This paper presents our contribution to PolEval 2019 Task 6: Hate speech and bullying detection. We describe three parallel approaches that we followed: fine-tuning a pre-trained ULMFiT model to our classification task, fine-tuning a pre-trained BERT model to our classification task, and using the TPOT library to find the optimal pipeline. We present results achieved by these three tools and review their advantages and disadvantages in terms of user experience. Our team placed second in subtask 2 with a shallow model found by TPOT: a~logistic regression classifier with non-trivial feature engineering.

We consider minimizing a nonconvex, smooth function $f$ on a Riemannian manifold $\mathcal{M}$. We show that a perturbed version of Riemannian gradient descent algorithm converges to a second-order stationary point (and hence is able to escape saddle points on the manifold). The rate of convergence depends as $1/\epsilon^2$ on the accuracy $\epsilon$, which matches a rate known only for unconstrained smooth minimization. The convergence rate depends polylogarithmically on the manifold dimension $d$, hence is almost dimension-free. The rate also has a polynomial dependence on the parameters describing the curvature of the manifold and the smoothness of the function. While the unconstrained problem (Euclidean setting) is well-studied, our result is the first to prove such a rate for nonconvex, manifold-constrained problems.

Traditionally, most of the existing attribute learning methods are trained based on the consensus of annotations aggregated from a limited number of annotators. However, the consensus might fail in settings, especially when a wide spectrum of annotators with different interests and comprehension about the attribute words are involved. In this paper, we develop a novel multi-task method to understand and predict personalized attribute annotations. Regarding the attribute preference learning for each annotator as a specific task, we first propose a multi-level task parameter decomposition to capture the evolution from a highly popular opinion of the mass to highly personalized choices that are special for each person. Meanwhile, for personalized learning methods, ranking prediction is much more important than accurate classification. This motivates us to employ an Area Under ROC Curve (AUC) based loss function to improve our model. On top of the AUC-based loss, we propose an efficient method to evaluate the loss and gradients. Theoretically, we propose a novel closed-form solution for one of our non-convex subproblem, which leads to provable convergence behaviors. Furthermore, we also provide a generalization bound to guarantee a reasonable performance. Finally, empirical analysis consistently speaks to the efficacy of our proposed method.

Many recent machine learning tools rely on differentiable game formulations. While several numerical methods have been proposed for these types of games, most of the work has been on convergence proofs or on upper bounds for the rate of convergence of those methods. In this work, we approach the question of fundamental iteration complexity by providing lower bounds. We generalise Nesterov's argument -- used in single-objective optimisation to derive a lower bound for a class of first-order black box optimisation algorithms -- to games. Moreover, we extend to games the p-SCLI framework used to derive spectral lower bounds for a large class of derivative-based single-objective optimisers. Finally, we propose a definition of the condition number arising from our lower bound analysis that matches the conditioning observed in upper bounds. Our condition number is more expressive than previously used definitions, as it covers a wide range of games, including bilinear games that lack strong convex-concavity.

Autonomy and adaptation of machines requires that they be able to measure their own errors. We consider the advantages and limitations of such an approach when a machine has to measure the error in a regression task. How can a machine measure the error of regression sub-components when it does not have the ground truth for the correct predictions? A compressed sensing approach applied to the error signal of the regressors can recover their precision error without any ground truth. It allows for some regressors to be \emph{strongly correlated} as long as not too many are so related. Its solutions, however, are not unique - a property of ground truth inference solutions. Adding $\ell_1$--minimization as a condition can recover the correct solution in settings where error correction is possible. We briefly discuss the similarity of the mathematics of ground truth inference for regressors to that for classifiers.

Understanding if classifiers generalize to out-of-sample datasets is a central problem in machine learning. Microscopy images provide a standardized way to measure the generalization capacity of image classifiers, as we can image the same classes of objects under increasingly divergent, but controlled factors of variation. We created a public dataset of 132,209 images of mouse cells, COOS-7 (Cells Out Of Sample 7-Class). COOS-7 provides a classification setting where four test datasets have increasing degrees of covariate shift: some images are random subsets of the training data, while others are from experiments reproduced months later and imaged by different instruments. We benchmarked a range of classification models using different representations, including transferred neural network features, end-to-end classification with a supervised deep CNN, and features from a self-supervised CNN. While most classifiers perform well on test datasets similar to the training dataset, all classifiers failed to generalize their performance to datasets with greater covariate shifts. These baselines highlight the challenges of covariate shifts in image data, and establish metrics for improving the generalization capacity of image classifiers.

ADHD is being recognized as a diagnosis which persists into adulthood impacting economic, occupational, and educational outcomes. There is an increased need to accurately diagnose and recommend interventions for this population. One consideration is the development and implementation of reliable and valid outcome measures which reflect core diagnostic criteria. For example, adults with ADHD have reduced working memory capacity when compared to their peers (Michalek et al., 2014). A reduction in working memory capacity indicates attentional control deficits which align with many symptoms outlined on behavioral checklists used to diagnose ADHD. Using computational methods, such as eye tracking technology, to generate a relationship between ADHD and measures of working memory capacity would be useful to advancing our understanding and treatment of the diagnosis in adults. This chapter will outline a feasibility study in which eye tracking was used to measure eye gaze metrics during a working memory capacity task for adults with and without ADHD and machine learning algorithms were applied to generate a feature set unique to the ADHD diagnosis. The chapter will summarize the purpose, methods, results, and impact of this study.

We present a technique to improve the transferability of deep representations learned on small labeled datasets by introducing self-supervised tasks as auxiliary loss functions. While recent approaches for self-supervised learning have shown the benefits of training on large unlabeled datasets, we find improvements in generalization even on small datasets and when combined with strong supervision. Learning representations with self-supervised losses reduces the relative error rate of a state-of-the-art meta-learner by 5-25% on several few-shot learning benchmarks, as well as off-the-shelf deep networks on standard classification tasks when training from scratch. We find the benefits of self-supervision increase with the difficulty of the task. Our approach utilizes the images within the dataset to construct self-supervised losses and hence is an effective way of learning transferable representations without relying on any external training data.

The policy gradient theorem describes the gradient of the expected discounted return with respect to an agent's policy parameters. However, most policy gradient methods do not use the discount factor in the manner originally prescribed, and therefore do not optimize the discounted objective. It has been an open question in RL as to which, if any, objective they optimize instead. We show that the direction followed by these methods is not the gradient of any objective, and reclassify them as semi-gradient methods with respect to the undiscounted objective. Further, we show that they are not guaranteed to converge to a locally optimal policy, and construct an counterexample where they will converge to the globally pessimal policy with respect to both the discounted and undiscounted objectives.

De-identification of clinical records is an extremely important process which enables the use of the wealth of information present in them. There are a lot of techniques available for this but none of the method implementation has evaluated the scalability, which is an important benchmark. We evaluated numerous deep learning techniques such as BiLSTM-CNN, IDCNN, CRF, BiLSTM-CRF, SpaCy, etc. on both the performance and efficiency. We propose that the SpaCy model implementation for scrubbing sensitive PHI data from medical records is both well performing and extremely efficient compared to other published models.