Research connected to &quot;machine learning&quot;

CHALLENGER: Training with Attribution Maps

We show that utilizing attribution maps for training neural networks can improve regularization of models and thus increase performance. Regularization is key in deep learning, especially when training complex models on relatively small datasets. In order to understand inner workings of neural networks, attribution methods such as Layer-wise Relevance Propagation (LRP) have been extensively studied, particularly for interpreting the relevance of input features. We introduce Challenger, a module that leverages the explainable power of attribution maps in order to manipulate particularly relevant input patterns. Therefore, exposing and subsequently resolving regions of ambiguity towards separating classes on the ground-truth data manifold, an issue that arises particularly when training models on rather small datasets. Our Challenger module increases model performance through building more diverse filters within the network and can be applied to any input data domain. We demonstrate that our approach results in substantially better classification as well as calibration performance on datasets with only a few samples up to datasets with thousands of samples. In particular, we show that our generic domain-independent approach yields state-of-the-art results in vision, natural language processing and on time series tasks.

preprint2020arXiv

The Benefits of Pairwise Discriminators for Adversarial Training

Adversarial training methods typically align distributions by solving two-player games. However, in most current formulations, even if the generator aligns perfectly with data, a sub-optimal discriminator can still drive the two apart. Absent additional regularization, the instability can manifest itself as a never-ending game. In this paper, we introduce a family of objectives by leveraging pairwise discriminators, and show that only the generator needs to converge. The alignment, if achieved, would be preserved with any discriminator. We provide sufficient conditions for local convergence; characterize the capacity balance that should guide the discriminator and generator choices; and construct examples of minimally sufficient discriminators. Empirically, we illustrate the theory and the effectiveness of our approach on synthetic examples. Moreover, we show that practical methods derived from our approach can better generate higher-resolution images.

preprint2016arXiv

Active Regression with Adaptive Huber Loss

This paper addresses the scalar regression problem through a novel solution to exactly optimize the Huber loss in a general semi-supervised setting, which combines multi-view learning and manifold regularization. We propose a principled algorithm to 1) avoid computationally expensive iterative schemes while 2) adapting the Huber loss threshold in a data-driven fashion and 3) actively balancing the use of labelled data to remove noisy or inconsistent annotations at the training stage. In a wide experimental evaluation, dealing with diverse applications, we assess the superiority of our paradigm which is able to combine robustness towards noise with both strong performance and low computational cost.

Understanding Hyperdimensional Computing for Parallel Single-Pass Learning

Hyperdimensional computing (HDC) is an emerging learning paradigm that computes with high dimensional binary vectors. It is attractive because of its energy efficiency and low latency, especially on emerging hardware -- but HDC suffers from low model accuracy, with little theoretical understanding of what limits its performance. We propose a new theoretical analysis of the limits of HDC via a consideration of what similarity matrices can be "expressed" by binary vectors, and we show how the limits of HDC can be approached using random Fourier features (RFF). We extend our analysis to the more general class of vector symbolic architectures (VSA), which compute with high-dimensional vectors (hypervectors) that are not necessarily binary. We propose a new class of VSAs, finite group VSAs, which surpass the limits of HDC. Using representation theory, we characterize which similarity matrices can be "expressed" by finite group VSA hypervectors, and we show how these VSAs can be constructed. Experimental results show that our RFF method and group VSA can both outperform the state-of-the-art HDC model by up to 7.6\% while maintaining hardware efficiency.

FAQNAS: FLOPs-aware Hybrid Quantum Neural Architecture Search using Genetic Algorithm

Hybrid Quantum Neural Networks (HQNNs), which combine parameterized quantum circuits with classical neural layers, are emerging as promising models in the noisy intermediate-scale quantum (NISQ) era. While quantum circuits are not naturally measured in floating point operations (FLOPs), most HQNNs (in NISQ era) are still trained on classical simulators where FLOPs directly dictate runtime and scalability. Hence, FLOPs represent a practical and viable metric to measure the computational complexity of HQNNs. In this work, we introduce FAQNAS, a FLOPs-aware neural architecture search (NAS) framework that formulates HQNN design as a multi-objective optimization problem balancing accuracy and FLOPs. Unlike traditional approaches, FAQNAS explicitly incorporates FLOPs into the optimization objective, enabling the discovery of architectures that achieve strong performance while minimizing computational cost. Experiments on five benchmark datasets (MNIST, Digits, Wine, Breast Cancer, and Iris) show that quantum FLOPs dominate accuracy improvements, while classical FLOPs remain largely fixed. Pareto-optimal solutions reveal that competitive accuracy can often be achieved with significantly r

