Research papers and code for "Zhong Liu":
MicroRNAs (miRNAs) are small RNA molecules composed of 19-22 nt, which play important regulatory roles in post-transcriptional gene regulation by inhibiting the translation of the mRNA into proteins or otherwise cleaving the target mRNA. Inferring miRNA targets provides useful information for understanding the roles of miRNA in biological processes that are potentially involved in complex diseases. Statistical methodologies for point estimation, such as the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm, have been proposed to identify the interactions of miRNA and mRNA based on sequence and expression data. In this paper, we propose using the Bayesian LASSO (BLASSO) and the non-negative Bayesian LASSO (nBLASSO) to analyse the interactions between miRNA and mRNA using expression data. The proposed Bayesian methods explore the posterior distributions for those parameters required to model the miRNA-mRNA interactions. These approaches can be used to observe the inferred effects of the miRNAs on the targets by plotting the posterior distributions of those parameters. For comparison purposes, the Least Squares Regression (LSR), Ridge Regression (RR), LASSO, non-negative LASSO (nLASSO), and the proposed Bayesian approaches were applied to four public datasets. We concluded that nLASSO and nBLASSO perform best in terms of sensitivity and specificity. Compared to the point estimate algorithms, which only provide single estimates for those parameters, the Bayesian methods are more meaningful and provide credible intervals, which take into account the uncertainty of the inferred interactions of the miRNA and mRNA. Furthermore, Bayesian methods naturally provide statistical significance to select convincing inferred interactions, while point estimate algorithms require a manually chosen threshold, which is less meaningful, to choose the possible interactions.

* 21 pages, 11 figures, 8 tables
Click to Read Paper and Get Code
Recently, there have been great interests in Monte Carlo Tree Search (MCTS) in AI research. Although the sequential version of MCTS has been studied widely, its parallel counterpart still lacks systematic study. This leads us to the following questions: \emph{how to design efficient parallel MCTS (or more general cases) algorithms with rigorous theoretical guarantee? Is it possible to achieve linear speedup?} In this paper, we consider the search problem on a more general acyclic one-root graph (namely, Monte Carlo Graph Search (MCGS)), which generalizes MCTS. We develop a parallel algorithm (P-MCGS) to assign multiple workers to investigate appropriate leaf nodes simultaneously. Our analysis shows that P-MCGS algorithm achieves linear speedup and that the sample complexity is comparable to its sequential counterpart.

Click to Read Paper and Get Code
We propose a collection of three shift-based primitives for building efficient compact CNN-based networks. These three primitives (channel shift, address shift, shortcut shift) can reduce the inference time on GPU while maintains the prediction accuracy. These shift-based primitives only moves the pointer but avoids memory copy, thus very fast. For example, the channel shift operation is 12.7x faster compared to channel shuffle in ShuffleNet but achieves the same accuracy. The address shift and channel shift can be merged into the point-wise group convolution and invokes only a single kernel call, taking little time to perform spatial convolution and channel shift. Shortcut shift requires no time to realize residual connection through allocating space in advance. We blend these shift-based primitives with point-wise group convolution and built two inference-efficient CNN architectures named AddressNet and Enhanced AddressNet. Experiments on CIFAR100 and ImageNet datasets show that our models are faster and achieve comparable or better accuracy.

Click to Read Paper and Get Code
The remarkable success of machine learning, especially deep learning, has produced a variety of cloud-based services for mobile users. Such services require an end user to send data to the service provider, which presents a serious challenge to end-user privacy. To address this concern, prior works either add noise to the data or send features extracted from the raw data. They struggle to balance between the utility and privacy because added noise reduces utility and raw data can be reconstructed from extracted features. This work represents a methodical departure from prior works: we balance between a measure of privacy and another of utility by leveraging adversarial learning to find a sweeter tradeoff. We design an encoder that optimizes against the reconstruction error (a measure of privacy), adversarially by a Decoder, and the inference accuracy (a measure of utility) by a Classifier. The result is RAN, a novel deep model with a new training algorithm that automatically extracts features for classification that are both private and useful. It turns out that adversarially forcing the extracted features to only conveys the intended information required by classification leads to an implicit regularization leading to better classification accuracy than the original model which completely ignores privacy. Thus, we achieve better privacy with better utility, a surprising possibility in machine learning! We conducted extensive experiments on five popular datasets over four training schemes, and demonstrate the superiority of RAN compared with existing alternatives.

