New Computational Techniques for a Faster Variation of BM3D Image Denoising

BM3D has been considered the standard for comparison in the image denoising literature for the last decade. Though it has been shown to be surpassed numerous times by alternative algorithms in terms of PSNR, the margins are very thin, and denoising is approaching a limiting point. The reason for the continued use of BM3D within the literature is due to its off-the-shelf ease-of-use in any application, which alternative improved denoising algorithms sometimes fail to match. This article proposes a new variation of BM3D, which maintains its ease of use but is notably faster. This development brings us closer to real-time ease-of-use application of new state-of-the-art image reconstruction algorithms such as plug-and-play priors. We refer to our variation of BM3D as G-BM3D. In terms of image quality, our algorithm attains very similar denoising performance to the original algorithm. Though our algorithm is written completely in MATLAB software, it is already between 5-20 times faster than the original algorithm, and the modifications to the algorithm are such that it is expected to be significantly faster when ported to CUDA language and with more powerful GPUs. The improved processing time is achieved by two main components. The first component is a new computational strategy that achieves faster block matching, and the second is a new global approach to the 3D wavelet filtering step that allows for significantly improved processing times on GPUs. The fast block matching strategy could also be applied to any of the vast number of nonlocal self-similarity (NSS) denoisers to improve processing times.

Video Frame Interpolation via Structure-Motion based Iterative Fusion

Video Frame Interpolation synthesizes non-existent images between adjacent frames, with the aim of providing a smooth and consistent visual experience. Two approaches for solving this challenging task are optical flow based and kernel-based methods. In existing works, optical flow based methods can provide accurate point-to-point motion description, however, they lack constraints on object structure. On the contrary, kernel-based methods focus on structural alignment, which relies on semantic and apparent features, but tends to blur results. Based on these observations, we propose a structure-motion based iterative fusion method. The framework is an end-to-end learnable structure with two stages. First, interpolated frames are synthesized by structure-based and motion-based learning branches respectively, then, an iterative refinement module is established via spatial and temporal feature integration. Inspired by the observation that audiences have different visual preferences on foreground and background objects, we for the first time propose to use saliency masks in the evaluation processes of the task of video frame interpolation. Experimental results on three typical benchmarks show that the proposed method achieves superior performance on all evaluation metrics over the state-of-the-art methods, even when our models are trained with only one-tenth of the data other methods use.

Learning Fast and Slow: PROPEDEUTICA for Real-time Malware Detection

In this paper, we introduce and evaluate PROPEDEUTICA, a novel methodology and framework for efficient and effective real-time malware detection, leveraging the best of conventional machine learning (ML) and deep learning (DL) algorithms. In PROPEDEUTICA, all software processes in the system start execution subjected to a conventional ML detector for fast classification. If a piece of software receives a borderline classification, it is subjected to further analysis via more performance expensive and more accurate DL methods, via our newly proposed DL algorithm DEEPMALWARE. Further, we introduce delays to the execution of software subjected to deep learning analysis as a way to "buy time" for DL analysis and to rate-limit the impact of possible malware in the system. We evaluated PROPEDEUTICA with a set of 9,115 malware samples and 877 commonly used benign software samples from various categories for the Windows OS. Our results show that the false positive rate for conventional ML methods can reach 20%, and for modern DL methods it is usually below 6%. However, the classification time for DL can be 100X longer than conventional ML methods. PROPEDEUTICA improved the detection F1-score from 77.54% (conventional ML method) to 90.25%, and reduced the detection time by 54.86%. Further, the percentage of software subjected to DL analysis was approximately 40% on average. Further, the application of delays in software subjected to ML reduced the detection time by approximately 10%. Finally, we found and discussed a discrepancy between the detection accuracy offline (analysis after all traces are collected) and on-the-fly (analysis in tandem with trace collection). Our insights show that conventional ML and modern DL-based malware detectors in isolation cannot meet the needs of efficient and effective malware detection: high accuracy, low false positive rate, and short classification time.

Early Performance Prediction using Interpretable Patterns in Programming Process Data

Instructors have limited time and resources to help struggling students, and these resources should be directed to the students who most need them. To address this, researchers have constructed models that can predict students' final course performance early in a semester. However, many predictive models are limited to static and generic student features (e.g. demographics, GPA), rather than computing-specific evidence that assesses a student's progress in class. Many programming environments now capture complete time-stamped records of students' actions during programming. In this work, we leverage this rich, fine-grained log data to build a model to predict student course outcomes. From the log data, we extract patterns of behaviors that are predictive of students' success using an approach called differential sequence mining. We evaluate our approach on a dataset from 106 students in a block-based, introductory programming course. The patterns extracted from our approach can predict final programming performance with 79% accuracy using only the first programming assignment, outperforming two baseline methods. In addition, we show that the patterns are interpretable and correspond to concrete, effective -- and ineffective -- novice programming behaviors. We also discuss these patterns and their implications for classroom instruction.

Reinforced Attention for Few-Shot Learning and Beyond

Few-shot learning aims to correctly recognize query samples from unseen classes given a limited number of support samples, often by relying on global embeddings of images. In this paper, we propose to equip the backbone network with an attention agent, which is trained by reinforcement learning. The policy gradient algorithm is employed to train the agent towards adaptively localizing the representative regions on feature maps over time. We further design a reward function based on the prediction of the held-out data, thus helping the attention mechanism to generalize better across the unseen classes. The extensive experiments show, with the help of the reinforced attention, that our embedding network has the capability to progressively generate a more discriminative representation in few-shot learning. Moreover, experiments on the task of image classification also show the effectiveness of the proposed design.

