of the Media Equation, especially in the field of conversational agents, was supported by incomplete data; too often omitted in its purview is the occurrence of unsavoury user responses. Our study shows that conversations with Jabberwacky often bring about the expression of negative verbal disinhibition. We discovered that 10% of the total stems in the corpus reflected abusive language, and approximately 11% of the sample addressed hard-core sex. Users were often rude and violated the conversation maxims of manner, quantity, and relevance. Also particularly pronounced in the conversations was a persistent need of the user to define the speakers’ identities (human vs. machine). Users were also curious to understand and test the cognitive capabilities of the chatterbot. Our analysis indicates that the Media Equation may need qualifying, that users treat computers that talk, less as they do people and more as they might treat something not quite an object yet not quite human.

“You see, there’s a primal joy in hitting a thing in motion. It’s one of the oldest pleasures there is. Something moves, boo, you wing it. Beast, bird or human, the thing to do is to knock it down. It’s primal, Davy. It’s basic to the origin of the species.” – Don DeLillo, Americana Coming across an unknown creature generates anxiety. Its autonomy gives rise to fears; its agency requires vigilance. In the primitive world of predators and prey, it is wise to be leery of anything that moves and to learn as quickly as possible whether a creature is aggressive or possesses dangerous weapons of self-defence. What better way to find that out then to throw a rock at it from a safe distance to see how it reacts and what it is capable of doing to protect itself. Testing, teasing, and poking creatures is so crucial for human and animal survival that it is no wonder these behaviours are also pleasurable, as are play fighting and play chasing (for a review of the literature, see Aldis, 1975). Apparently an elementary predator–prey schema runs across cultures and most species (Lorenz, 1939).