Awesome Papers: 2016-09-1

Distributed Bayesian Learning with Stochastic Natural-gradient Expectation Propagation and the Posterior Server

Leonard Hasenclever, Stefan Webb, Thibaut Lienart, Yee Whye Teh, Sebastian Vollmer, Balaji Lakshminarayanan, Charles Blundell $\sum\limits_{\alpha }{\left[ \frac{\partial \theta }{\Xi \eta } \right]}$ (Submitted on 31 Dec 2015 (v1), last revised 1 Sep 2016 (this version, v2)) This paper makes two contributions to Bayesian machine learning algorithms. Firstly, we propose stochastic natural gradient expectation propagation (SNEP), a novel alternative to expectation propagation (EP), a popular variational inference algorithm. SNEP is a black box variational algorithm, in that it does not require any simplifying assumptions on the distribution of interest, beyond the existence of some Monte Carlo sampler for estimating the moments of the EP tilted distributions. Further, as opposed to EP which has no guarantee of convergence, SNEP can be shown to be convergent, even when using Monte Carlo moment estimates. Secondly, we propose a novel architecture for distributed Bayesian learning which we call the posterior server. The posterior server allows scalable and robust Bayesian learning in cases where a dataset is stored in a distributed manner across a cluster, with each compute node containing a disjoint subset of data. An independent Monte Carlo sampler is run on each compute node, with direct access only to the local data subset, but which targets an approximation to the global posterior distribution given all data across the whole cluster. This is achieved by using a distributed asynchronous implementation of SNEP to pass messages across the cluster. We demonstrate SNEP and the posterior server on distributed Bayesian learning of logistic regression and neural networks.

A novel progressive learning technique for multi-class classification

Authors: Rajasekar Venkatesan, Meng Joo Er (Submitted on 1 Sep 2016) In this paper, a progressive learning technique for multi-class classification is proposed. This newly developed learning technique is independent of the number of class constraints and it can learn new classes while still retaining the knowledge of previous classes. Whenever a new class (non-native to the knowledge learnt thus far) is encountered, the neural network structure gets remodeled automatically by facilitating new neurons and interconnections, and the parameters are calculated in such a way that it retains the knowledge learnt thus far. This technique is suitable for real-world applications where the number of classes is often unknown and online learning from real-time data is required. The consistency and the complexity of the progressive learning technique are analyzed. Several standard datasets are used to evaluate the performance of the developed technique. A comparative study shows that the developed technique is superior.

Employing traditional machine learning algorithms for big data streams analysis: the case of object trajectory prediction

Angelos Valsamis, Konstantinos Tserpes, Dimitrios Zissis, Dimosthenis Anagnostopoulos, Theodora Varvarigou (Submitted on 1 Sep 2016) In this paper, we model the trajectory of sea vessels and provide a service that predicts in near-real time the position of any given vessel in 4’, 10’, 20’ and 40’ time intervals. We explore the necessary tradeoffs between accuracy, performance and resource utilization are explored given the large volume and update rates of input data. We start with building models based on well-established machine learning algorithms using static datasets and multi-scan training approaches and identify the best candidate to be used in implementing a single-pass predictive approach, under real-time constraints. The results are measured in terms of accuracy and performance and are compared against the baseline kinematic equations. Results show that it is possible to efficiently model the trajectory of multiple vessels using a single model, which is trained and evaluated using an adequately large, static dataset, thus achieving a significant gain in terms of resource usage while not compromising accuracy.

A Unified View of Multi-Label Performance Measures

Xi-Zhu Wu, Zhi-Hua Zhou (Submitted on 1 Sep 2016) Multi-label classification deals with the problem where each instance is associated with multiple class labels. Because evaluation in multi-label classification is more complicated than single-label setting, a number of performance measures have been proposed. It is noticed that an algorithm usually performs differently on different measures. Therefore, it is important to understand which algorithms perform well on which measure(s) and why. In this paper, we propose a unified margin view to revisit eleven performance measures in multi-label classification. In particular, we define label-wise margin and instance-wise margin, and prove that through maximizing these margins, different corresponding performance measures will be optimized. Based on the defined margins, a max-margin approach called LIMO is designed and empirical results verify our theoretical findings.

