Improving Multilingual Language Models by Aligning Representations through Steering

Humans excel at analogical reasoning - applying knowledge from one task to a related one with minimal relearning. In contrast, reinforcement learning (RL) agents typically require extensive retraining even when new tasks share structural similarities with previously learned ones. In this work, we propose MAGIK, a novel framework that enables RL agents to transfer knowledge to analogous tasks without interacting with the target environment. Our approach leverages an imagination mechanism to map entities in the target task to their analogues in the source domain, allowing the agent to reuse its original policy. Experiments on custom MiniGrid and MuJoCo tasks show that MAGIK achieves effective zero-shot transfer using only a small number of human-labelled examples. We compare our approach to related baselines and highlight how it offers a novel and effective mechanism for knowledge transfer via imagination-based analogy mapping.

Aly M. Kassem, Omar Mahmoud, Niloofar Mireshghallah, Hyunwoo Kim, Yulia Tsvetkov, Yejin Choi, Sherif Saad, Santu Rana. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

This paper explores the use of Hermite transform based reproducing kernel Banach space methods to construct exact or un-approximated models of feedforward neural networks of arbitrary width, depth and topology, including ResNet and Transformers networks, assuming only a feedforward topology, finite energy activations and finite (spectral-) norm weights and biases. Using this model, two straightforward but surprisingly tight bounds on Rademacher complexity are derived, precisely (1) a general bound that is width-independent and scales exponentially with depth; and (2) a width- and depth-independent bound for networks with appropriately constrained (below threshold) weights and biases.

Autonomously learning diverse behaviours without an extrinsic reward signal has been a problem of interest in reinforcement learning. However, the nature of learning in such mechanisms is unconstrained, often resulting in the accumulation of several unusable, unsafe or misaligned skills. In order to avoid such issues and to ensure the discovery of safe and human-aligned skills, it is necessary to incorporate humans into the unsupervised training process, which remains a largely unexplored topic. In this work, we propose Controlled Diversity with Preference (CDP), a novel, collaborative human-guided mechanism for an agent to learn a set of skills that is diverse as well as desirable. The key principle is to restrict the discovery of skills to regions that are deemed to be desirable as per a preference model trained using human preference labels on trajectory pairs. We evaluate our approach on 2D navigation and Mujoco environments and demonstrate the ability to discover diverse, yet desirable skills. We also provide principled guidelines for selecting suitable hyperparameter values along with comprehensive sensitivity analyses of the various factors influencing the performance of our approach.

Abstract Large language models (LLMs) have recently demonstrated their impressive ability to provide context-aware responses via text. This ability could potentially be used to predict plausible solutions in sequential decision making tasks pertaining to pattern completion. For example, by observing a partial stack of cubes, LLMs can predict the correct sequence in which the remaining cubes should be stacked by extrapolating the observed patterns (e.g., cube sizes, colors or other attributes) in the partial stack. In this work, we introduce LaGR (language-guided reinforcement learning), which uses this predictive ability of LLMs to propose solutions to tasks that have been partially completed by a primary reinforcement learning (RL) agent, in order to subsequently guide the latter’s training. However, as RL training is generally not sample-efficient, deploying this approach would inherently imply that the LLM be repeatedly queried for solutions; a process that can be expensive and infeasible. To address this issue, we introduce SEQ (sample-efficient querying), where we simultaneously train a secondary RL agent to decide when the LLM should be queried for solutions. Specifically, we use the quality of the solutions emanating from the LLM as the reward to train this agent. We show that our proposed framework LaGR-SEQ enables more efficient primary RL training, while simultaneously minimizing the number of queries to the LLM. We demonstrate our approach on a series of tasks and highlight the advantages of our approach, along with its limitations and potential future research directions.

Automated outdoor vision-based applications have become increasingly in demand for day-to-day life. Bad weather like haze, rain, snow, etc. may limit the reliability of these applications due to degradation in the overall video quality. So, there is a dire need to pre-process the weather-degraded videos before they are fed to downstream applications. Researchers generally adopt synthetically generated paired hazy frames for learning the task of video de-hazing. The models trained solely on synthetic data may have limited performance on different types of real-world hazy scenarios due to significant domain gap between synthetic and real-world hazy videos. One possible solution is to prove the generalization ability by training on unpaired data for video de-hazing. Some unpaired learning approaches are proposed for single image de-hazing. However, these unpaired single image de-hazing approaches compromise the performance in terms of temporal consistency, which is important for video de-hazing tasks. With this motivation, we have proposed a lightweight and temporally consistent architecture for video de-hazing tasks. To achieve this, diverse receptive and multi-scale features at various input resolutions are mixed and aggregated with multi-kernel attention to extract significant haze information. Furthermore, we propose a recurrent multi-attentive feature alignment concept to maintain temporal consistency with recurrent feedback of previously restored frames for temporal consistent video restoration . Comprehensive experiments are conducted on real-world and synthetic video databases (REVIDE and RSA100Haze). Both the qualitative and quantitative results show significant improvement of the proposed network with better temporal consistency over state-of-the-art methods for detailed video restoration in hazy weather. Source code is available at: https://github.com/pwp1208/UnpairedVideoDehazing .

In this paper, we study the “dataset bias” problem from a statistical standpoint, and identify the main cause of the problem as the strong correlation between a class attribute u and a non-class attribute b in the input x, represented by p(u|b) differing significantly from p(u). Since p(u|b) appears as part of the sampling distributions in the standard maximum log-likelihood (MLL) objective, a model trained on a biased dataset via MLL inherently incorporates such correlation into its parameters, leading to poor generalization to unbiased test data. From this observation, we propose to mitigate dataset bias via either weighting the objective of each sample n by 1 / p(un|bn) or sampling that sample with a weight proportional to 1 / p(un|bn). While both methods are statistically equivalent, the former proves more stable and effective in practice. Additionally, we establish a connection between our debiasing approach and causal reasoning, reinforcing our method’s theoretical foundation. However, when the bias label is unavailable, computing p(u|b) exactly is difficult. To overcome this challenge, we propose to approximate 1 / p(u|b) using a biased classifier trained with “bias amplification” losses. Extensive experiments on various biased datasets demonstrate the superiority of our method over existing debiasing techniques in most settings, validating our theoretical analysis.