Federated Learning for Data Streams

Federated learning (FL) is an effective solution to train machine learning models on the increasing amount of data generated by IoT devices and smartphones while keeping such data localized. Most previous work on federated learning assumes that clients operate on static datasets collected before training starts. This approach may be inefficient because 1) it ignores new samples clients collect during training, and 2) it may require a potentially long preparatory phase for clients to collect enough data. Moreover, learning on static datasets may be simply impossible in scenarios with small aggregate storage across devices. It is, therefore, necessary to design federated algorithms able to learn from data streams. In this work, we formulate and study the problem of \emph{federated learning for data streams}. We propose a general FL algorithm to learn from data streams through an opportune weighted empirical risk minimization. Our theoretical analysis provides insights to configure such an algorithm, and we evaluate its performance on a wide range of machine learning tasks.

preprint2020arXiv

Exploring Restart Distributions

We consider the generic approach of using an experience memory to help exploration by adapting a restart distribution. That is, given the capacity to reset the state with those corresponding to the agent's past observations, we help exploration by promoting faster state-space coverage via restarting the agent from a more diverse set of initial states, as well as allowing it to restart in states associated with significant past experiences. This approach is compatible with both on-policy and off-policy methods. However, a caveat is that altering the distribution of initial states could change the optimal policies when searching within a restricted class of policies. To reduce this unsought learning bias, we evaluate our approach in deep reinforcement learning which benefits from the high representational capacity of deep neural networks. We instantiate three variants of our approach, each inspired by an idea in the context of experience replay. Using these variants, we show that performance gains can be achieved, especially in hard exploration problems.

Online Algorithmic Recourse by Collective Action

Research on algorithmic recourse typically considers how an individual can reasonably change an unfavorable automated decision when interacting with a fixed decision-making system. This paper focuses instead on the online setting, where system parameters are updated dynamically according to interactions with data subjects. Beyond the typical individual-level recourse, the online setting opens up new ways for groups to shape system decisions by leveraging the parameter update rule. We show empirically that recourse can be improved when users coordinate by jointly computing their feature perturbations, underscoring the importance of collective action in mitigating adverse automated decisions.

GDI: Rethinking What Makes Reinforcement Learning Different From Supervised Learning

Deep Q Network (DQN) firstly kicked the door of deep reinforcement learning (DRL) via combining deep learning (DL) with reinforcement learning (RL), which has noticed that the distribution of the acquired data would change during the training process. DQN found this property might cause instability for training, so it proposed effective methods to handle the downside of the property. Instead of focusing on the unfavourable aspects, we find it critical for RL to ease the gap between the estimated data distribution and the ground truth data distribution while supervised learning (SL) fails to do so. From this new perspective, we extend the basic paradigm of RL called the Generalized Policy Iteration (GPI) into a more generalized version, which is called the Generalized Data Distribution Iteration (GDI). We see massive RL algorithms and techniques can be unified into the GDI paradigm, which can be considered as one of the special cases of GDI. We provide theoretical proof of why GDI is better than GPI and how it works. Several practical algorithms based on GDI have been proposed to verify the effectiveness and extensiveness of it. Empirical experiments prove our state-of-the-art (SOTA) performance on Arcade Learning Environment (ALE), wherein our algorithm has achieved 9620.98% mean human normalized score (HNS), 1146.39% median HNS and 22 human world record breakthroughs (HWRB) using only 200M training frames. Our work aims to lead the RL research to step into the journey of conquering the human world records and seek real superhuman agents on both performance and efficiency.

preprint2015arXiv

Significance Analysis of High-Dimensional, Low-Sample Size Partially Labeled Data