Click to Read Paper and Get Code
Convolutional Neural Network (CNN) has gained state-of-the-art results in many pattern recognition and computer vision tasks. However, most of the CNN structures are manually designed by experienced researchers. Therefore, auto- matically building high performance networks becomes an important problem. In this paper, we introduce the idea of using DNA computing algorithm to automatically learn high-performance architectures. In DNA computing algorithm, we use short DNA strands to represent layers and long DNA strands to represent overall networks. We found that most of the learned models perform similarly, and only those performing worse during the first runs of training will perform worse finally than others. The indicates that: 1) Using DNA computing algorithm to learn deep architectures is feasible; 2) Local minima should not be a problem of deep networks; 3) We can use early stop to kill the models with the bad performance just after several runs of training. In our experiments, an accuracy 99.73% was obtained on the MNIST data set and an accuracy 95.10% was obtained on the CIFAR-10 data set.

Click to Read Paper and Get Code
Probabilistic matrix factorization (PMF) plays a crucial role in recommendation systems. It requires a large amount of user data (such as user shopping records and movie ratings) to predict personal preferences, and thereby provides users high-quality recommendation services, which expose the risk of leakage of user privacy. Differential privacy, as a provable privacy protection framework, has been applied widely to recommendation systems. It is common that different individuals have different levels of privacy requirements on items. However, traditional differential privacy can only provide a uniform level of privacy protection for all users. In this paper, we mainly propose a probabilistic matrix factorization recommendation scheme with personalized differential privacy (PDP-PMF). It aims to meet users' privacy requirements specified at the item-level instead of giving the same level of privacy guarantees for all. We then develop a modified sampling mechanism (with bounded differential privacy) for achieving PDP. We also perform a theoretical analysis of the PDP-PMF scheme and demonstrate the privacy of the PDP-PMF scheme. In addition, we implement the probabilistic matrix factorization schemes both with traditional and with personalized differential privacy (DP-PMF, PDP-PMF) and compare them through a series of experiments. The results show that the PDP-PMF scheme performs well on protecting the privacy of each user and its recommendation quality is much better than the DP-PMF scheme.

* 24 pages, 12 figures, 4 tables
Click to Read Paper and Get Code
In recent years, research on image generation methods has been developing fast. The auto-encoding variational Bayes method (VAEs) was proposed in 2013, which uses variational inference to learn a latent space from the image database and then generates images using the decoder. The generative adversarial networks (GANs) came out as a promising framework, which uses adversarial training to improve the generative ability of the generator. However, the generated pictures by GANs are generally blurry. The deep convolutional generative adversarial networks (DCGANs) were then proposed to leverage the quality of generated images. Since the input noise vectors are randomly sampled from a Gaussian distribution, the generator has to map from a whole normal distribution to the images. This makes DCGANs unable to reflect the inherent structure of the training data. In this paper, we propose a novel deep model, called generative adversarial networks with decoder-encoder output noise (DE-GANs), which takes advantage of both the adversarial training and the variational Bayesain inference to improve the performance of image generation. DE-GANs use a pre-trained decoder-encoder architecture to map the random Gaussian noise vectors to informative ones and pass them to the generator of the adversarial networks. Since the decoder-encoder architecture is trained by the same images as the generators, the output vectors could carry the intrinsic distribution information of the original images. Moreover, the loss function of DE-GANs is different from GANs and DCGANs. A hidden-space loss function is added to the adversarial loss function to enhance the robustness of the model. Extensive empirical results show that DE-GANs can accelerate the convergence of the adversarial training process and improve the quality of the generated images.

Click to Read Paper and Get Code
Evolutionary many-objective optimization has been gaining increasing attention from the evolutionary computation research community. Much effort has been devoted to addressing this issue by improving the scalability of multiobjective evolutionary algorithms, such as Pareto-based, decomposition-based, and indicator-based approaches. Different from current work, we propose a novel algorithm in this paper called AnD, which consists of an angle-based selection strategy and a shift-based density estimation strategy. These two strategies are employed in the environmental selection to delete the poor individuals one by one. Specifically, the former is devised to find a pair of individuals with the minimum vector angle, which means that these two individuals share the most similar search direction. The latter, which takes both the diversity and convergence into account, is adopted to compare these two individuals and to delete the worse one. AnD has a simple structure, few parameters, and no complicated operators. The performance of AnD is compared with that of seven state-of-the-art many-objective evolutionary algorithms on a variety of benchmark test problems with up to 15 objectives. The experimental results suggest that AnD can achieve highly competitive performance. In addition, we also verify that AnD can be readily extended to solve constrained many-objective optimization problems.