Federated Learning systems are increasingly subjected to a multitude of model poisoning attacks from clients. Among these, edge-case attacks that target a small fraction of the input space are nearly impossible to detect using existing defenses, leading to a high attack success rate. We propose an effective defense using an external defense dataset, which provides information about the attack target. The defense dataset contains a mix of poisoned and clean examples, with only a few known to be clean. The proposed method, DataDefense, uses this dataset to learn a poisoned data detector model which marks each example in the defense dataset as poisoned or clean. It also learns a client importance model that estimates the probability of a client update being malicious. The global model is then updated as a weighted average of the client models’ updates. The poisoned data detector and the client importance model parameters are updated using an alternating minimization strategy over the Federated Learning rounds. Extensive experiments on standard attack scenarios demonstrate that DataDefense can defend against model poisoning attacks where other state-of-the-art defenses fail. In particular, DataDefense is able to reduce the attack success rate by at least ∼ 40% on standard attack setups and by more than 80% on some setups. Furthermore, DataDefense requires very few defense examples (as few as five) to achieve a near-optimal reduction in attack success rate.

Reward-optimal policies obtained by training deep reinforcement learning agents may not be aligned with one’s personal preferences. Rectifying this by retraining the agent with a user-specific reward function is impractical, as such functions are not readily available. In addition, retraining is associated with high costs. Instead, we propose to adapt via policy fusion, the already trained policy with the user’s intent, which we in turn infer from trajectory-level feedback. We design the policy fusion process to be dynamic, such that the resulting policy is neither dominated by the task goals nor the user needs. We empirically demonstrate that our method consistently balances these objectives across various environments.

Recent data-privacy laws have sparked interest in machine unlearning, which involves removing the effect of specific training samples from a learnt model as if they were never present in the original training dataset. The challenge of machine unlearning is to discard information about the "forget" data in the learnt model without altering the knowledge about the remaining dataset and to do so more efficiently than the naive retraining approach. To achieve this, we adopt a projected-gradient based learning method, named as Projected-Gradient Unlearning (PGU), in which the model takes steps in the orthogonal direction to the gradient subspaces deemed unimportant for the retaining dataset, so as to its knowledge is preserved. By utilizing Stochastic Gradient Descent (SGD) to update the model weights, our method can efficiently scale to any model and dataset size. We provide empirically evidence to demonstrate that our unlearning method can produce models that behave similar to models retrained from scratch across various metrics even when the training dataset is no longer accessible. Our code is available at https://github.com/hnanhtuan/projected_gradient_unlearning.

Weather degraded conditions such as rain, haze, snow, etc. may degrade the performance of most computer vision systems. Therefore, effective restoration of multi-weather degraded images is an essential prerequisite for successful functioning of such systems. The current multi-weather image restoration approaches utilize a model that is trained on a combined dataset consisting of individual images for rainy, snowy, and hazy weather degradations. These methods may face challenges when dealing with real-world situations where the images may have multiple, more intricate weather conditions. To address this issue, we propose a domain translation-based unified method for multi-weather image restoration. In this approach, the proposed network learns multiple weather degradations simultaneously, making it immune for real-world conditions. Specifically, we first propose an instance-level domain (weather) translation with multi-attentive feature learning approach to get different weather-degraded variants of the same scenario. Next, the original and translated images are used as input to the proposed novel multi-weather restoration network which utilizes a progressive multi-domain deformable alignment (PMDA) with cascaded multi-head attention (CMA). The proposed PMDA facilitates the restoration network to learn weather-invariant clues effectively. Further, PMDA and respective decoder features are merged via proposed CMA module for restoration. Extensive experimental results on synthetic and real-world hazy, rainy, and snowy image databases clearly demonstrate that our model outperforms the state-of-the-art multi-weather image restoration methods. Code is available at https://github.com/pwp1208/Domain_Translation_Multi-weather_Restoration.

The study of Neural Tangent Kernels (NTKs) has provided much needed insight into convergence and generalization properties of neural networks in the over-parametrized (wide) limit by approximating the network using a first-order Taylor expansion with respect to its weights in the neighborhood of their initialization values. This allows neural network training to be analyzed from the perspective of reproducing kernel Hilbert spaces (RKHS), which is informative in the over-parametrized regime, but a poor approximation for narrower networks as the weights change more during training. Our goal is to extend beyond the limits of NTK toward a more general theory. We construct an exact power-series representation of the neural network in a finite neighborhood of the initial weights as an inner product of two feature maps, respectively from data and weight-step space, to feature space, allowing neural network training to be analyzed from the perspective of reproducing kernel Banach space (RKBS). We prove that, regardless of width, the training sequence produced by gradient descent can be exactly replicated by regularized sequential learning in RKBS. Using this, we present novel bound on uniform convergence where the iterations count and learning rate play a central role, giving new theoretical insight into neural network training.

The optimistic nature of the Q−learning target leads to an overestimation bias, which is an inherent problem associated with standard Q−learning. Such a bias fails to account for the possibility of low returns, particularly in risky scenarios. However, the existence of biases, whether overestimation or underestimation, need not necessarily be undesirable. In this paper, we analytically examine the utility of biased learning, and show that specific types of biases may be preferable, depending on the scenario. Based on this finding, we design a novel reinforcement learning algorithm, Balanced Q-learning, in which the target is modified to be a convex combination of a pessimistic and an optimistic term, whose associated weights are determined online, analytically. Such a balanced target inherently promotes risk-averse behavior, which we examine through the lens of the agent's exploration. We prove the convergence of this algorithm in a tabular setting, and empirically demonstrate its consistently good learning performance in various environments.

The expected improvement (EI) algorithm is one of the most popular strategies for optimization under uncertainty due to its simplicity and efficiency. Despite its popularity, the theoretical aspects of this algorithm have not been properly analyzed. In particular, whether in the noisy setting, the EI strategy with a standard incumbent converges is still an open question of the Gaussian process bandit optimization problem. We aim to answer this question by proposing a variant of EI with a standard incumbent defined via the GP predictive mean. We prove that our algorithm converges, and achieves a cumulative regret bound of $\mathcal O(\gamma_T\sqrt{T})$, where $\gamma_T$ is the maximum information gain between $T$ observations and the Gaussian process model. Based on this variant of EI, we further propose an algorithm called Improved GP-EI that converges faster than previous counterparts. In particular, our proposed variants of EI do not require the knowledge of the RKHS norm and the noise’s sub-Gaussianity parameter as in previous works. Empirical validation in our paper demonstrates the effectiveness of our algorithms compared to several baselines.

Adapting an agent’s behaviour to new environments has been one of the primary focus areas of physics based reinforcement learning. Although recent approaches such as universal policy networks partially address this issue by enabling the storage of multiple policies trained in simulation on a wide range of dynamic/latent factors, efficiently identifying the most appropriate policy for a given environment remains a challenge. In this work, we propose a Gaussian Process-based prior learned in simulation, that captures the likely performance of a policy when transferred to a previously unseen environment. We integrate this prior with a Bayesian Optimisation-based policy search process to improve the efficiency of identifying the most appropriate policy from the universal policy network. We empirically evaluate our approach in a range of continuous and discrete control environments, and show that it outperforms other competing baselines.

