We provide larger step-size restrictions for which gradient descent based algorithms (almost surely) avoid strict saddle points. In particular, consider a twice differentiable (non-convex) objective function whose gradient has Lipschitz constant L and whose Hessian is well-behaved. We prove that the probability of initial conditions for gradient descent with step-size up to 2/L converging to a strict saddle point, given one uniformly random initialization, is zero. This extends previous results up to the sharp limit imposed by the convex case. In addition, the arguments hold in the case when a learning rate schedule is given, with either a continuous decaying rate or a piece-wise constant schedule.

With the industry trend of shifting from a traditional hierarchical approach to flatter management structure, crowdsourced performance assessment gained mainstream popularity. One fundamental challenge of crowdsourced performance assessment is the risks that personal interest can introduce distortions of facts, especially when the system is used to determine merit pay or promotion. In this paper, we developed a method to identify bias and strategic behavior in crowdsourced performance assessment, using a rich dataset collected from a professional service firm in China. We find a pattern of "discriminatory generosity" on the part of peer evaluation, where raters downgrade their peer coworkers who have passed objective promotion requirements while overrating their peer coworkers who have not yet passed. This introduces two types of biases: the first aimed against more competent competitors, and the other favoring less eligible peers which can serve as a mask of the first bias. This paper also aims to bring angles of fairness-aware data mining to talent and management computing. Historical decision records, such as performance ratings, often contain subjective judgment which is prone to bias and strategic behavior. For practitioners of predictive talent analytics, it is important to investigate potential bias and strategic behavior underlying historical decision records.

The paper presents Imbalance-XGBoost, a Python package that combines the powerful XGBoost software with weighted and focal losses to tackle binary label-imbalanced classification tasks. Though a small-scale program in terms of size, the package is, to the best of the authors' knowledge, the first of its kind which provides an integrated implementation for the two losses on XGBoost and brings a general-purpose extension on XGBoost for label-imbalanced scenarios. In this paper, the design and usage of the package are described with exemplar code listings, and its convenience to be integrated into Python-driven Machine Learning projects is illustrated. Furthermore, as the first- and second-order derivatives of the loss functions are essential for the implementations, the algebraic derivation is discussed and it can be deemed as a separate algorithmic contribution. The performances of the algorithms implemented in the package are empirically evaluated on Parkinson's disease classification data set, and multiple state-of-the-art performances have been observed. Given the scalable nature of XGBoost, the package has great potentials to be applied to real-life binary classification tasks, which are usually of large-scale and label-imbalanced.

Deep neural networks are vulnerable to adversarial perturbations: small changes in the input easily lead to misclassification. In this work, we propose an attack methodology catered not only for cases where the perturbations are measured by $\ell_p$ norms, but in fact any adversarial dissimilarity metric with a closed proximal form. This includes, but is not limited to, $\ell_1$, $\ell_2$, $\ell_\infty$ perturbations, and the $\ell_0$ counting "norm", i.e. true sparseness. Our approach to generating perturbations is a natural extension of our recent work, the LogBarrier attack, which previously required the metric to be differentiable. We demonstrate our new algorithm, ProxLogBarrier, on the MNIST, CIFAR10, and ImageNet-1k datasets. We attack undefended and defended models, and show that our algorithm transfers to various datasets with little parameter tuning. In particular, in the $\ell_0$ case, our algorithm finds significantly smaller perturbations compared to multiple existing methods

We provide a comprehensive review of classical algorithms for compressive sensing of images, focused on Total variation methods, with a view to application in LiDAR systems. Our primary focus is providing a full review for beginners in the field, as well as simulating the kind of noise found in real LiDAR systems. To this end, we provide an overview of the theoretical background, a brief discussion of various considerations that come in to play in compressive sensing, and a standardized comparison of off-the-shelf methods, intended as a quick-start guide to choosing algorithms for compressive sensing applications.