Importance weighting without importance weights: An efficient algorithm for combinatorial semi-bandits

Gergely Neu, Gábor Bartók We propose a sample-efficient alternative for importance weighting for situations where one only has sample access to the probability distribution that generates the observations. Our new method, called Geometric Resampling (GR), is described and analyzed in the context of online combinatorial optimization under semi-bandit feedback, where a learner sequentially selects its actions from a combinatorial decision set so as to minimize its cumulative loss. In particular, we show that the well-known Follow-the-Perturbed-Leader (FPL) prediction method coupled with Geometric Resampling yields the first computationally efficient reduction from offline to online optimization in this setting. We provide a thorough theoretical analysis for the resulting algorithm, showing that its performance is on par with previous, inefficient solutions. Our main contribution is showing that, despite the relatively large variance induced by the GR procedure, our performance guarantees hold with high probability rather than only in expectation. As a side result, we also improve the best known regret bounds for FPL in online combinatorial optimization with full feedback, closing the perceived performance gap between FPL and exponential weights in this sett。

Business Process Deviance Mining: Review and Evaluation

Hoang Nguyen, Marlon Dumas, Marcello La Rosa, Fabrizio Maria Maggi, Suriadi Suriadi (Submitted on 29 Aug 2016) Business process deviance refers to the phenomenon whereby a subset of the executions of a business process deviate, in a negative or positive way, with respect to its expected or desirable outcomes. Deviant executions of a business process include those that violate compliance rules, or executions that undershoot or exceed performance targets. Deviance mining is concerned with uncovering the reasons for deviant executions by analyzing business process event logs. This article provides a systematic review and comparative evaluation of deviance mining approaches based on a family of data mining techniques known as sequence classification. Using real-life logs from multiple domains, we evaluate a range of feature types and classification methods in terms of their ability to accurately discriminate between normal and deviant executions of a process. We also analyze the interestingness of the rule sets extracted using different methods. We observe that feature sets extracted using pattern mining techniques only slightly outperform simpler feature sets based on counts of individual activity occurrences in a trace.

Modelling Cyber-Security Experts’ Decision Making Processes using Aggregation Operators

Simon Miller, Christian Wagner, Uwe Aickelin, Jonathan M. Garibaldi (Submitted on 30 Aug 2016) An important role carried out by cyber-security experts is the assessment of proposed computer systems, during their design stage. This task is fraught with difficulties and uncertainty, making the knowledge provided by human experts essential for successful assessment. Today, the increasing number of progressively complex systems has led to an urgent need to produce tools that support the expert-led process of system-security assessment. In this research, we use weighted averages (WAs) and ordered weighted averages (OWAs) with evolutionary algorithms (EAs) to create aggregation operators that model parts of the assessment process. We show how individual overall ratings for security components can be produced from ratings of their characteristics, and how these individual overall ratings can be aggregated to produce overall rankings of potential attacks on a system. As well as the identification of salient attacks and weak points in a prospective system, the proposed method also highlights which factors and security components contribute most to a component’s difficulty and attack ranking respectively. A real world scenario is used in which experts were asked to rank a set of technical attacks, and to answer a series of questions about the security components that are the subject of the attacks. The work shows how finding good aggregation operators, and identifying important components and factors of a cyber-security problem can be automated. The resulting operators have the potential for use as decision aids for systems designers and cyber-security experts, increasing the amount of assessment that can be achieved with the limited resources available.

Multi-Label Classification Method Based on Extreme Learning Machines

Rajasekar Venkatesan, Meng Joo Er (Submitted on 30 Aug 2016) In this paper, an Extreme Learning Machine (ELM) based technique for Multi-label classification problems is proposed and discussed. In multi-label classification, each of the input data samples belongs to one or more than one class labels. The traditional binary and multi-class classification problems are the subset of the multi-label problem with the number of labels corresponding to each sample limited to one. The proposed ELM based multi-label classification technique is evaluated with six different benchmark multi-label datasets from different domains such as multimedia, text and biology. A detailed comparison of the results is made by comparing the proposed method with the results from nine state of the arts techniques for five different evaluation metrics. The nine methods are chosen from different categories of multi-label methods. The comparative results shows that the proposed Extreme Learning Machine based multi-label classification technique is a better alternative than the existing state of the art methods for multi-label problems.

工具

深入深度学习四步走(相关资源介绍)

《4 Steps for Learning Deep Learning - A handy list of resources to help you become a deep learning expert》by Vivek Kumar

基于OCR任务的机器学习简单实践

《The Simple + Practical Path to Machine Learning Capability: A Common Benchmark Task》by Dan Kuster

Email

fengyanghe@gmail.com

07318457661