Sim2real transfer is primarily concerned with transferring policies trained in simulation to potentially noisy real world environments. A common problem associated with sim2real transfer is estimating the real-world environmental parameters to ground the simulated environment to. Although existing methods such as Domain Randomisation (DR) can produce robust policies by sampling from a distribution of parameters during training, there is no established method for identifying the parameters of the corresponding distribution for a given real-world setting. In this work, we propose Uncertainty-aware policy search (UncAPS), where we use Universal Policy Network (UPN) to store simulation-trained task-specific policies across the full range of environmental parameters and then subsequently employ robust Bayesian optimisation to craft robust policies for the given environment by combining relevant UPN policies in a DR like fashion. Such policy-driven grounding is expected to be more efficient as it estimates only task-relevant sets of parameters. Further, we also account for the estimation uncertainties in the search process to produce policies that are robust against both aleatoric and epistemic uncertainties. We empirically evaluate our approach in a range of noisy, continuous control environments, and show its improved performance compared to competing baselines.

Data-free Knowledge Distillation (DFKD) has attracted attention recently thanks to its appealing capability of transferring knowledge from a teacher network to a student network without using training data. The main idea is to use a generator to synthesize data for training the student. As the generator gets updated, the distribution of synthetic data will change. Such distribution shift could be large if the generator and the student are trained adversarially, causing the student to forget the knowledge it acquired at the previous steps. To alleviate this problem, we propose a simple yet effective method called Momentum Adversarial Distillation (MAD) which maintains an exponential moving average (EMA) copy of the generator and uses synthetic samples from both the generator and the EMA generator to train the student. Since the EMA generator can be considered as an ensemble of the generator's old versions and often undergoes a smaller change in updates compared to the generator, training on its synthetic samples can help the student recall the past knowledge and prevent the student from adapting too quickly to the new updates of the generator. Our experiments on six benchmark datasets including big datasets like ImageNet and Places365 demonstrate the superior performance of MAD over competing methods for handling the large distribution shift problem. Our method also compares favorably to existing DFKD methods and even achieves state-of-the-art results in some cases.

The expected improvement (EI) is a popular technique to handle the tradeoff between exploration and exploitation under uncertainty. This technique has been widely used in Bayesian optimization but it is not applicable for the contextual bandit problem which is a generalization of the standard bandit and Bayesian optimization. In this paper, we initiate and study the EI technique for contextual bandits from both theoretical and practical perspectives. We propose two novel EI-based algorithms, one when the reward function is assumed to be linear and the other for more general reward functions. With linear reward functions, we demonstrate that our algorithm achieves a near-optimal regret. Notably, our regret improves that of LinTS \cite{agrawal13} by a factor $\sqrt{d}$ while avoiding to solve a NP-hard problem at each iteration as in LinUCB \cite{Abbasi11}. For more general reward functions which are modeled by deep neural networks, we prove that our algorithm achieves a $\tilde{\mathcal O} (\tilde{d}\sqrt{T})$ regret, where $\tilde{d}$ is the effective dimension of a neural tangent kernel (NTK) matrix, and $T$ is the number of iterations. Our experiments on various benchmark datasets show that both proposed algorithms work well and consistently outperform existing approaches, especially in high dimensions.

Abstract Human-AI collaboration looks at harnessing the complementary strengths of both humans and AI. We propose a new method for human-AI collaboration in Bayesian optimisation where the optimum is mainly pursued by the Bayesian optimisation algorithm following complex computation, whilst getting occasional help from the accompanying expert having a deeper knowledge of the underlying physical phenomenon. We expect experts to have some understanding of the correlation structures of the experimental system, but not the location of the optimum. The expert provides feedback by either changing the current recommendation or providing her belief on the good and bad regions of the search space based on the current observations. Our proposed method takes such feedback to build a model that aligns with the expert’s model and then uses it for optimisation. We provide theoretical underpinning on why such an approach may be more efficient than the one without expert’s feedback. The empirical results show the robustness and superiority of our method with promising efficiency gains.

Level Set Estimation (LSE) is an important problem with applications in various fields such as material design, biotechnology, machine operational testing, etc. Existing techniques suffer from the scalability issue, that is, these methods do not work well with high dimensional inputs. This paper proposes novel methods to solve the high dimensional LSE problems using Bayesian Neural Networks. In particular, we consider two types of LSE problems: (1) explicit LSE problem where the threshold level is a fixed user-specified value, and, (2) implicit LSE problem where the threshold level is defined as a percentage of the (unknown) maximum of the objective function. For each problem, we derive the corresponding theoretic information based acquisition function to sample the data points so as to maximally increase the level set accuracy. Furthermore, we also analyse theoretical time complexity of our proposed acquisition functions, and suggest a practical methodology to efficiently tune the network hyper-parameters to achieve high model accuracy. Numerical experiments on both synthetic and real-world datasets show that our proposed methods can achieve better results compared to existing state-of-the-art approaches.

Transfer in reinforcement learning is usually achieved through generalisation across tasks. Whilst many studies have investigated transferring knowledge when the reward function changes, they have assumed that the dynamics of the environments remain consistent. Many real-world RL problems require transfer among environments with different dynamics. To address this problem, we propose an approach based on successor features in which we model successor feature functions with Gaussian Processes permitting the source successor features to be treated as noisy measurements of the target successor feature function. Our theoretical analysis proves the convergence of this approach as well as the bounded error on modelling successor feature functions with Gaussian Processes in environments with both different dynamics and rewards. We demonstrate our method on benchmark datasets and show that it outperforms current baselines.

Bayesian optimisation (BO) is a well known algorithm for finding the global optimum of expensive, black-box functions. The current practical BO algorithms have regret bounds ranging from $\mathcal{O}(\frac{logN}{\sqrt{N}})$ to $\mathcal O(e^{-\sqrt{N}})$, where $N$ is the number of evaluations. This paper explores the possibility of improving the regret bound in the noise-free setting by intertwining concepts from BO and optimistic optimisation methods which are based on partitioning the search space. We propose the BOO algorithm, a first practical approach which can achieve an exponential regret bound with order $\mathcal O(N^{-\sqrt{N}})$ under the assumption that the objective function is sampled from a Gaussian process with a Matérn kernel with smoothness parameter $\nu > 4 +\frac{D}{2}$, where $D$ is the number of dimensions. We perform experiments on optimisation of various synthetic functions and machine learning hyperparameter tuning tasks and show that our algorithm outperforms baselines.