Click to Read Paper and Get Code
In this paper, we propose a novel unsupervised deep learning model, called PCA-based Convolutional Network (PCN). The architecture of PCN is composed of several feature extraction stages and a nonlinear output stage. Particularly, each feature extraction stage includes two layers: a convolutional layer and a feature pooling layer. In the convolutional layer, the filter banks are simply learned by PCA. In the nonlinear output stage, binary hashing is applied. For the higher convolutional layers, the filter banks are learned from the feature maps that were obtained in the previous stage. To test PCN, we conducted extensive experiments on some challenging tasks, including handwritten digits recognition, face recognition and texture classification. The results show that PCN performs competitive with or even better than state-of-the-art deep learning models. More importantly, since there is no back propagation for supervised finetuning, PCN is much more efficient than existing deep networks.

* 8 pages,5 figures
Click to Read Paper and Get Code
In the evolutionary computation research community, the performance of most evolutionary algorithms (EAs) depends strongly on their implemented coordinate system. However, the commonly used coordinate system is fixed and not well suited for different function landscapes, EAs thus might not search efficiently. To overcome this shortcoming, in this paper we propose a framework, named ACoS, to adaptively tune the coordinate systems in EAs. In ACoS, an Eigen coordinate system is established by making use of the cumulative population distribution information, which can be obtained based on a covariance matrix adaptation strategy and an additional archiving mechanism. Since the population distribution information can reflect the features of the function landscape to some extent, EAs in the Eigen coordinate system have the capability to identify the modality of the function landscape. In addition, the Eigen coordinate system is coupled with the original coordinate system, and they are selected according to a probability vector. The probability vector aims to determine the selection ratio of each coordinate system for each individual, and is adaptively updated based on the collected information from the offspring. ACoS has been applied to two of the most popular EA paradigms, i.e., particle swarm optimization (PSO) and differential evolution (DE), for solving 30 test functions with 30 and 50 dimensions at the 2014 IEEE Congress on Evolutionary Computation. The experimental studies demonstrate its effectiveness.

* This paper provides a new point of view toward how to describe an evolutionary operator in the original coordinate system, and also offers a convenient transformation from an evolutionary operator in the original coordinate system to the corresponding evolutionary operator in the Eigen coordinate system
Click to Read Paper and Get Code
The recent years have seen remarkable success in the use of deep neural networks on text summarization. However, there is no clear understanding of \textit{why} they perform so well, or \textit{how} they might be improved. In this paper, we seek to better understand how neural extractive summarization systems could benefit from different types of model architectures, transferable knowledge and learning schemas. Additionally, we find an effective way to improve current frameworks and achieve the state-of-the-art result on CNN/DailyMail by a large margin based on our observations and analyses. Hopefully, our work could provide more clues for future research on extractive summarization.

* Accepted by ACL 2019 (oral); Project homepage: http://pfliu.com/InterpretSum/
Click to Read Paper and Get Code
Recently, a number of learning-based optimization methods that combine data-driven architectures with the classical optimization algorithms have been proposed and explored, showing superior empirical performance in solving various ill-posed inverse problems, but there is still a scarcity of rigorous analysis about the convergence behaviors of learning-based optimization. In particular, most existing analyses are specific to unconstrained problems but cannot apply to the more general cases where some variables of interest are subject to certain constraints. In this paper, we propose Differentiable Linearized ADMM (D-LADMM) for solving the problems with linear constraints. Specifically, D-LADMM is a K-layer LADMM inspired deep neural network, which is obtained by firstly introducing some learnable weights in the classical Linearized ADMM algorithm and then generalizing the proximal operator to some learnable activation function. Notably, we rigorously prove that there exist a set of learnable parameters for D-LADMM to generate globally converged solutions, and we show that those desired parameters can be attained by training D-LADMM in a proper way. To the best of our knowledge, we are the first to provide the convergence analysis for the learning-based optimization method on constrained problems.