Classification and clustering are both important topics in statistical learning. A natural question herein is whether predefined classes are really different from one another, or whether clusters are really there. Specifically, we may be interested in knowing whether the two classes defined by some class labels (when they are provided), or the two clusters tagged by a clustering algorithm (where class labels are not provided), are from the same underlying distribution. Although both are challenging questions for the high-dimensional, low-sample size data, there has been some recent development for both. However, when it is costly to manually place labels on observations, it is often that only a small portion of the class labels is available. In this article, we propose a significance analysis approach for such type of data, namely partially labeled data. Our method makes use of the whole data and tries to test the class difference as if all the labels were observed. Compared to a testing method that ignores the label information, our method provides a greater power, meanwhile, maintaining the size, illustrated by a comprehensive simulation study. Theoretical properties of the propo

Physics-Informed Gaussian Process Regression for the Constitutive Modeling of Concrete: A Data-Driven Improvement to Phenomenological Models

Understanding and modeling the constitutive behavior of concrete is crucial for civil and defense applications, yet widely used phenomenological models such as Karagozian \& Case concrete (KCC) model depend on empirically calibrated failure surfaces that lack flexibility in model form and associated uncertainty quantification. This work develops a physics-informed framework that retains the modular elastoplastic structure of KCC model while replacing its empirical failure surface with a constrained Gaussian Process Regression (GPR) surrogate that can be learned directly from experimentally accessible observables. Triaxial compression data under varying confinement levels are used for training, and the surrogate is then evaluated at confinement levels not included in the training set to assess its generalization capability. Results show that an unconstrained GPR interpolates well near training conditions but deteriorates and violates essential physical constraints under extrapolation, even when augmented with simulated data. In contrast, a physics-informed GPR that incorporates derivative-based constraints aligned with known material behavior yields markedly better accuracy and reli

Maximum Mean Discrepancy for Generalization in the Presence of Distribution and Missingness Shift

Covariate shifts are a common problem in predictive modeling on real-world problems. This paper proposes addressing the covariate shift problem by minimizing Maximum Mean Discrepancy (MMD) statistics between the training and test sets in either feature input space, feature representation space, or both. We designed three techniques that we call MMD Representation, MMD Mask, and MMD Hybrid to deal with the scenarios where only a distribution shift exists, only a missingness shift exists, or both types of shift exist, respectively. We find that integrating an MMD loss component helps models use the best features for generalization and avoid dangerous extrapolation as much as possible for each test sample. Models treated with this MMD approach show better performance, calibration, and extrapolation on the test set.

preprint2012arXiv

The Kernel Pitman-Yor Process

In this work, we propose the kernel Pitman-Yor process (KPYP) for nonparametric clustering of data with general spatial or temporal interdependencies. The KPYP is constructed by first introducing an infinite sequence of random locations. Then, based on the stick-breaking construction of the Pitman-Yor process, we define a predictor-dependent random probability measure by considering that the discount hyperparameters of the Beta-distributed random weights (stick variables) of the process are not uniform among the weights, but controlled by a kernel function expressing the proximity between the location assigned to each weight and the given predictors.

Efficient and Accurate Estimation of Lipschitz Constants for Deep Neural Networks

Tight estimation of the Lipschitz constant for deep neural networks (DNNs) is useful in many applications ranging from robustness certification of classifiers to stability analysis of closed-loop systems with reinforcement learning controllers. Existing methods in the literature for estimating the Lipschitz constant suffer from either lack of accuracy or poor scalability. In this paper, we present a convex optimization framework to compute guaranteed upper bounds on the Lipschitz constant of DNNs both accurately and efficiently. Our main idea is to interpret activation functions as gradients of convex potential functions. Hence, they satisfy certain properties that can be described by quadratic constraints. This particular description allows us to pose the Lipschitz constant estimation problem as a semidefinite program (SDP). The resulting SDP can be adapted to increase either the estimation accuracy (by capturing the interaction between activation functions of different layers) or scalability (by decomposition and parallel implementation). We illustrate the utility of our approach with a variety of experiments on randomly generated networks and on classifiers trained on the MNIST and Iris datasets. In particular, we experimentally demonstrate that our Lipschitz bounds are the most accurate compared to those in the literature. We also study the impact of adversarial training methods on the Lipschitz bounds of the resulting classifiers and show that our bounds can be used to efficiently provide robustness guarantees.