Traditional methods for kernel selection rely on parametric kernel functions or a combination thereof and although the kernel hyperparameters are tuned, these methods often provide sub-optimal results due to the limitations induced by the parametric forms. In this paper, we propose a novel formulation for kernel selection using efficient Bayesian optimisation to find the best fitting non-parametric kernel. The kernel is expressed using a linear combination of functions sampled from a prior Gaussian Process (GP) defined by a hyperkernel. We also provide a mechanism to ensure the positive definiteness of the Gram matrix constructed using the resultant kernels. Our experimental results on GP regression and Support Vector Machine (SVM) classification tasks involving both synthetic functions and several real-world datasets show the superiority of our approach over the state-of-the-art.

Many real-world functions are defined over both categorical and category-specific continuous variables and thus cannot be optimized by traditional Bayesian optimization (BO) methods. To optimize such functions, we propose a new method that formulates the problem as a multi-armed bandit problem, wherein each category corresponds to an arm with its reward distribution centered around the optimum of the objective function in continuous variables. Our goal is to identify the best arm and the maximizer of the corresponding continuous function simultaneously. Our algorithm uses a Thompson sampling scheme that helps connecting both multi-arm bandit and BO in a unified framework. We extend our method to batch BO to allow parallel optimization when multiple resources are available. We theoretically analyze our method for convergence and prove sub-linear regret bounds. We perform a variety of experiments: optimization of several benchmark functions, hyper-parameter tuning of a neural network, and automatic selection of the best machine learning model along with its optimal hyper-parameters (a.k.a automated machine learning). Comparisons with other methods demonstrate the effectiveness of our proposed method.

Scaling Bayesian optimisation (BO) to high-dimensional search spaces is a active and open research problems particularly when no assumptions are made on function structure. The main reason is that at each iteration, BO requires to find global maximisation of acquisition function, which itself is a non-convex optimization problem in the original search space. With growing dimensions, the computational budget for this maximisation gets increasingly short leading to inaccurate solution of the maximisation. This inaccuracy adversely affects both the convergence and the efficiency of BO. We propose a novel approach where the acquisition function only requires maximisation on a discrete set of low dimensional subspaces embedded in the original high-dimensional search space. Our method is free of any low dimensional structure assumption on the function unlike many recent high-dimensional BO methods. Optimising acquisition function in low dimensional subspaces allows our method to obtain accurate solutions within limited computational budget. We show that in spite of this convenience, our algorithm remains convergent. In particular, cumulative regret of our algorithm only grows sub-linearly with the number of iterations. More importantly, as evident from our regret bounds, our algorithm provides a way to trade the convergence rate with the number of subspaces used in the optimisation. Finally, when the number of subspaces is "sufficiently large", our algorithm's cumulative regret is at most O*(√TγT) as opposed to O*(√DTγT) for the GP-UCB of Srinivas et al. (2012), reducing a crucial factor √D where D being the dimensional number of input space. We perform empirical experiments to evaluate our method extensively, showing that its sample efficiency is better than the existing methods for many optimisation problems involving dimensions up to 5000.

Bayesian quadrature optimization (BQO) maximizes the expectation of an expensive black-box integrand taken over a known probability distribution. In this work, we study BQO under distributional uncertainty in which the underlying probability distribution is unknown except for a limited set of its i.i.d samples. A standard BQO approach maximizes the Monte Carlo estimate of the true expected objective given the fixed sample set. Though Monte Carlo estimate is unbiased, it has high variance given a small set of samples; thus can result in a spurious objective function. We adopt the distributionally robust optimization perspective to this problem by maximizing the expected objective under the most adversarial distribution. In particular, we propose a novel posterior sampling based algorithm, namely distributionally robust BQO (DRBQO) for this purpose. We demonstrate the empirical effectiveness of our proposed framework in synthetic and real-world problems, and characterize its theoretical convergence via Bayesian regret.

Bayesian optimization (BO) is a widely-used method for optimizing expensive (to evaluate) problems. At the core of most BO methods is the modeling of the objective function using a Gaussian Process (GP) whose covariance is selected from a set of standard covariance functions. From a weight-space view, this models the objective as a linear function in a feature space implied by the given covariance $K$, with an arbitrary Gaussian weight prior ${\bf w} \sim ormdist ({\bf 0},{\bf I})$. In many practical applications there is data available that has a similar (covariance) structure to the objective, but which, having different form, cannot be used directly in standard transfer learning. In this paper we show how such auxiliary data may be used to construct a GP covariance corresponding to a more appropriate weight prior for the objective function. Building on this, we show that we may accelerate BO by modeling the objective function using this (learned) weight prior, which we demonstrate on both test functions and a practical application to short-polymer fibre manufacture.

Abstract Interpretability allows the domain-expert to directly evaluate the model’s relevance and reliability, a practice that offers assurance and builds trust. In the healthcare setting, interpretable models should implicate relevant biological mechanisms independent of technical factors like data pre-processing. We define personalized interpretability as a measure of sample-specific feature attribution, and view it as a minimum requirement for a precision health model to justify its conclusions. Some health data, especially those generated by high-throughput sequencing experiments, have nuances that compromise precision health models and their interpretation. These data are compositional, meaning that each feature is conditionally dependent on all other features. We propose the Deep Compositional Data Analysis (DeepCoDA) framework to extend precision health modelling to high-dimensional compositional data, and to provide personalized interpretability through patient-specific weights. Our architecture maintains state-of-the-art performance across 25 real-world data sets, all while producing interpretations that are both personalized and fully coherent for compositional data.

In this paper we propose a data-efficient Bayesian active learning framework for factor screening, which is important when dealing with systems which are expensive to evaluate, such as combat simulations. We use Gaussian Process meta-modelling with the Automatic Relevance Determination covariance kernel, which measures the importance of each factor by the inverse of their associated length-scales in the kernel. This importance measures the degree of non-linearity in the simulation response with respect to the corresponding factor. We initially place a prior over the length-scale values, then use the estimated posterior to select the next datum to simulate which maximises the mutual entropy between the length-scales and the unknown simulation response. Our goal-driven Bayesian active learning strategy ensures that we are data-efficient in discovering the correct values of the length-scales compared to either a random-sampling or uncertainty-sampling based approach. We apply our method to an expensive combat simulation and demonstrate the superiority of our approach.

In order to improve the performance of Bayesian optimisation, we develop a modified Gaussian process upper confidence bound (GP-UCB) acquisition function. This is done by sampling the exploration-exploitation trade-off parameter from a distribution. We prove that this allows the expected trade-off parameter to be altered to better suit the problem without compromising a bound on the function's Bayesian regret. We also provide results showing that our method achieves better performance than GP-UCB in a range of real-world and synthetic problems.

