Language-independent tokenisation (LIT) methods that do not require labelled language resources or lexicons have recently gained popularity because of their applicability in resource-poor languages. Moreover, they compactly represent a language using a fixed size vocabulary and can efficiently handle unseen or rare words. On the other hand, language-specific tokenisation (LST) methods have a long and established history, and are developed using carefully created lexicons and training resources. Unlike subtokens produced by LIT methods, LST methods produce valid morphological subwords. Despite the contrasting trade-offs between LIT vs. LST methods, their performance on downstream NLP tasks remain unclear. In this paper, we empirically compare the two approaches using semantic similarity measurement as an evaluation task across a diverse set of languages. Our experimental results covering eight languages show that LST consistently outperforms LIT when the vocabulary size is large, but LIT can produce comparable or better results than LST in many languages with comparatively smaller (i.e. less than 100K words) vocabulary sizes, encouraging the use of LIT when language-specific resources are unavailable, incomplete or a smaller model is required. Moreover, we find that smoothed inverse frequency (SIF) to be an accurate method to create word embeddings from subword embeddings for multilingual semantic similarity prediction tasks. Further analysis of the nearest neighbours of tokens show that semantically and syntactically related tokens are closely embedded in subword embedding spaces
Dominance move (DoM) is a binary quality indicator to compare solution sets in multiobjective optimization. The indicator allows a more natural and intuitive relation when comparing solution sets. It is Pareto compliant and does not demand any parameters or reference sets. In spite of its advantages, the combinatorial calculation nature is a limitation. The original formulation presents an efficient method to calculate it in a biobjective case only. This work presents an assignment formulation to calculate DoM in problems with three objectives or more. Some initial experiments, in the biobjective space, were done to present the model correctness. Next, other experiments, using three dimensions, were also done to show how DoM could be compared with other indicators: inverted generational distance (IGD) and hypervolume (HV). Results show the assignment formulation for DoM is valid for more than three objectives. However, there are some strengths and weaknesses, which are discussed and detailed. Some notes, considerations, and future research paths conclude this work.
We show that, while Minesweeper is NP-complete, its hyperbolic variant is in P. Our proof does not rely on the rules of Minesweeper, but is valid for any puzzle based on satisfying local constraints on a graph embedded in the hyperbolic plane.
In Constraint Programming, constraints are usually represented as predicates allowing or forbidding combinations of values. However, some algorithms exploit a finer representation: error functions. Their usage comes with a price though: it makes problem modeling significantly harder. Here, we propose a method to automatically learn an error function corresponding to a constraint, given a function deciding if assignments are valid or not. This is, to the best of our knowledge, the first attempt to automatically learn error functions for hard constraints. Our method uses a variant of neural networks we named Interpretable Compositional Networks, allowing us to get interpretable results, unlike regular artificial neural networks. Experiments on 5 different constraints show that our system can learn functions that scale to high dimensions, and can learn fairly good functions over incomplete spaces.
The Bayesian learning rule is a natural-gradient variational inference method, which not only contains many existing learning algorithms as special cases but also enables the design of new algorithms. Unfortunately, when variational parameters lie in an open constraint set, the rule may not satisfy the constraint and requires line-searches which could slow down the algorithm. In this work, we address this issue for positive-definite constraints by proposing an improved rule that naturally handles the constraints. Our modification is obtained by using Riemannian gradient methods, and is valid when the approximation attains a \emph{block-coordinate natural parameterization} (e.g., Gaussian distributions and their mixtures). Our method outperforms existing methods without any significant increase in computation. Our work makes it easier to apply the rule in the presence of positive-definite constraints in parameter spaces.
Canonical correlation analysis (CCA) has become a key tool for population neuroimaging, allowing investigation of associations between many imaging and non-imaging measurements. As other variables are often a source of variability not of direct interest, previous work has used CCA on residuals from a model that removes these effects, then proceeded directly to permutation inference. We show that such a simple permutation test leads to inflated error rates. The reason is that residualisation introduces dependencies among the observations that violate the exchangeability assumption. Even in the absence of nuisance variables, however, a simple permutation test for CCA also leads to excess error rates for all canonical correlations other than the first. The reason is that a simple permutation scheme does not ignore the variability already explained by previous canonical variables. Here we propose solutions for both problems: in the case of nuisance variables, we show that transforming the residuals to a lower dimensional basis where exchangeability holds results in a valid permutation test; for more general cases, with or without nuisance variables, we propose estimating the canonical correlations in a stepwise manner, removing at each iteration the variance already explained, while dealing with different number of variables in both sides. We also discuss how to address the multiplicity of tests, proposing an admissible test that is not conservative, and provide a complete algorithm for permutation inference for CCA.