Transient Chaos in BERT

Language is an outcome of our complex and dynamic human-interactions and the technique of natural language processing (NLP) is hence built on human linguistic activities. Bidirectional Encoder Representations from Transformers (BERT) has recently gained its popularity by establishing the state-of-the-art scores in several NLP benchmarks. A Lite BERT (ALBERT) is literally characterized as a lightweight version of BERT, in which the number of BERT parameters is reduced by repeatedly applying the same neural network called Transformer's encoder layer. By pre-training the parameters with a massive amount of natural language data, ALBERT can convert input sentences into versatile high-dimensional vectors potentially capable of solving multiple NLP tasks. In that sense, ALBERT can be regarded as a well-designed high-dimensional dynamical system whose operator is the Transformer's encoder, and essential structures of human language are thus expected to be encapsulated in its dynamics. In this study, we investigated the embedded properties of ALBERT to reveal how NLP tasks are effectively solved by exploiting its dynamics. We thereby aimed to explore the nature of human language from the dynamical expressions of the NLP model. Our short-term analysis clarified that the pre-trained model stably yields trajectories with higher dimensionality, which would enhance the expressive capacity required for NLP tasks. Also, our long-term analysis revealed that ALBERT intrinsically shows transient chaos, a typical nonlinear phenomenon showing chaotic dynamics only in its transient, and the pre-trained ALBERT model tends to produce the chaotic trajectory for a significantly longer time period compared to a randomly-initialized one. Our results imply that local chaoticity would contribute to improving NLP performance, uncovering a novel aspect in the role of chaotic dynamics in human language behaviors.

Pattern Localization in Time Series through Signal-To-Model Alignment in Latent Space

In this paper, we study the problem of locating a predefined sequence of patterns in a time series. In particular, the studied scenario assumes a theoretical model is available that contains the expected locations of the patterns. This problem is found in several contexts, and it is commonly solved by first synthesizing a time series from the model, and then aligning it to the true time series through dynamic time warping. We propose a technique that increases the similarity of both time series before aligning them, by mapping them into a latent correlation space. The mapping is learned from the data through a machine-learning setup. Experiments on data from non-destructive testing demonstrate that the proposed approach shows significant improvements over the state of the art.

Discussion of Ensemble Learning under the Era of Deep Learning

Due to the dominant position of deep learning (mostly deep neural networks) in various artificial intelligence applications, recently, ensemble learning based on deep neural networks (ensemble deep learning) has shown significant performances in improving the generalization of learning system. However, since modern deep neural networks usually have millions to billions of parameters, the time and space overheads for training multiple base deep learners and testing with the ensemble deep learner are far greater than that of traditional ensemble learning. Though several algorithms of fast ensemble deep learning have been proposed to promote the deployment of ensemble deep learning in some applications, further advances still need to be made for many applications in specific fields, where the developing time and computing resources are usually restricted or the data to be processed is of large dimensionality. An urgent problem needs to be solved is how to take the significant advantages of ensemble deep learning while reduce the required time and space overheads so that many more applications in specific fields can benefit from it. For the alleviation of this problem, it is necessary to know about how ensemble learning has developed under the era of deep learning. Thus, in this article, we present discussion focusing on data analyses of published works, the methodology and unattainability of traditional ensemble learning, and recent developments of ensemble deep learning. We hope this article will be helpful to realize the technical challenges faced by future developments of ensemble learning under the era of deep learning.

Adaptable Hamiltonian neural networks

The rapid growth of research in exploiting machine learning to predict chaotic systems has revived a recent interest in Hamiltonian Neural Networks (HNNs) with physical constraints defined by the Hamilton's equations of motion, which represent a major class of physics-enhanced neural networks. We introduce a class of HNNs capable of adaptable prediction of nonlinear physical systems: by training the neural network based on time series from a small number of bifurcation-parameter values of the target Hamiltonian system, the HNN can predict the dynamical states at other parameter values, where the network has not been exposed to any information about the system at these parameter values. The architecture of the HNN differs from the previous ones in that we incorporate an input parameter channel, rendering the HNN parameter--cognizant. We demonstrate, using paradigmatic Hamiltonian systems, that training the HNN using time series from as few as four parameter values bestows the neural machine with the ability to predict the state of the target system in an entire parameter interval. Utilizing the ensemble maximum Lyapunov exponent and the alignment index as indicators, we show that our parameter-cognizant HNN can successfully predict the route of transition to chaos. Physics-enhanced machine learning is a forefront area of research, and our adaptable HNNs provide an approach to understanding machine learning with broad applications.

Unsupervised Instance Selection with Low-Label, Supervised Learning for Outlier Detection

The laborious process of labeling data often bottlenecks projects that aim to leverage the power of supervised machine learning. Active Learning (AL) has been established as a technique to ameliorate this condition through an iterative framework that queries a human annotator for labels of instances with the most uncertain class assignment. Via this mechanism, AL produces a binary classifier trained on less labeled data but with little, if any, loss in predictive performance. Despite its advantages, AL can have difficulty with class-imbalanced datasets and results in an inefficient labeling process. To address these drawbacks, we investigate our unsupervised instance selection (UNISEL) technique followed by a Random Forest (RF) classifier on 10 outlier detection datasets under low-label conditions. These results are compared to AL performed on the same datasets. Further, we investigate the combination of UNISEL and AL. Results indicate that UNISEL followed by an RF performs comparably to AL with an RF and that the combination of UNISEL and AL demonstrates superior performance. The practical implications of these findings in terms of time savings and generalizability afforded by UNISEL are discussed.