Circadian rhythms are oscillations of behavior, physiology and metabolism in many organisms. Recent advancements in omics technology make it possible for genome-wide profiling of circadian rhythms. Here, we conducted a comprehensive analysis of seven existing algorithms commonly used for circadian rhythm detection. Using gold-standard circadian and non-circadian genes, we systematically evaluated the accuracy and reproducibility of the algorithms on empirical datasets generated from various omics platforms under different experimental designs. We also carried out extensive simulation studies to test each algorithm's robustness to key variables, including sampling patterns, replicates, waveforms, signal-to-noise ratios, uneven samplings and missing values. Furthermore, we examined the distributions of the nominal $P$-values under the null and raised issues with multiple testing corrections using traditional approaches. With our assessment, we provide method selection guidelines for circadian rhythm detection, which are applicable to different types of high-throughput omics data.

This work is to improve deep learning training performances via compression-based denoising algorithms. The deep learning models we focus on are Deep Boltzmann Machines (DBM) including Restricted Boltzmann Machines (RBM), Bernoulli DBM, Gaussian Bernoulli RBM, and Gaussian Bernoulli DBM. In theory, DBMs are universal appropximators, however, there is a gap between DBMs' promised theoretical properties and their practical performances. In this paper, we propose a post-hoc signal processing technique, i.e., denoise, to trained DBM. The proposed technique denoises DBMs, and thus the resulting models, denoising DBMs, remain more information about the training data. The proposed denoising DBM has its roots in compression-based denoising techniques, and is unlike other denoising techniques widely used in deep learning areas.

Click-through rate (CTR) prediction is a critical task in online advertising and marketing. For this problem, existing approaches, with shallow or deep architectures, have three major drawbacks. First, they typically lack persuasive rationales to explain the outcomes of the models. Unexplainable predictions and recommendations may be difficult to validate and thus unreliable and untrustworthy. In many applications, inappropriate suggestions may even bring severe consequences. Second, existing approaches have poor efficiency in analyzing high-order feature interactions. Third, the polysemy of feature interactions in different semantic subspaces is largely ignored. In this paper, we propose InterHAt that employs a Transformer with multi-head self-attention for feature learning. On top of that, hierarchical attention layers are utilized for predicting CTR while simultaneously providing interpretable insights of the prediction results. InterHAt captures high-order feature interactions by an efficient attentional aggregation strategy with low computational complexity. Extensive experiments on four public real datasets and one synthetic dataset demonstrate the effectiveness and efficiency of InterHAt.

We explore the connections between rate distortion/lossy source coding and deep learning models, the Restricted Boltzmann Machines (RBMs) and Deep Belief Networks (DBNs). We show that rate distortion is a function of the RBM log partition function and that RBM/DBN can be used to learn the rate distortion approaching posterior as in the Blahut-Arimoto algorithm. We propose an algorithm for lossy compressing of binary sources. The algorithm consists of two stages, a training stage that learns the posterior with training data of the same class as the source, and a compression/reproduction stage that is comprised of a lossless compression and a lossless reproduction. Theoretical results show that the proposed algorithm achieves the optimum rate distortion function for stationary ergodic sources asymptotically. Numerical experiments show that the proposed algorithm outperforms the reported best results.

We answer two questions in this work: what Deep Boltzmann Machines (DBMs) can do for compression and vise versa. We show that (1) DBMs can be applied to learn the rate distortion approaching posterior as in the Blahut-Arimoto (BA) algorithm, and to construct a lossy source compression scheme based on the Deep AutoEncoder; (2) compression can improve DBMs’ training performances via compression-based denoising algorithms. The implementation of the BA algorithm in the form of DBMs is the foundation of the two applications.

Motivated by a recent work of learning rate distortion approaching posterior via Restricted Boltzmann Machines, we generalize the result to Deep Belief Networks and propose a deep learning based lossy compression for stationary ergodic sources. The compression algorithm consists of two stages, a training stage, which is to learn the posterior with the training data of the same class as the source, and a compression/reproduction stage, which consists of a lossless compression and a lossless reproduction. The theoretical result shows that our algorithm asymptotically achieves the optimum rate-distortion function for stationary ergodic sources, and the experimental results outperform the reported best results.