Prior access to domain knowledge could significantly improve the performance of a reinforcement learning agent. In particular, it could help agents avoid potentially catastrophic exploratory actions, which would otherwise have to be experienced during learning. In this work, we identify consistently undesirable actions in a set of previously learned tasks, and use pseudo-rewards associated with them to learn a prior policy. In addition to enabling safer exploratory behaviors in subsequent tasks in the domain, we show that these priors are transferable to similar environments, and can be learned off-policy and in parallel with the learning of other tasks in the domain. We compare our approach to established, state-of-the-art algorithms in both discrete as well as continuous environments, and demonstrate that it exhibits a safer exploratory behavior while learning to perform arbitrary tasks in the domain. We also present a theoretical analysis to support these results, and briefly discuss the implications and some alternative formulations of this approach, which could also be useful in certain scenarios.

Bayesian optimisation is a popular method for efficient optimisation of expensive black-box functions. Traditionally, BO assumes that the search space is known. However, in many problems, this assumption does not hold. To this end, we propose a novel BO algorithm which expands (and shifts) the search space over iterations based on controlling the expansion rate thought a \emph{hyperharmonic series}. Further, we propose another variant of our algorithm that scales to high dimensions. We show theoretically that for both our algorithms, the cumulative regret grows at sub-linear rates. Our experiments with synthetic and real-world optimisation tasks demonstrate the superiority of our algorithms over the current state-of-the-art methods for Bayesian optimisation in unknown search space.

Real world experiments are expensive, and thus it is important to reach a target in a minimum number of experiments. Experimental processes often involve control variables that change over time. Such problems can be formulated as functional optimisation problem. We develop a novel Bayesian optimisation framework for such functional optimisation of expensive black-box processes. We represent the control function using Bernstein polynomial basis and optimise in the coefficient space. We derive the theory and practice required to dynamically adjust the order of the polynomial degree, and show how prior information about shape can be integrated. We demonstrate the effectiveness of our approach for short polymer fibre design and optimising learning rate schedules for deep networks.

Bayesian optimization (BO) is concerned with efficient optimization using probabilistic methods. Predictive entropy search (PES) is a popular and successful BO strategy to find a point that maximizes the information gained about the optima location of an unknown function. Since the PES analytical form is intractable, it requires approximations and is computationally expensive. These approximations may degrade PES performance in terms of accuracy and efficiency. In this paper, we propose an alternative scheme - predictive variance reduction search (PVRS) - to find a point that maximally reduces the uncertainty at the perceived optima locations. The optimization converges to the true optimum when the uncertainty at all perceived optima locations is vanished. Our novel modification is beneficial in two ways. First, PVRS can be computed in closed-form, unlike the approximations made in PES. Second, PVRS is simple and easy to implement. As a result, the proposed PVRS gains huge speed up for scalable BO whilst showing favorable optimization efficiency. Furthermore, we extend our PVRS framework for batch setting where we select multiple experiments for parallel evaluations at each iteration. Empirically, we demonstrate the effectiveness of the PVRS on both benchmark functions and real-world applications in standard and batch BO settings.

We present a multi-objective Bayesian optimisation algorithm that allows the user to express preference-order constraints on the objectives of the type objective A is more important than objective B. These preferences are defined based on the stability of the obtained solutions with respect to preferred objective functions. Rather than attempting to find a representative subset of the complete Pareto front, our algorithm selects those Pareto-optimal points that satisfy these constraints. We formulate a new acquisition function based on expected improvement in dominated hypervolume (EHI) to ensure that the subset of Pareto front satisfying the constraints is thoroughly explored. The hypervolume calculation is weighted by the probability of a point satisfying the constraints from a gradient Gaussian Process model. We demonstrate our algorithm on both synthetic and real-world problems.

Applying Bayesian optimization in problems wherein the search space is unknown is challenging. To address this problem, we propose a systematic volume expansion strategy for the Bayesian optimization. We devise a strategy to guarantee that in iterative expansions of the search space, our method can find a point whose function value within epsilon of the objective function maximum. Without the need to specify any parameters, our algorithm automatically triggers a minimal expansion required iteratively. We derive analytic expressions for when to trigger the expansion and by how much to expand. We also provide theoretical analysis to show that our method achieves epsilon-accuracy after a finite number of iterations. We demonstrate our method on both benchmark test functions and machine learning hyper-parameter tuning tasks and demonstrate that our method outperforms baselines.

Designing new physical products and processes requires enormous experimentation. The scientific simulators play a fundamental role for such design tasks. To design a new product with certain target characteristics, a search is performed in the design space by trying out a large number of design combinations through simulators before reaching to the target characteristics. However, searching for the target design using simulators is generally expensive and becomes prohibitive when the target is either revised or only partially specified. To address this problem, we use a machine learning model to predict the design in single step using the target product specifications as input. We overcome two technical challenges: the first caused due to one-to-many mapping when learning the inverse problem and the second caused due to a user specifying the target specifications only partially. We unify a conditional variational auto-encoder model (to address the partial target specification) with mixture density networks (to address the one-to-many mapping) and train an end-to-end model to predict the optimum design.

This paper proposes a novel approach to batch Bayesian optimisation using a multi-objective optimisation framework with exploitation and exploration forming two objectives. The key advantage of this approach is that it uses a suite of strategies to balance exploration and exploitation and thus can efficiently handle the optimisation of a variety of functions with small to large number of local extrema. Another advantage is that it automatically determines the batch size within a specified budget avoiding unnecessary function evaluations. Theoretical analysis shows that the regret not only reduces sub-linearly but also by an additional reduction factor determined by the batch size. We demonstrate the efficiency of our algorithm by optimising a variety of benchmark functions, performing hyperparameter tuning of support vector regression and classification, and finally heat treatment process of an Al-Sc alloy. Comparisons with recent baseline algorithms confirm the usefulness of our algorithm.

Experimental design is a process of obtaining a product with target property via experimentation. Bayesian optimization offers a sample-efficient tool for experimental design when experiments are expensive. Often, expert experimenters have 'hunches' about the behavior of the experimental system, offering potentials to further improve the efficiency. In this paper, we consider per-variable monotonic trend in the underlying property that results in a unimodal trend in those variables for a target value optimization. For example, sweetness of a candy is monotonic to the sugar content. However, to obtain a target sweetness, the utility of the sugar content becomes a unimodal function, which peaks at the value giving the target sweetness and falls off both ways. In this paper, we propose a novel method to solve such problems that achieves two main objectives: (a) the monotonicity information is used to the fullest extent possible, whilst ensuring that (b) the convergence guarantee remains intact. This is achieved by a two-stage Gaussian process modeling, where the first stage uses the monotonicity trend to model the underlying property, and the second stage uses 'virtual' samples, sampled from the first, to model the target value optimization function. The process is made theoretically consistent by adding appropriate adjustment factor in the posterior computation, necessitated because of using the 'virtual' samples. The proposed method is evaluated through both simulations and real world experimental design problems of (a) new short polymer fiber with the target length, and (b) designing of a new three dimensional porous scaffolding with a target porosity. In all scenarios our method demonstrates faster convergence than the basic Bayesian optimization approach not using such 'hunches'.