A fundamental component of neural network verification is the computation of bounds on the values their outputs can take. Previous methods have either used off-the-shelf solvers, discarding the problem structure, or relaxed the problem even further, making the bounds unnecessarily loose. We propose a novel approach based on Lagrangian Decomposition. Our formulation admits an efficient supergradient ascent algorithm, as well as an improved proximal algorithm. Both the algorithms offer three advantages: (i) they yield bounds that are provably at least as tight as previous dual algorithms relying on Lagrangian relaxations; (ii) they are based on operations analogous to forward/backward pass of neural networks layers and are therefore easily parallelizable, amenable to GPU implementation and able to take advantage of the convolutional structure of problems; and (iii) they allow for anytime stopping while still providing valid bounds. Empirically, we show that we obtain bounds comparable with off-the-shelf solvers in a fraction of their running time, and obtain tighter bounds in the same time as previous dual algorithms. This results in an overall speed-up when employing the bounds for formal verification. Code for our algorithms is available at https://github.com/oval-group/decomposition-plnn-bounds.
There is a vast body of literature related to methods for detecting changepoints (CP). However, less attention has been paid to assessing the statistical reliability of the detected CPs. In this paper, we introduce a novel method to perform statistical inference on the significance of the CPs, estimated by a Dynamic Programming (DP)-based optimal CP detection algorithm. Based on the selective inference (SI) framework, we propose an exact (non-asymptotic) approach to compute valid p-values for testing the significance of the CPs. Although it is well-known that SI has low statistical power because of over-conditioning, we address this disadvantage by introducing parametric programming techniques. Then, we propose an efficient method to conduct SI with the minimum amount of conditioning, leading to high statistical power. We conduct experiments on both synthetic and real-world datasets, through which we offer evidence that our proposed method is more powerful than existing methods, has decent performance in terms of computational efficiency, and provides good results in many practical applications.
We study generalised linear regression and classification for a synthetically generated dataset encompassing different problems of interest, such as learning with random features, neural networks in the lazy training regime, and the hidden manifold model. We consider the high-dimensional regime and using the replica method from statistical physics, we provide a closed-form expression for the asymptotic generalisation performance in these problems, valid in both the under- and over-parametrised regimes and for a broad choice of generalised linear model loss functions. In particular, we show how to obtain analytically the so-called double descent behaviour for logistic regression with a peak at the interpolation threshold, we illustrate the superiority of orthogonal against random Gaussian projections in learning with random features, and discuss the role played by correlations in the data generated by the hidden manifold model. Beyond the interest in these particular problems, the theoretical formalism introduced in this manuscript provides a path to further extensions to more complex tasks.
Self-supervision is key to extending use of deep learning for label scarce domains. For most of self-supervised approaches data transformations play an important role. However, up until now the impact of transformations have not been studied. Furthermore, different transformations may have different impact on the system. We provide novel insights into the use of data transformation in self-supervised tasks, specially pertaining to clustering. We show theoretically and empirically that certain set of transformations are helpful in convergence of self-supervised clustering. We also show the cases when the transformations are not helpful or in some cases even harmful. We show faster convergence rate with valid transformations for convex as well as certain family of non-convex objectives along with the proof of convergence to the original set of optima. We have synthetic as well as real world data experiments. Empirically our results conform with the theoretical insights provided.