* Accepted by ICML2019
Click to Read Paper and Get Code
This work develops a novel end-to-end deep unsupervised learning method based on convolutional neural network (CNN) with pseudo-classes for remote sensing scene representation. First, we introduce center points as the centers of the pseudo classes and the training samples can be allocated with pseudo labels based on the center points. Therefore, the CNN model, which is used to extract features from the scenes, can be trained supervised with the pseudo labels. Moreover, a pseudo-center loss is developed to decrease the variance between the samples and the corresponding pseudo center point. The pseudo-center loss is important since it can update both the center points with the training samples and the CNN model with the center points in the training process simultaneously. Finally, joint learning of the pseudo-center loss and the pseudo softmax loss which is formulated with the samples and the pseudo labels is developed for unsupervised remote sensing scene representation to obtain discriminative representations from the scenes. Experiments are conducted over two commonly used remote sensing scene datasets to validate the effectiveness of the proposed method and the experimental results show the superiority of the proposed method when compared with other state-of-the-art methods.

* Submitted to IJCNN 2019
Click to Read Paper and Get Code
Recent breakthroughs in Go play and strategic games have witnessed the great potential of reinforcement learning in intelligently scheduling in uncertain environment, but some bottlenecks are also encountered when we generalize this paradigm to universal complex tasks. Among them, the low efficiency of data utilization in model-free reinforcement algorithms is of great concern. In contrast, the model-based reinforcement learning algorithms can reveal underlying dynamics in learning environments and seldom suffer the data utilization problem. To address the problem, a model-based reinforcement learning algorithm with attention mechanism embedded is proposed as an extension of World Models in this paper. We learn the environment model through Mixture Density Network Recurrent Network(MDN-RNN) for agents to interact, with combinations of variational auto-encoder(VAE) and attention incorporated in state value estimates during the process of learning policy. In this way, agent can learn optimal policies through less interactions with actual environment, and final experiments demonstrate the effectiveness of our model in control problem.

Click to Read Paper and Get Code
AutoML has become a popular service that is provided by most leading cloud service providers today. In this paper, we focus on the AutoML problem from the \emph{service provider's perspective}, motivated by the following practical consideration: When an AutoML service needs to serve {\em multiple users} with {\em multiple devices} at the same time, how can we allocate these devices to users in an efficient way? We focus on GP-EI, one of the most popular algorithms for automatic model selection and hyperparameter tuning, used by systems such as Google Vizer. The technical contribution of this paper is the first multi-device, multi-tenant algorithm for GP-EI that is aware of \emph{multiple} computation devices and multiple users sharing the same set of computation devices. Theoretically, given $N$ users and $M$ devices, we obtain a regret bound of $O((\text{\bf {MIU}}(T,K) + M)\frac{N^2}{M})$, where $\text{\bf {MIU}}(T,K)$ refers to the maximal incremental uncertainty up to time $T$ for the covariance matrix $K$. Empirically, we evaluate our algorithm on two applications of automatic model selection, and show that our algorithm significantly outperforms the strategy of serving users independently. Moreover, when multiple computation devices are available, we achieve near-linear speedup when the number of users is much larger than the number of devices.

Click to Read Paper and Get Code
User profiling means exploiting the technology of machine learning to predict attributes of users, such as demographic attributes, hobby attributes, preference attributes, etc. It's a powerful data support of precision marketing. Existing methods mainly study network behavior, personal preferences, post texts to build user profile. Through our data analysis of micro-blog, we find that females show more positive and have richer emotions than males in online social platform. This difference is very conducive to the distinction between genders. Therefore, we argue that sentiment context is important as well for user profiling.This paper focuses on exploiting microblog user posts to predict one of the demographic labels: gender. We propose a Sentiment Representation Learning based Multi-Layer Perceptron(SRL-MLP) model to classify gender. First we build a sentiment polarity classifier in advance by training Long Short-Term Memory(LSTM) model on e-commerce review corpus. Next we transfer sentiment representation to a basic MLP network. Last we conduct experiments on gender classification by sentiment representation. Experimental results show that our approach can improve gender classification accuracy by 5.53\%, from 84.20\% to 89.73\%.