Privacy preservation is important. Prescriptive analytics is a method to extract corrective actions to avoid undesirable outcomes. We propose a privacy preserving prescriptive analytics algorithm to protect the data used during the construction of the prescriptive analytics algorithm. We use differential privacy mechanism to achieve strong privacy guarantee. Differential privacy mechanism requires computation of sensitivity: maximum change in the output between two training datasets, which is differed by only one instance. The main challenge we addressed is the computation of sensitivity of the prescription vector. In absence of any analytical form, we construct a nested global optimization problem to compute the sensitivity. We solve the optimization problem using constrained Bayesian optimization, as the nested structure makes the objective function expensive. We demonstrate our algorithm on two real world datasets and observe that the prescription vectors remains useful even after making them private.

Bayesian optimisation has become a powerful framework for global optimisation of black-box functions that are expensive to evaluate and possibly noisy. In addition to expensive evaluation of objective functions, many real-world optimisation problems deal with similarly expensive black-box constraints. However, there are few studies regarding the role of constraints in multi-objective Bayesian optimisation. In this paper, we extend the Expected Hypervolume Improvement by introducing expectation of constraints satisfaction and merging them into a new acquisition function called Expected Hypervolume Improvement with Constraints (EHVIC). We analyse the performance of our algorithm by estimating the feasible region dominated by Pareto front using 4 benchmark functions. The proposed method is also evaluated on a realworld problem of Alloy Design. We demonstrate that EHVIC is an effective algorithm that provides a promising performance by comparing to a well-known related method.

We introduce algorithmic assurance, the problem of testing whether machine learning algorithms are conforming to their intended design goal. We address this problem by proposing an efficient framework for algorithmic testing. To provide assurance, we need to efficiently discover scenarios where an algorithm decision deviates maximally from its intended gold standard. We mathematically formulate this task as an optimisation problem of an expensive, black-box function. We use an active learning approach based on Bayesian optimisation to solve this optimisation problem. We extend this framework to algorithms with vector-valued outputs by making appropriate modification in Bayesian optimisation via the EXP3 algorithm. We theoretically analyse our methods for convergence. Using two real-world applications, we demonstrate the efficiency of our methods. The significance of our problem formulation and initial solutions is that it will serve as the foundation in assuring humans about machines making complex decisions.

Bayesian optimization (BO) is a sample-efficient method for global optimization of expensive, noisy, black-box functions using probabilistic methods. The performance of a BO method depends on its selection strategy through the acquisition function. Expected improvement (EI) is one of the most widely used acquisition functions for BO that finds the expectation of the improvement function over the incumbent. The incumbent is usually selected as the best-observed value so far, termed as $y^\max$ (for the maximizing problem). Recent work has studied the convergence rate for EI under some mild assumptions or zero noise of observations. Especially, the work of Wang and de Freitas (2014) has derived the sublinear regret for EI under a stochastic noise. However, due to the difficulty in stochastic noise setting and to make the convergent proof feasible, they use an alternative choice for the incumbent as the maximum of the Gaussian process predictive mean, $μ^\max$. This modification makes the algorithm computationally inefficient because it requires an additional global optimization step to estimate $μ^\max$ that is costly and may be inaccurate. To address this issue, we derive a sublinear convergence rate for EI using the commonly used $y^\max$. Moreover, our analysis is the first to study a stopping criteria for EI to prevent unnecessary evaluations. Our analysis complements the results of Wang and de Freitas (2014) to theoretically cover two incumbent settings for EI. Finally, we demonstrate empirically that EI using $y^\max$ is both more computationally efficiency and more accurate than EI using $μ^\max$.

This paper studies the regret bound of two transfer learning algorithms in Bayesian optimisation. The first algorithm models any difference between the source and target functions as a noise process. The second algorithm proposes a new way to model the difference between the source and target as a Gaussian process which is then used to adapt the source data. We show that in both cases the regret bounds are tighter than in the no transfer case. We also experimentally compare the performance of these algorithms relative to no transfer learning and demonstrate benefits of transfer learning.

Bayesian optimization (BO) has recently emerged as a powerful and flexible tool for hyper-parameter tuning and more generally for the efficient global optimization of expensive black-box functions. Systems implementing BO has successfully solved difficult problems in automatic design choices and machine learning hyper-parameters tunings. Many recent advances in the methodologies and theories underlying Bayesian optimization have extended the framework to new applications and provided greater insights into the behavior of these algorithms. Still, these established techniques always require a user-defined space to perform optimization. This pre-defined space specifies the ranges of hyper-parameter values. In many situations, however, it can be difficult to prescribe such spaces, as a prior knowledge is often unavailable. Setting these regions arbitrarily can lead to inefficient optimization - if a space is too large, we can miss the optimum with a limited budget, on the other hand, if a space is too small, it may not contain the optimum point that we want to get. The unknown search space problem is intractable to solve in practice. Therefore, in this paper, we narrow down to consider specifically the setting of "weakly specified" search space for Bayesian optimization. By weakly specified space, we mean that the pre-defined space is placed at a sufficiently good region so that the optimization can expand and reach to the optimum. However, this pre-defined space need not include the global optimum. We tackle this problem by proposing the filtering expansion strategy for Bayesian optimization. Our approach starts from the initial region and gradually expands the search space. Wedevelop an efficient algorithm for this strategy and derive its regret bound. These theoretical results are complemented by an extensive set of experiments on benchmark functions and tworeal-world applications which demonstrate the benefits of our proposed approach.

Bayesian optimization is an efficient way to optimize expensive black-box functions such as designing a new product with highest quality or hyperparameter tuning of a machine learning algorithm. However, it has a serious limitation when the parameter space is high-dimensional as Bayesian optimization crucially depends on solving a global optimization of a surrogate utility function in the same sized dimensions. The surrogate utility function, known commonly as acquisition function is a continuous function but can be extremely sharp at high dimension - having only a few peaks marooned in a large terrain of almost flat surface. Global optimization algorithms such as DIRECT are infeasible at higher dimensions and gradient-dependent methods cannot move if initialized in the flat terrain. We propose an algorithm that enables local gradient-dependent algorithms to move through the flat terrain by using a sequence of gross-to-finer Gaussian process priors on the objective function as we leverage two underlying facts - a) there exists a large enough length-scales for which the acquisition function can be made to have a significant gradient at any location in the parameter space, and b) the extrema of the consecutive acquisition functions are close although they are different only due to a small difference in the length-scales. Theoretical guarantees are provided and experiments clearly demonstrate the utility of the proposed method at high dimension using both benchmark test functions and real-world case studies.