The continuous expansion of the urban traffic sensing infrastructure has led to a surge in the volume of widely available road related data. Consequently, increasing effort is being dedicated to the creation of intelligent transportation systems, where decisions on issues ranging from city-wide road maintenance planning to improving the commuting experience are informed by computational models of urban traffic instead of being left entirely to humans. The automation of traffic management has received substantial attention from the research community, however, most approaches target highways, produce predictions valid for a limited time window or require expensive retraining of available models in order to accurately forecast traffic at a new location. In this article, we propose a novel and accurate traffic flow prediction method based on symbolic regression enhanced with a lag operator. Our approach produces robust models suitable for the intricacies of urban roads, much more difficult to predict than highways. Additionally, there is no need to retrain the model for a period of up to 9 weeks. Furthermore, the proposed method generates models that are transferable to other segments of the road network, similar to, yet geographically distinct from the ones they were initially trained on. We demonstrate the achievement of these claims by conducting extensive experiments on data collected from the Darmstadt urban infrastructure.
We study how to efficiently estimate average treatment effects (ATEs) using adaptive experiments. In adaptive experiments, experimenters sequentially assign treatments to experimental units while updating treatment assignment probabilities based on past data. We start by defining the efficient treatment-assignment probability, which minimizes the semiparametric efficiency bound for ATE estimation. Our proposed experimental design estimates and uses the efficient treatment-assignment probability to assign treatments. At the end of the proposed design, the experimenter estimates the ATE using a newly proposed Adaptive Augmented Inverse Probability Weighting (A2IPW) estimator. We show that the asymptotic variance of the A2IPW estimator using data from the proposed design achieves the minimized semiparametric efficiency bound. We also analyze the estimator's finite-sample properties and develop nonparametric and nonasymptotic confidence intervals that are valid at any round of the proposed design. These anytime valid confidence intervals allow us to conduct rate-optimal sequential hypothesis testing, allowing for early stopping and reducing necessary sample size.
Despite the tremendous efforts to democratize machine learning, especially in applied-science, the application is still often hampered by the lack of coding skills. As we consider programmatic understanding key to building effective and efficient machine learning solutions, we argue for a novel educational approach that builds upon the accessibility and acceptance of graphical user interfaces to convey programming skills to an applied-science target group. We outline a proof-of-concept, open-source web application, the PHOTON Wizard, which dynamically translates GUI interactions into valid source code for the Python machine learning framework PHOTON. Thereby, users possessing theoretical machine learning knowledge gain key insights into the model development workflow as well as an intuitive understanding of custom implementations. Specifically, the PHOTON Wizard integrates the concept of Educational Machine Learning Code Generators to teach users how to write code for designing, training, optimizing and evaluating custom machine learning pipelines.
The stochastic gradient descent (SGD) algorithm is widely used for parameter estimation, especially for huge data sets and online learning. While this recursive algorithm is popular for computation and memory efficiency, quantifying variability and randomness of the solutions has been rarely studied. This paper aims at conducting statistical inference of SGD-based estimates in an online setting. In particular, we propose a fully online estimator for the covariance matrix of averaged SGD iterates (ASGD) only using the iterates from SGD. We formally establish our online estimator's consistency and show that the convergence rate is comparable to offline counterparts. Based on the classic asymptotic normality results of ASGD, we construct asymptotically valid confidence intervals for model parameters. Upon receiving new observations, we can quickly update the covariance matrix estimate and the confidence intervals. This approach fits in an online setting and takes full advantage of SGD: efficiency in computation and memory.
Given only positive (P) and unlabeled (U) data, PU learning can train a binary classifier without any negative data. It has two building blocks: PU class-prior estimation (CPE) and PU classification; the latter has been well studied while the former has received less attention. Hitherto, the distributional-assumption-free CPE methods rely on a critical assumption that the support of the positive data distribution cannot be contained in the support of the negative data distribution. If this is violated, those CPE methods will systematically overestimate the class prior; it is even worse that we cannot verify the assumption based on the data. In this paper, we rethink CPE for PU learning-can we remove the assumption to make CPE always valid? We show an affirmative answer by proposing Regrouping CPE (ReCPE) that builds an auxiliary probability distribution such that the support of the positive data distribution is never contained in the support of the negative data distribution. ReCPE can work with any CPE method by treating it as the base method. Theoretically, ReCPE does not affect its base if the assumption already holds for the original probability distribution; otherwise, it reduces the positive bias of its base. Empirically, ReCPE improves all state-of-the-art CPE methods on various datasets, implying that the assumption has indeed been violated here.