This paper aims to analyze knowledge isomorphism between pre-trained deep neural networks. We propose a generic definition for knowledge isomorphism between neural networks at different fuzziness levels, and design a task-agnostic and model-agnostic method to disentangle and quantify isomorphic features from intermediate layers of a neural network. As a generic tool, our method can be broadly used for different applications. In preliminary experiments, we have used knowledge isomorphism as a tool to diagnose feature representations of neural networks. Knowledge isomorphism provides new insights to explain the success of existing deep-learning techniques, such as knowledge distillation and network compression. More crucially, it has been shown that knowledge isomorphism can also be used to refine pre-trained networks and boost performance.

2D path planning in static environment is a well-known problem and one of the common ways to solve it is to 1) represent the environment as a grid and 2) perform a heuristic search for a path on it. At the same time 2D grid resembles much a digital image, thus an appealing idea comes to being -- to treat the problem as an image generation task and to solve it utilizing the recent advances in deep learning. In this work we make an attempt to apply a generative neural network as a path finder and report preliminary results, convincing enough to claim that this direction of research is worth further exploration.

Quantum algorithms can enhance machine learning in different aspects. Here, we study quantum-enhanced least-square support vector machine (LS-SVM). Firstly, a novel quantum algorithm that uses continuous variable to assist matrix inversion is introduced to simplify the algorithm for quantum LS-SVM, while retaining exponential speed-up. Secondly, we propose a hybrid quantum-classical version for sparse solutions of LS-SVM. By encoding a large dataset into a quantum state, a much smaller transformed dataset can be extracted using quantum matrix toolbox, which is further processed in classical SVM. We also incorporate kernel methods into the above quantum algorithms, which uses both exponential growth Hilbert space of qubits and infinite dimensionality of continuous variable for quantum feature maps. The quantum LS-SVM exploits quantum properties to explore important themes for SVM such as sparsity and kernel methods, and stresses its quantum advantages ranging from speed-up to the potential capacity to solve classically difficult machine learning tasks.

We propose Style Conditioned Recommendations (SCR) and introduce style injection as a method to diversify recommendations. We use Conditional Variational Autoencoder (CVAE) architecture, where both the encoder and decoder are conditioned on a user profile learned from item content data. This allows us to apply style transfer methodologies to the task of recommendations, which we refer to as injection. To enable style injection, user profiles are learned to be interpretable such that they express users' propensities for specific predefined styles. These are learned via label-propagation from a dataset of item content, with limited labeled points. To perform injection, the condition on the encoder is learned while the condition on the decoder is selected per explicit feedback. Explicit feedback can be taken either from a user's response to a style or interest quiz, or from item ratings. In the absence of explicit feedback, the condition at the encoder is applied to the decoder. We show a 12% improvement on NDCG@20 over the traditional VAE based approach and an average 22% improvement on AUC across all classes for predicting user style profiles against our best performing baseline. After injecting styles we compare the user style profile to the style of the recommendations and show that injected styles have an average +133% increase in presence. Our results show that style injection is a powerful method to diversify recommendations while maintaining personal relevance. Our main contribution is an application of a semi-supervised approach that extends item labels to interpretable user profiles.

Learning from triplet comparison data has been extensively studied in the context of metric learning, where we want to learn a distance metric between two instances, and ordinal embedding, where we want to learn an embedding in an Euclidean space of the given instances that preserves the comparison order as well as possible. Unlike fully-labeled data, triplet comparison data can be collected in a more accurate and human-friendly way. Although learning from triplet comparison data has been considered in many applications, an important fundamental question of whether we can learn a classifier only from triplet comparison data has remained unanswered. In this paper, we give a positive answer to this important question by proposing an unbiased estimator for the classification risk under the empirical risk minimization framework. Since the proposed method is based on the empirical risk minimization framework, it inherently has the advantage that any surrogate loss function and any model, including neural networks, can be easily applied. Furthermore, we theoretically establish an estimation error bound for the proposed empirical risk minimizer. Finally, we provide experimental results to show that our method empirically works well and outperforms various baseline methods.