Scaling Bayesian optimization to high dimensions is challenging task as the global optimization of high-dimensional acquisition function can be expensive and often infeasible. Existing methods depend either on limited “active” variables or the additive form of the objective function. We propose a new method for high-dimensional Bayesian optimization, that uses a drop-out strategy to optimize only a subset of variables at each iteration. We derive theoretical bounds for the regret and show how it can inform the derivation of our algorithm. We demonstrate the efficacy of our algorithms for optimization on two benchmark functions and two real-world applications - training cascade classifiers and optimizing alloy composition.

Abstract Prevailing batch Bayesian optimisation methods allow all control variables to be freely altered at each iteration. Real-world experiments, however, often have physical limitations making it time-consuming to alter all settings for each recommendation in a batch. This gives rise to a unique problem in BO: in a recommended batch, a set of variables that are expensive to experimentally change need to be fixed, while the remaining control variables can be varied. We formulate this as a process-constrained batch Bayesian optimisation problem. We propose two algorithms, pc-BO(basic) and pc-BO(nested). pc-BO(basic) is simpler but lacks convergence guarantee. In contrast pc-BO(nested) is slightly more complex, but admits convergence analysis. We show that the regret of pc-BO(nested) is sublinear. We demonstrate the performance of both pc-BO(basic) and pc-BO(nested) by optimising benchmark test functions, tuning hyper-parameters of the SVM classifier, optimising the heat-treatment process for an Al-Sc alloy to achieve target hardness, and optimising the short polymer fibre production process.

Many real-world applications require multi-label classification where multiple target labels are assigned to each instance. In multi-label classification, there exist the intrinsic correlations between the labels and features. These correlations are beneficial for multi-label classification task since they reflect the coexistence of the input and output spaces that can be exploited for prediction. Traditional classification methods have attempted to reveal these correlations in different ways. However, existing methods demand expensive computation complexity for finding such correlation structures. Furthermore, these approaches can not identify the suitable number of label-feature correlation patterns. In this paper, we propose a Bayesian nonparametric (BNP) framework for multi-label classification that can automatically learn and exploit the unknown number of multi-label correlation. We utilize the recent techniques in stochastic inference to derive the cheap (but efficient) posterior inference algorithm for the model. In addition, our model can naturally exploit the useful information from missing label samples. Furthermore, we extend the model to update parameters in an online fashion that highlights the flexibility of our model against the existing approaches. We compare our method with the state-of-the-art multi-label classification algorithms on real-world datasets using both complete and missing label settings. Our model achieves better classification accuracy while our running time is consistently much faster than the baselines in an order of magnitude.

Parameter settings profoundly impact the performance of machine learning algorithms and laboratory experiments. The classical trial-error methods are exponentially expensive in large parameter spaces, and Bayesian optimization (BO) offers an elegant alternative for global optimization of black box functions. In situations where the functions can be evaluated at multiple points simultaneously, batch Bayesian optimization is used. Current batch BO approaches are restrictive in fixing the number of evaluations per batch, and this can be wasteful when the number of specified evaluations is larger than the number of real maxima in the underlying acquisition function. We present the budgeted batch Bayesian optimization (B3O) for hyper-parameter tuning and experimental design - we identify the appropriate batch size for each iteration in an elegant way. In particular, we use the infinite Gaussian mixture model (IGMM) for automatically identifying the number of peaks in the underlying acquisition functions. We solve the intractability of estimating the IGMM directly from the acquisition function by formulating the batch generalized slice sampling to efficiently draw samples from the acquisition function. We perform extensive experiments for benchmark functions and two real world applications - machine learning hyper-parameter tuning and experimental design for alloy hardening. We show empirically that the proposed B3O outperforms the existing fixed batch BO approaches in finding the optimum whilst requiring a fewer number of evaluations, thus saving cost and time.

Hyperparameters play a crucial role in the model selection of machine learning algorithms. Tuning these hyperparameters can be exhaustive when the data is large. Bayesian optimisation has emerged as an efficient tool for hyperparameter tuning of machine learning algorithms. In this paper, we propose a novel framework for tuning the hyperparameters for big data using Bayesian optimisation. We divide the big data into chunks and generate hyperparameter configurations for the chunks using the standard Bayesian optimisation. We utilise this information from the chunks for hyperparameter tuning on big data using a transfer learning setting. We evaluate the performance of the proposed method on the task of tuning hyperparameters of two machine learning algorithms. We show that our method achieves the best available hyperparameter configuration within less computational time compared to the state-of-art hyperparameter tuning methods.

Adverse effects, such as voice change and fatigue, are prevalent in cancer treatment duration. These adverse effects have been significant burden for patients physically and emotionally. Predicting multiple adverse effects becomes important for patients and oncologists. In this paper, we formulate the prediction of multiple adverse effects in cancer treatment as a longitudinal multiple-output regression problem. The correlated multiple outputs are first decoupled to uncorrelated ones in a new output space. We then propose a comprehensive framework to capture the empirical loss between the predicted value and the ground truth in the transformed space and the temporal smoothness at neighboring prediction points. Experiments were performed on one synthetic data and two real-world datasets including radiotherapy and chemotherapy treatments. Results in terms of root mean square errors (RMSE) and R-value show that our proposed approach is promising for the longitudinal multiple-output regression problem.

Multiple Instance Regression jointly models a set of instances and its corresponding real-valued output. We present a novel multiple instance regression model that infers a subset of instances in each bag that best describes the bag label and uses them to learn a predictive model in a unified framework. We assume that instances in each bag are drawn from a mixture distribution and thus naturally form groups, and instances from one of this group explain the bag label. The largest cluster is assumed to be correlated with the label. We evaluate this model on the crop yield prediction and aerosol depth prediction problems. The predictive accuracy of our model is better than the state of the art MIR methods.

Privacy-preserving data mining has become an active focus of the research community in the domains where data are sensitive and personal in nature. For example, highly sensitive digital repositories of medical or financial records offer enormous values for risk prediction and decision making. However, prediction models derived from such repositories should maintain strict privacy of individuals. We propose a novel random forest algorithm under the framework of differential privacy. Unlike previous works that strictly follow differential privacy and keep the complete data distribution approximately invariant to change in one data instance, we only keep the necessary statistics (e.g. variance of the estimate) invariant. This relaxation results in significantly higher utility. To realize our approach, we propose a novel differentially private decision tree induction algorithm and use them to create an ensemble of decision trees. We also propose feasible adversary models to infer about the attribute and class label of unknown data in presence of the knowledge of all other data. Under these adversary models, we derive bounds on the maximum number of trees that are allowed in the ensemble while maintaining privacy. We focus on binary classification problem and demonstrate our approach on four real-world datasets. Compared to the existing privacy preserving approaches we achieve significantly higher utility.