HyperNTF: A Hypergraph Regularized Nonnegative Tensor Factorization for Dimensionality Reduction

Tensor decomposition is an effective tool for learning multi-way structures and heterogeneous features from high-dimensional data, such as the multi-view images and multichannel electroencephalography (EEG) signals, are often represented by tensors. However, most of tensor decomposition methods are the linear feature extraction techniques, which are unable to reveal the nonlinear structure within high-dimensional data. To address such problem, a lot of algorithms have been proposed for simultaneously performs linear and non-linear feature extraction. A representative algorithm is the Graph Regularized Non-negative Matrix Factorization (GNMF) for image clustering. However, the normal 2-order graph can only models the pairwise similarity of objects, which cannot sufficiently exploit the complex structures of samples. Thus, we propose a novel method, named Hypergraph Regularized Non-negative Tensor Factorization (HyperNTF), which utilizes hypergraph to encode the complex connections among samples and employs the factor matrix corresponding with last mode of Canonical Polyadic (CP) decomposition as low-dimensional representation. Extensive experiments on synthetic manifolds, real-world image datasets, and EEG signals, demonstrating that HyperNTF outperforms the state-of-the-art methods in terms of dimensionality reduction, clustering, and classification.

preprint2020arXiv

Distributional Individual Fairness in Clustering

In this paper, we initiate the study of fair clustering that ensures distributional similarity among similar individuals. In response to improving fairness in machine learning, recent papers have investigated fairness in clustering algorithms and have focused on the paradigm of statistical parity/group fairness. These efforts attempt to minimize bias against some protected groups in the population. However, to the best of our knowledge, the alternative viewpoint of individual fairness, introduced by Dwork et al. (ITCS 2012) in the context of classification, has not been considered for clustering so far. Similar to Dwork et al., we adopt the individual fairness notion which mandates that similar individuals should be treated similarly for clustering problems. We use the notion of $f$-divergence as a measure of statistical similarity that significantly generalizes the ones used by Dwork et al. We introduce a framework for assigning individuals, embedded in a metric space, to probability distributions over a bounded number of cluster centers. The objective is to ensure (a) low cost of clustering in expectation and (b) individuals that are close to each other in a given fairness space

Robust Checkpoint Selection for Multimodal LLMs via Agentic Evaluation and Stability-Aware Ranking

Checkpoint selection for multimodal large language models (MLLMs) presents significant challenges when performance differentials are marginal and evaluation signals are prone to noise. Existing methodologies rely heavily on static benchmarks or pointwise scoring, which frequently misalign with in-the-wild usage and lack robust uncertainty estimation, particularly in OCR-heavy scenarios. In this work, we formulate checkpoint selection as a robust decision problem under evaluation uncertainty. We propose a multi-stage framework that integrates curated real-world data, structured LLM-based judgment, and multi-stage ranking protocols. The evaluation system orchestrates progressive refinement via pointwise filtering, listwise ranking, and pairwise comparison. To enhance reliability, we introduce subsampling-based confidence estimation and a percentile-based scoring formulation that captures distributional characteristics while penalizing tail failures. Furthermore, we demonstrate that data quality, specifically OCR readability, is a critical determinant of evaluation validity.

DP-PCA: Statistically Optimal and Differentially Private PCA

We study the canonical statistical task of computing the principal component from $n$ i.i.d.~data in $d$ dimensions under $(\varepsilon,δ)$-differential privacy. Although extensively studied in literature, existing solutions fall short on two key aspects: ($i$) even for Gaussian data, existing private algorithms require the number of samples $n$ to scale super-linearly with $d$, i.e., $n=Ω(d^{3/2})$, to obtain non-trivial results while non-private PCA requires only $n=O(d)$, and ($ii$) existing techniques suffer from a non-vanishing error even when the randomness in each data point is arbitrarily small. We propose DP-PCA, which is a single-pass algorithm that overcomes both limitations. It is based on a private minibatch gradient ascent method that relies on {\em private mean estimation}, which adds minimal noise required to ensure privacy by adapting to the variance of a given minibatch of gradients. For sub-Gaussian data, we provide nearly optimal statistical error rates even for $n=\tilde O(d)$. Furthermore, we provide a lower bound showing that sub-Gaussian style assumption is necessary in obtaining the optimal error rate.