Click to Read Paper and Get Code
Stochastic gradient methods are dominant in nonconvex optimization especially for deep models but have low asymptotical convergence due to the fixed smoothness. To address this problem, we propose a simple yet effective method for improving stochastic gradient methods named predictive local smoothness (PLS). First, we create a convergence condition to build a learning rate which varies adaptively with local smoothness. Second, the local smoothness can be predicted by the latest gradients. Third, we use the adaptive learning rate to update the stochastic gradients for exploring linear convergence rates. By applying the PLS method, we implement new variants of three popular algorithms: PLS-stochastic gradient descent (PLS-SGD), PLS-accelerated SGD (PLS-AccSGD), and PLS-AMSGrad. Moreover, we provide much simpler proofs to ensure their linear convergence. Empirical results show that the variants have better performance gains than the popular algorithms, such as, faster convergence and alleviating explosion and vanish of gradients.

* 14 pages, 7 figures
Click to Read Paper and Get Code
We present ease.ml, a declarative machine learning service platform we built to support more than ten research groups outside the computer science departments at ETH Zurich for their machine learning needs. With ease.ml, a user defines the high-level schema of a machine learning application and submits the task via a Web interface. The system automatically deals with the rest, such as model selection and data movement. In this paper, we describe the ease.ml architecture and focus on a novel technical problem introduced by ease.ml regarding resource allocation. We ask, as a "service provider" that manages a shared cluster of machines among all our users running machine learning workloads, what is the resource allocation strategy that maximizes the global satisfaction of all our users? Resource allocation is a critical yet subtle issue in this multi-tenant scenario, as we have to balance between efficiency and fairness. We first formalize the problem that we call multi-tenant model selection, aiming for minimizing the total regret of all users running automatic model selection tasks. We then develop a novel algorithm that combines multi-armed bandits with Bayesian optimization and prove a regret bound under the multi-tenant setting. Finally, we report our evaluation of ease.ml on synthetic data and on one service we are providing to our users, namely, image classification with deep neural networks. Our experimental evaluation results show that our proposed solution can be up to 9.8x faster in achieving the same global quality for all users as the two popular heuristics used by our users before ease.ml.

Click to Read Paper and Get Code
In many applications, ideas that are described by a set of words often flow between different groups. To facilitate users in analyzing the flow, we present a method to model the flow behaviors that aims at identifying the lead-lag relationships between word clusters of different user groups. In particular, an improved Bayesian conditional cointegration based on dynamic time warping is employed to learn links between words in different groups. A tensor-based technique is developed to cluster these linked words into different clusters (ideas) and track the flow of ideas. The main feature of the tensor representation is that we introduce two additional dimensions to represent both time and lead-lag relationships. Experiments on both synthetic and real datasets show that our method is more effective than methods based on traditional clustering techniques and achieves better accuracy. A case study was conducted to demonstrate the usefulness of our method in helping users understand the flow of ideas between different user groups on social media

* 8 pages, AAAI 2016
Click to Read Paper and Get Code
Facial MicroExpressions (MEs) are spontaneous, involuntary facial movements when a person experiences an emotion but deliberately or unconsciously attempts to conceal his or her genuine emotions. Recently, ME recognition has attracted increasing attention due to its potential applications such as clinical diagnosis, business negotiation, interrogations and security. However, it is expensive to build large scale ME datasets, mainly due to the difficulty of naturally inducing spontaneous MEs. This limits the application of deep learning techniques which require lots of training data. In this paper, we propose a simple, efficient yet robust descriptor called Extended Local Binary Patterns on Three Orthogonal Planes (ELBPTOP) for ME recognition. ELBPTOP consists of three complementary binary descriptors: LBPTOP and two novel ones Radial Difference LBPTOP (RDLBPTOP) and Angular Difference LBPTOP (ADLBPTOP), which explore the local second order information along radial and angular directions contained in ME video sequences. ELBPTOP is a novel ME descriptor inspired by the unique and subtle facial movements. It is computationally efficient and only marginally increases the cost of computing LBPTOP, yet is extremely effective for ME recognition. In addition, by firstly introducing Whitened Principal Component Analysis (WPCA) to ME recognition, we can further obtain more compact and discriminative feature representations, and achieve significantly computational savings. Extensive experimental evaluation on three popular spontaneous ME datasets SMIC, CASMEII and SAMM show that our proposed ELBPTOP approach significantly outperforms previous state of the art on all three evaluated datasets. Our proposed ELBPTOP achieves 73.94% on CASMEII, which is 6.6% higher than state of the art on this dataset. More impressively, ELBPTOP increases recognition accuracy from 44.7% to 63.44% on the SAMM dataset.

Click to Read Paper and Get Code