Multi-task learning is a learning paradigm that improves the performance of “related” tasks through their joint learning. To do this each task answers the question “Which other task should I share with”? This task relatedness can be complex - a task may be related to one set of tasks based on one subset of features and to other tasks based on other subsets. Existing multi-task learning methods do not explicitly model this reality, learning a single-faceted task relationship over all the features. This degrades performance by forcing a task to become similar to other tasks even on their unrelated features. Addressing this gap, we propose a novel multi-task learning model that learns multi-faceted task relationship, allowing tasks to collaborate differentially on different feature subsets. This is achieved by simultaneously learning a low dimensional subspace for task parameters and inducing task groups over each latent subspace basis using a novel combination of L1 and pairwise L∞ norms. Further, our model can induce grouping across both positively and negatively related tasks, which helps towards exploiting knowledge from all types of related tasks. We validate our model on two synthetic and five real datasets, and show significant performance improvements over several state-of-the-art multi-task learning techniques. Thus our model effectively answers for each task: What shall I share and with whom?

We propose a novel hierarchical Bayesian framework, word-distance-dependent Chinese restaurant franchise (wd-dCRF) for topic discovery from a document corpus regularized by side information in the form of word-to-word relations, with an application on Electronic Medical Records (EMRs). Typically, a EMRs dataset consists of several patients (documents) and each patient contains many diagnosis codes (words). We exploit the side information available in the form of a semantic tree structure among the diagnosis codes for semantically-coherent disease topic discovery. We introduce novel functions to compute word-to-word distances when side information is available in the form of tree structures. We derive an efficient inference method for the wddCRF using MCMC technique. We evaluate on a real world medical dataset consisting of about 1000 patients with PolyVascular disease. Compared with the popular topic analysis tool, hierarchical Dirichlet process (HDP), our model discovers topics which are superior in terms of both qualitative and quantitative measures.

Medical outcomes are inexorably linked to patient illness and clinical interventions. Interventions change the course of disease, crucially determining outcome. Traditional outcome prediction models build a single classifier by augmenting interventions with disease information. Interventions, however, differentially affect prognosis, thus a single prediction rule may not suffice to capture variations. Interventions also evolve over time as more advanced interventions replace older ones. To this end, we propose a Bayesian nonparametric, supervised framework that models a set of intervention groups through a mixture distribution building a separate prediction rule for each group, and allows the mixture distribution to change with time. This is achieved by using a hierarchical Dirichlet process mixture model over the interventions. The outcome is then modeled as conditional on both the latent grouping and the disease information through a Bayesian logistic regression. Experiments on synthetic and medical cohorts for 30-day readmission prediction demonstrate the superiority of the proposed model over clinical and data mining baselines.

BACKGROUND: Feature engineering is a time consuming component of predictive modeling. We propose a versatile platform to automatically extract features for risk prediction, based on a pre-defined and extensible entity schema. The extraction is independent of disease type or risk prediction task. We contrast auto-extracted features to baselines generated from the Elixhauser comorbidities. RESULTS: Hospital medical records was transformed to event sequences, to which filters were applied to extract feature sets capturing diversity in temporal scales and data types. The features were evaluated on a readmission prediction task, comparing with baseline feature sets generated from the Elixhauser comorbidities. The prediction model was through logistic regression with elastic net regularization. Predictions horizons of 1, 2, 3, 6, 12 months were considered for four diverse diseases: diabetes, COPD, mental disorders and pneumonia, with derivation and validation cohorts defined on non-overlapping data-collection periods. For unplanned readmissions, auto-extracted feature set using socio-demographic information and medical records, outperformed baselines derived from the socio-demographic information and Elixhauser comorbidities, over 20 settings (5 prediction horizons over 4 diseases). In particular over 30-day prediction, the AUCs are: COPD-baseline: 0.60 (95% CI: 0.57, 0.63), auto-extracted: 0.67 (0.64, 0.70); diabetes-baseline: 0.60 (0.58, 0.63), auto-extracted: 0.67 (0.64, 0.69); mental disorders-baseline: 0.57 (0.54, 0.60), auto-extracted: 0.69 (0.64,0.70); pneumonia-baseline: 0.61 (0.59, 0.63), auto-extracted: 0.70 (0.67, 0.72). CONCLUSIONS: The advantages of auto-extracted standard features from complex medical records, in a disease and task agnostic manner were demonstrated. Auto-extracted features have good predictive power over multiple time horizons. Such feature sets have potential to form the foundation of complex automated analytic tasks.

Multimedia contents often possess weakly annotated data such as tags, links and interactions. The weakly annotated data is called side information. It is the auxiliary information of data and provides hints for exploring the link structure of data. Most clustering algorithms utilize pure data for clustering. A model that combines pure data and side information, such as images and tags, documents and keywords, can perform better at understanding the underlying structure of data. We demonstrate how to incorporate different types of side information into a recently proposed Bayesian nonparametric model, the distance dependent Chinese restaurant process (DD-CRP). Our algorithm embeds the affinity of this information into the decay function of the DD-CRP when side information is in the form of subsets of discrete labels. It is flexible to measure distance based on arbitrary side information instead of only the spatial layout or time stamp of observations. At the same time, for noisy and incomplete side information, we set the decay function so that the DD-CRP reduces to the traditional Chinese restaurant process, thus not inducing side effects of noisy and incomplete side information. Experimental evaluations on two real-world datasets NUS WIDE and 20 Newsgroups show exploiting side information in DD-CRP significantly improves the clustering performance.

This paper addresses the limitation of current multilinear techniques (multilinear PCA, multilinear ICA) when applied to face recognition for handling faces in unseen illumination and viewpoints. We propose a new recognition method, exploiting the interaction of all the subspaces resulting from multilinear decomposition (for both multilinear PCA and ICA), to produce a new basis called multilinear-eigenmodes. This basis offers the flexibility to handle face images at unseen illumination or viewpoints. Experiments on benchmarked datasets yield superior performance in terms of both accuracy and computational cost.

This paper addresses the limitation of current multilinear PCA based techniques, in terms of prohibitive computational cost of testing and poor generalisation in some scenarios, when applied to large training databases. We define person-specific eigenmodes to obtain a set of projection bases, wherein a particular basis captures variation across lightings and viewpoints for a particular person. A new recognition approach is developed utilizing these bases. The proposed approach performs on a par with the existing multilinear approaches, whilst significantly reducing the complexity order of the testing algorithm.