preprint2014arXiv

Composite Likelihood Estimation for Restricted Boltzmann machines

Learning the parameters of graphical models using the maximum likelihood estimation is generally hard which requires an approximation. Maximum composite likelihood estimations are statistical approximations of the maximum likelihood estimation which are higher-order generalizations of the maximum pseudo-likelihood estimation. In this paper, we propose a composite likelihood method and investigate its property. Furthermore, we apply our composite likelihood method to restricted Boltzmann machines.

preprint2016arXiv

On the Prior Sensitivity of Thompson Sampling

The empirically successful Thompson Sampling algorithm for stochastic bandits has drawn much interest in understanding its theoretical properties. One important benefit of the algorithm is that it allows domain knowledge to be conveniently encoded as a prior distribution to balance exploration and exploitation more effectively. While it is generally believed that the algorithm's regret is low (high) when the prior is good (bad), little is known about the exact dependence. In this paper, we fully characterize the algorithm's worst-case dependence of regret on the choice of prior, focusing on a special yet representative case. These results also provide insights into the general sensitivity of the algorithm to the choice of priors. In particular, with $p$ being the prior probability mass of the true reward-generating model, we prove $O(\sqrt{T/p})$ and $O(\sqrt{(1-p)T})$ regret upper bounds for the bad- and good-prior cases, respectively, as well as \emph{matching} lower bounds. Our proofs rely on the discovery of a fundamental property of Thompson Sampling and make heavy use of martingale theory, both of which appear novel in the literature, to the best of our knowledge.

On Completeness-aware Concept-Based Explanations in Deep Neural Networks

Human explanations of high-level decisions are often expressed in terms of key concepts the decisions are based on. In this paper, we study such concept-based explainability for Deep Neural Networks (DNNs). First, we define the notion of completeness, which quantifies how sufficient a particular set of concepts is in explaining a model's prediction behavior based on the assumption that complete concept scores are sufficient statistics of the model prediction. Next, we propose a concept discovery method that aims to infer a complete set of concepts that are additionally encouraged to be interpretable, which addresses the limitations of existing methods on concept explanations. To define an importance score for each discovered concept, we adapt game-theoretic notions to aggregate over sets and propose ConceptSHAP. Via proposed metrics and user studies, on a synthetic dataset with apriori-known concept explanations, as well as on real-world image and language datasets, we validate the effectiveness of our method in finding concepts that are both complete in explaining the decisions and interpretable. (The code is released at https://github.com/chihkuanyeh/concept_exp)

Towards Building A Group-based Unsupervised Representation Disentanglement Framework

Disentangled representation learning is one of the major goals of deep learning, and is a key step for achieving explainable and generalizable models. A well-defined theoretical guarantee still lacks for the VAE-based unsupervised methods, which are a set of popular methods to achieve unsupervised disentanglement. The Group Theory based definition of representation disentanglement mathematically connects the data transformations to the representations using the formalism of group. In this paper, built on the group-based definition and inspired by the n-th dihedral group, we first propose a theoretical framework towards achieving unsupervised representation disentanglement. We then propose a model, based on existing VAE-based methods, to tackle the unsupervised learning problem of the framework. In the theoretical framework, we prove three sufficient conditions on model, group structure, and data respectively in an effort to achieve, in an unsupervised way, disentangled representation per group-based definition. With the first two of the conditions satisfied and a necessary condition derived for the third one, we offer additional constraints, from the perspective of the group-based definition, for the existing VAE-based models. Experimentally, we train 1800 models covering the most prominent VAE-based methods on five datasets to verify the effectiveness of our theoretical framework. Compared to the original VAE-based methods, these Groupified VAEs consistently achieve better mean performance with smaller variances.

preprint2021arXiv

AutoFITS: Automatic Feature Engineering for Irregular Time Series

A time series represents a set of observations collected over time. Typically, these observations are captured with a uniform sampling frequency (e.g. daily). When data points are observed in uneven time intervals the time series is referred to as irregular or intermittent. In such scenarios, the most common solution is to reconstruct the time series to make it regular, thus removing its intermittency. We hypothesise that, in irregular time series, the time at which each observation is collected may be helpful to summarise the dynamics of the data and improve forecasting performance. We study this idea by developing a novel automatic feature engineering framework, which focuses on extracting information from this point of view, i.e., when each instance is collected. We study how valuable this information is by integrating it in a time series forecasting workflow and investigate how it compares to or complements state-of-the-art methods for regular time series forecasting. In the end, we contribute by providing a novel framework that tackles feature engineering for time series from an angle previously vastly ignored. We show that our approach has the potential to further extract more information about time series that significantly improves forecasting performance.

