Francisco Ibarrola

Francisco Ibarrola, Rohan Lulham, Kazjon Grace

In creativity support and computational co-creativity contexts, the task of discovering appropriate prompts for use with text-to-image generative models remains difficult. In many cases the creator wishes to evoke a certain impression with the image, but the task of conferring that succinctly in a text prompt poses a challenge: affective language is nuanced, complex, and model-specific. In this work we introduce a method for generating images conditioned on desired affect, quantified using a psychometrically validated three-component approach, that can be combined with conditioning on text descriptions. We first train a neural network for estimating the affect content of text and images from semantic embeddings, and then demonstrate how this can be used to exert control over a variety of generative models. We show examples of how affect modifies the outputs, provide quantitative and qualitative analysis of its capabilities, and discuss possible extensions and use cases.

Francisco Ibarrola, Kazjon Grace

Quality and diversity have been proposed as reasonable heuristics for assessing content generated by co-creative systems, but to date there has been little agreement around what constitutes the latter or how to measure it. Proposed approaches for assessing generative models in terms of diversity have limitations in that they compare the model's outputs to a ground truth that in the era of large pre-trained generative models might not be available, or entail an impractical number of computations. We propose an alternative based on entropy of neural network encodings for comparing diversity between sets of images that does not require ground-truth knowledge and is easy to compute. We also compare two pre-trained networks and show how the choice relates to the notion of diversity that we want to evaluate. We conclude with a discussion of the potential applications of these measures for ideation in interactive systems, model evaluation, and more broadly within computational creativity.

Francisco J. Ibarrola, Nishant Ravikumar, Alejandro F. Frangi

Generative models are undoubtedly a hot topic in Artificial Intelligence, among which the most common type is Generative Adversarial Networks (GANs). These architectures let one synthesise artificial datasets by implicitly modelling the underlying probability distribution of a real-world training dataset. With the introduction of Conditional GANs and their variants, these methods were extended to generating samples conditioned on ancillary information available for each sample within the dataset. From a practical standpoint, however, one might desire to generate data conditioned on partial information. That is, only a subset of the ancillary conditioning variables might be of interest when synthesising data. In this work, we argue that standard Conditional GANs are not suitable for such a task and propose a new Adversarial Network architecture and training strategy to deal with the ensuing problems. Experiments illustrating the value of the proposed approach in digit and face image synthesis under partial conditioning information are presented, showing that the proposed method can effectively outperform the standard approach under these circumstances.

Nicolás Nieto, Francisco Ibarrola, Victoria Peterson et al.

Extreme Learning Machines (ELMs) have become a popular tool in the field of Artificial Intelligence due to their very high training speed and generalization capabilities. Another advantage is that they have a single hyper-parameter that must be tuned up: the number of hidden nodes. Most traditional approaches dictate that this parameter should be chosen smaller than the number of available training samples in order to avoid over-fitting. In fact, it has been proved that choosing the number of hidden nodes equal to the number of training samples yields a perfect training classification with probability 1 (w.r.t. the random parameter initialization). In this article we argue that in spite of this, in some cases it may be beneficial to choose a much larger number of hidden nodes, depending on certain properties of the data. We explain why this happens and show some examples to illustrate how the model behaves. In addition, we present a pruning algorithm to cope with the additional computational burden associated to the enlarged ELM. Experimental results using electroencephalography (EEG) signals show an improvement in performance with respect to traditional ELM approaches, while diminishing the extra computing time associated to the use of large architectures.

Francisco Ibarrola, Leandro Di Persia, Ruben Spies

When recorded in an enclosed room, a sound signal will most certainly get affected by reverberation. This not only undermines audio quality, but also poses a problem for many human-machine interaction technologies that use speech as their input. In this work, a new blind, two-stage dereverberation approach based in a generalized β-divergence as a fidelity term over a non-negative representation is proposed. The first stage consists of learning the spectral structure of the signal solely from the observed spectrogram, while the second stage is devoted to model reverberation. Both steps are taken by minimizing a cost function in which the aim is put either in constructing a dictionary or a good representation by changing the divergence involved. In addition, an approach for finding an optimal fidelity parameter for dictionary learning is proposed. An algorithm for implementing the proposed method is described and tested against state-of-the-art methods. Results show improvements for both artificial reverberation and real recordings.

Francisco J. Ibarrola, Leandro E. Di Persia, Ruben D. Spies

When a signal is recorded in an enclosed room, it typically gets affected by reverberation. This degradation represents a problem when dealing with audio signals, particularly in the field of speech signal processing, such as automatic speech recognition. Although there are some approaches to deal with this issue that are quite satisfactory under certain conditions, constructing a method that works well in a general context still poses a significant challenge. In this article, we propose a method based on convolutive nonnegative matrix factorization that mixes two penalizers in order to impose certain characteristics over the time-frequency components of the restored signal and the reverberant components. An algorithm for implementing the method is described and tested. Comparisons of the results against those obtained with state of the art methods are presented, showing significant improvement.