14 December 2020 in IEEE/ACM Transactions on Audio, Speech, and Language Processing
A recently introduced, effective feature extraction technique for computational paralinguistics is that of Bag-of-Audio-Words (BoAW), where we cluster the frame-level training vectors, and represent each speech utterance based on the cluster of its frames. Over the past few years, several improvements have been proposed for the original BoAW approach, but none of them has examined the impact of the stochastic nature of the clustering step. In this study we demonstrate experimentally that the random factor present in the BoAW clustering step is indeed propagated into the next classification step, eventually leading to suboptimal classification performance. As a solution, we propose to train an ensemble of classifiers; that is, we repeat the BoAW codebook selection step several times, train a separate classifier models for these BoAW representation versions and combine the predictions. Our results, obtained for three different paralinguistic datasets, demonstrate that this ensemble technique makes the whole paralinguistic classification process more robust, therefore it leads to improvements in the classification performance. We tested this technique on three different paralinguistic datasets, and achieved the highest Unweighted Average Recall score reported so far on the iHEARu-EAT corpus.
14 February 2023 in arXiv
We consider the problem of Learning from Label Proportions (LLP), a weakly supervised classification setup where instances are grouped into "bags", and only the frequency of class labels at each bag is available. Albeit, the objective of the learner is to achieve low task loss at an individual instance level. Here we propose Easyllp: a flexible and simple-to-implement debiasing approach based on aggregate labels, which operates on arbitrary loss functions. Our technique allows us to accurately estimate the expected loss of an arbitrary model at an individual level. We showcase the flexibility of our approach by applying it to popular learning frameworks, like Empirical Risk Minimization (ERM) and Stochastic Gradient Descent (SGD) with provable guarantees on instance level performance. More concretely, we exhibit a variance reduction technique that makes the quality of LLP learning deteriorate only by a factor of k (k being bag size) in both ERM and SGD setups, as compared to full supervision. Finally, we validate our theoretical results on multiple datasets demonstrating our algorithm performs as well or better than previous LLP approaches in spite of its simplicity.
22 November 2019 in arXiv
Web crawling is the problem of keeping a cache of webpages fresh, i.e., having the most recent copy available when a page is requested. This problem is usually coupled with the natural restriction that the bandwidth available to the web crawler is limited. The corresponding optimization problem was solved optimally by Azar et al. [2018] under the assumption that, for each webpage, both the elapsed time between two changes and the elapsed time between two requests follow a Poisson distribution with known parameters. In this paper, we study the same control problem but under the assumption that the change rates are unknown a priori, and thus we need to estimate them in an online fashion using only partial observations (i.e., single-bit signals indicating whether the page has changed since the last refresh). As a point of departure, we characterise the conditions under which one can solve the problem with such partial observability. Next, we propose a practical estimator and compute confidence intervals for it in terms of the elapsed time between the observations. Finally, we show that the explore-and-commit algorithm achieves an $\mathcal{O}(\sqrt{T})$ regret with a carefully chosen exploration horizon. Our simulation study shows that our online policy scales well and achieves close to optimal performance for a wide range of the parameters.