To achieve optimal human-system integration in the context of user-AI interaction it is important that users develop a valid representation of how AI works. In most of the everyday interaction with technical systems users construct mental models (i.e., an abstraction of the anticipated mechanisms a system uses to perform a given task). If no explicit explanations are provided by a system (e.g. by a self-explaining AI) or other sources (e.g. an instructor), the mental model is typically formed based on experiences, i.e. the observations of the user during the interaction. The congruence of this mental model and the actual systems functioning is vital, as it is used for assumptions, predictions and consequently for decisions regarding system use. A key question for human-centered AI research is therefore how to validly survey users' mental models. The objective of the present research is to identify suitable elicitation methods for mental model analysis. We evaluated whether mental models are suitable as an empirical research method. Additionally, methods of cognitive tutoring are integrated. We propose an exemplary method to evaluate explainable AI approaches in a human-centered way.
We study minimax methods for off-policy evaluation (OPE) using value functions and marginalized importance weights. Despite that they hold promises of overcoming the exponential variance in traditional importance sampling, several key problems remain: (1) They require function approximation and are generally biased. For the sake of trustworthy OPE, is there anyway to quantify the biases? (2) They are split into two styles ("weight-learning" vs "value-learning"). Can we unify them? In this paper we answer both questions positively. By slightly altering the derivation of previous methods (one from each style; Uehara et al., 2020), we unify them into a single value interval that comes with a special type of double robustness: when either the value-function or the importance-weight class is well specified, the interval is valid and its length quantifies the misspecification of the other class. Our interval also provides a unified view of and new insights to some recent methods, and we further explore the implications of our results on exploration and exploitation in off-policy policy optimization with insufficient data coverage.
Image inpainting technology can patch images with missing pixels. Existing methods propose convolutional neural networks to repair corrupted images. The networks focus on the valid pixels around the missing pixels, use the encoder-decoder structure to extract valuable information, and use the information to fix the vacancy. However, if the missing part is too large to provide useful information, the result will exist blur, color mixing, and object confusion. In order to patch the large hole image, we study the existing approaches and propose a new network, the compression-decompression network. The compression network takes responsibility for inpainting and generating a down-sample image. The decompression network takes responsibility for extending the down-sample image into the original resolution. We construct the compression network with the residual network and propose a similar texture selection algorithm to extend the image that is better than using the super-resolution network. We evaluate our model over Places2 and CelebA data set and use the similarity ratio as the metric. The result shows that our model has better performance when the inpainting task has many conflicts.
The Nadaraya-Watson kernel estimator is among the most popular nonparameteric regression technique thanks to its simplicity. Its asymptotic bias has been studied by Rosenblatt in 1969 and has been reported in a number of related literature. However, Rosenblatt's analysis is only valid for infinitesimal bandwidth. In contrast, we propose in this paper an upper bound of the bias which holds for finite bandwidths. Moreover, contrarily to the classic analysis we allow for discontinuous first order derivative of the regression function, we extend our bounds for multidimensional domains and we include the knowledge of the bound of the regression function when it exists and if it is known, to obtain a tighter bound. We believe that this work has potential applications in those fields where some hard guarantees on the error are needed
Stein importance sampling is a widely applicable technique based on kernelized Stein discrepancy, which corrects the output of approximate sampling algorithms by reweighting the empirical distribution of the samples. A general analysis of this technique is conducted for the previously unconsidered setting where samples are obtained via the simulation of a Markov chain, and applies to an arbitrary underlying Polish space. We prove that Stein importance sampling yields consistent estimators for quantities related to a target distribution of interest by using samples obtained from a geometrically ergodic Markov chain with a possibly unknown invariant measure that differs from the desired target. The approach is shown to be valid under conditions that are satisfied for a large number of unadjusted samplers, and is capable of retaining consistency when data subsampling is used. Along the way, a universal theory of reproducing Stein kernels is established, which enables the construction of kernelized Stein discrepancy on general Polish spaces, and provides sufficient conditions for kernels to be convergence-determining on such spaces. These results are of independent interest for the development of future methodology based on kernelized Stein discrepancies.