Diffusion-based Molecule Generation with Informative Prior Bridges

AI-based molecule generation provides a promising approach to a large area of biomedical sciences and engineering, such as antibody design, hydrolase engineering, or vaccine development. Because the molecules are governed by physical laws, a key challenge is to incorporate prior information into the training procedure to generate high-quality and realistic molecules. We propose a simple and novel approach to steer the training of diffusion-based generative models with physical and statistics prior information. This is achieved by constructing physically informed diffusion bridges, stochastic processes that guarantee to yield a given observation at the fixed terminal time. We develop a Lyapunov function based method to construct and determine bridges, and propose a number of proposals of informative prior bridges for both high-quality molecule generation and uniformity-promoted 3D point cloud generation. With comprehensive experiments, we show that our method provides a powerful approach to the 3D generation task, yielding molecule structures with better quality and stability scores and more uniformly distributed point clouds of high qualities.

Generalization Bounds using Lower Tail Exponents in Stochastic Optimizers

Despite the ubiquitous use of stochastic optimization algorithms in machine learning, the precise impact of these algorithms and their dynamics on generalization performance in realistic non-convex settings is still poorly understood. While recent work has revealed connections between generalization and heavy-tailed behavior in stochastic optimization, this work mainly relied on continuous-time approximations; and a rigorous treatment for the original discrete-time iterations is yet to be performed. To bridge this gap, we present novel bounds linking generalization to the lower tail exponent of the transition kernel associated with the optimizer around a local minimum, in both discrete- and continuous-time settings. To achieve this, we first prove a data- and algorithm-dependent generalization bound in terms of the celebrated Fernique-Talagrand functional applied to the trajectory of the optimizer. Then, we specialize this result by exploiting the Markovian structure of stochastic optimizers, and derive bounds in terms of their (data-dependent) transition kernels. We support our theory with empirical results from a variety of neural networks, showing correlations between generalization error and lower tail exponents.

Preference-Based Self-Distillation: Beyond KL Matching via Reward Regularization

On-policy distillation is an efficient alternative to reinforcement learning, offering dense token-level training signals. However, its reliance on a stronger external teacher has driven recent work on on-policy self-distillation, where the same model serves as both teacher and student under different prompt contexts. Yet, existing self-distillation methods largely reduce learning to KL matching toward the context-augmented teacher model. This approach often suffers from training instability and can degrade reasoning performance over time. Moreover, self-distillation from the same model with prompt augmentation lacks the exploratory diversity provided by a genuine external teacher. To address these limitations, we move beyond fixed-teacher KL matching and propose \textbf{P}reference-\textbf{B}ased \textbf{S}elf-\textbf{D}istillation (\textbf{PBSD}), which revisits on-policy self-distillation through a reward-regularized perspective. Instead of directly matching the teacher distribution, we derive a reward-regularized objective whose analytic optimum is a reward-reweighted teacher distribution, yielding a target policy provably superior to the original teacher under this objective. Practically, PBSD optimizes preference gaps between teacher and student samples while maintaining on-policy student sampling. We support this framework with a statistical analysis of the induced preference-learning problem, formally establishing when on policy self-distillation is preferable to learning from an external teacher in our setting. Experiments on mathematical reasoning and tool-use benchmarks across multiple model scales demonstrate that PBSD consistently achieves the strongest average performance among comparable baselines, showing improved training stability over prior self-distillation baselines while preserving token efficiency.

preprint2014arXiv

A Nonparametric Adaptive Nonlinear Statistical Filter

We use statistical learning methods to construct an adaptive state estimator for nonlinear stochastic systems. Optimal state estimation, in the form of a Kalman filter, requires knowledge of the system's process and measurement uncertainty. We propose that these uncertainties can be estimated from (conditioned on) past observed data, and without making any assumptions of the system's prior distribution. The system's prior distribution at each time step is constructed from an ensemble of least-squares estimates on sub-sampled sets of the data via jackknife sampling. As new data is acquired, the state estimates, process uncertainty, and measurement uncertainty are updated accordingly, as described in this manuscript.

preprint2021arXiv

Meta-Learning with MAML on Trees

In meta-learning, the knowledge learned from previous tasks is transferred to new ones, but this transfer only works if tasks are related. Sharing information between unrelated tasks might hurt performance, and it is unclear how to transfer knowledge across tasks with a hierarchical structure. Our research extends a model agnostic meta-learning model, MAML, by exploiting hierarchical task relationships. Our algorithm, TreeMAML, adapts the model to each task with a few gradient steps, but the adaptation follows the hierarchical tree structure: in each step, gradients are pooled across tasks clusters, and subsequent steps follow down the tree. We also implement a clustering algorithm that generates the tasks tree without previous knowledge of the task structure, allowing us to make use of implicit relationships between the tasks. We show that the new algorithm, which we term TreeMAML, performs better than MAML when the task structure is hierarchical for synthetic experiments. To study the performance of the method in real-world data, we apply this method to Natural Language Understanding, we use our algorithm to finetune Language Models taking advantage of the language phylogenetic tree. We show that TreeMAML improves the state of the art results for cross-lingual Natural Language Inference. This result is useful, since most languages in the world are under-resourced and the improvement on cross-lingual transfer allows the internationalization of NLP models. This results open the window to use this algorithm in other real-world hierarchical datasets.

preprint2013arXiv

Induction of Selective Bayesian Classifiers

In this paper, we examine previous work on the naive Bayesian classifier and review its limitations, which include a sensitivity to correlated features. We respond to this problem by embedding the naive Bayesian induction scheme within an algorithm that c arries out a greedy search through the space of features. We hypothesize that this approach will improve asymptotic accuracy in domains that involve correlated features without reducing the rate of learning in ones that do not. We report experimental results on six natural domains, including comparisons with decision-tree induction, that support these hypotheses. In closing, we discuss other approaches to extending naive Bayesian classifiers and outline some directions for future research.

Cost Effective MLaaS Federation: A Combinatorial Reinforcement Learning Approach

With the advancement of deep learning techniques, major cloud providers and niche machine learning service providers start to offer their cloud-based machine learning tools, also known as machine learning as a service (MLaaS), to the public. According to our measurement, for the same task, these MLaaSes from different providers have varying performance due to the proprietary datasets, models, etc. Federating different MLaaSes together allows us to improve the analytic performance further. However, naively aggregating results from different MLaaSes not only incurs significant momentary cost but also may lead to sub-optimal performance gain due to the introduction of possible false-positive results. In this paper, we propose Armol, a framework to federate the right selection of MLaaS providers to achieve the best possible analytic performance. We first design a word grouping algorithm to unify the output labels across different providers. We then present a deep combinatorial reinforcement learning based-approach to maximize the accuracy while minimizing the cost. The predictions from the selected providers are then aggregated together using carefully chosen ensemble strategies. The real-world trace-driven evaluation further demonstrates that Armol is able to achieve the same accuracy results with $67\%$ less inference cost.

Measuring the False Sense of Security

Recently, several papers have demonstrated how widespread gradient masking is amongst proposed adversarial defenses. Defenses that rely on this phenomenon are considered failed, and can easily be broken. Despite this, there has been little investigation into ways of measuring the phenomenon of gradient masking and enabling comparisons of its extent amongst different networks. In this work, we investigate gradient masking under the lens of its mensurability, departing from the idea that it is a binary phenomenon. We propose and motivate several metrics for it, performing extensive empirical tests on defenses suspected of exhibiting different degrees of gradient masking. These are computationally cheaper than strong attacks, enable comparisons between models, and do not require the large time investment of tailor-made attacks for specific models. Our results reveal metrics that are successful in measuring the extent of gradient masking across different networks