Search

Spike sorting – the process of separating spikes from different neurons – is often the first and most critical step in the neural data analysis pipeline. Spike-sorting techniques isolate a single neuron's activity from background electrical noise based on the shapes of the waveforms obtained from extracellular recordings. Despite several advancements in this area, an important remaining challenge in neuroscience is online spike sorting, which has the potential to significantly advance basic neuroscience research and the clinical setting by providing the means to produce real-time perturbations of neurons via closed-loop control. Current approaches to online spike sorting are not fully automated, are computationally expensive and are often outperformed by offline approaches. In this paper, we present a novel algorithm for fast and robust online classification of single neuron activity. This algorithm is based on a deep contractive autoencoder (CAE) architecture. CAEs are neural networks that can learn a latent state representation of their inputs. The main advantage of CAE-based approaches is that they are less sensitive to noise (i.e., small perturbations in their inputs). We therefore reasoned that they can form the basis for robust online spike sorting algorithms. Overall, our deep CAE-based online spike sorting algorithm achieves over 90% accuracy in sorting unseen spike waveforms, outperforming existing models and maintaining a performance close to the offline case. In the offline scenario, our method substantially outperforms the existing models, providing an average improvement of 40% in accuracy over different datasets. • A novel algorithm for spike sorting based on a Contractive Auto-encoder. • Produce representations of spike waveforms that are robust to additive noise. • Reliably classify spikes for small and large datasets. • Outperform SOTA approaches in various online and offline spike-sorting applications. [ABSTRACT FROM AUTHOR]

Web search engines play a significant role in answering users' information needs based on the huge amount of data available on the internet. Although evaluating the performance of these systems is very important for their improvement, there is no comprehensive, unbiased, low-cost, and reusable method for this purpose. Previous works used a small and limited set of queries for their evaluation process that restricts the assessment domain. Moreover, these methods mainly rely on human evaluators for manual assessment of search engines which makes the results of the evaluation subjective to the opinion of human evaluators and also prone to error. In addition, repeating the evaluation would be a problem, as it requires the same level of human effort as of the first evaluation. Another drawback of the existing evaluations is that they score a search result based on its position in the retrieved list of relevant pages. This implies that these methods are only evaluating the ranker component of a web search engine, leaving all other components unevaluated. In this research, we propose an automatic approach for web search engine evaluation that can run with a query set that is multiple times bigger than the query sets used in the manual evaluations. The automatic nature of our proposed method makes repetition of the evaluation to be low cost in terms of the required human effort. Moreover, we designed this approach to be component based, meaning that we have different evaluation tasks for assessing different components of web search engines. For each component, queries are designed differently and are meant to assess the functionality of that component only. Similarly, the way that the retrieved results are scored is different for each component. For example, for assessing the spell correction component, the input query would contain a typo, and in the results, only instances of that word with the correct form will be scored positively. Experimental results of applying thousands of queries on two Persian and two language-independent web search engines show that none of the selected search engines dominates the other three across all components; instead, each search engine has its own points of strength and weakness that are highlighted through this evaluation. [ABSTRACT FROM AUTHOR]

Advanced metering infrastructure has been widely recognized as a key enabling factor for delivering a range of benefits to the electricity industry as well as to energy consumers. The data is commonly used for billing and to educate consumers so they can make economically optimal decisions on choosing between electricity retail plans. More importantly, an in-depth analysis of this data can drive the development and implementation of new energy technologies and to enable evidence-based policy decisions to be made. In this paper, an Australian case study is presented on how residential battery installations lead to behavioral changes in the way energy is consumed. Furthermore, data-informed recommendations to both distribution network operators and policy-makers are also provided. The data comprises more than 5000 energy consumers with either distributed generation systems such as Photovoltaics (PV) and Energy Storage Systems (ESS), or without. The methodology focuses on the analysis of energy consumption of consumers with PV and ESS energy consumers and compares them against consumers without ESS. In addition, an economic analysis on the benefit of installing ESS is presented using payback period and internal rate of return. The main finding is that residential energy storage systems provide a range of benefits to the distribution network. It is recommended that Governments should continue supporting the rollout of such systems by subsidy programs and implementing necessary policy directives. [ABSTRACT FROM AUTHOR]

A fundamental component to managing a marketing campaign is identifying prospects and selection of leads. Current lead generation models focus on predicting the intention of a customer to purchase a product, however with financial products, particularly loans, this can be insufficient as there are many factors to consider, such as risk, utility, and financial maturity. Developing a marketing campaign for loan prospecting should consider not only customers who need a loan, but rather clients who need a loan and will also be approved. Otherwise, the marketing effort is deemed ineffective, if the lead cannot be converted into a sale. Although, a low response rate is expected for a marketing campaign for loans, we highlight a better approach in managing resources while maintaining a shortlist of high-quality leads. This manuscript introduces a bi-level approach to handle the complex nature of loan products. Two classifiers are built, one modelling loan intention and the other one modelling loan eligibility. We adopt convex combination to control weights of both problems, which in most cases results in an improved performance for identifying future successful loan applicants when compared to baseline models. Rank-based evaluation measures are also adopted to explore the performance of customer rankings. We find that soft classifiers, such as deep learning techniques, are ideal for ranking customers, achieving a superior performance when compared to other machine learning techniques. In addition, we conclude that an ideal cut-off for K customers is estimated to be between 20 to 25 customers, however our best model can maintain an Average Precision of greater than 0.85 when K approaches 50. • Lead generation for loans presents a challenging classification problem. • Existing methods for predicting leads have lacked consideration of loan approval. • Combining binary classifiers with varying objectives can improve model accuracy. • Rank-based measures are preferred for evaluating the quality of predicted leads. • Deep learning models are ideal for lead selection when ranking is considered. [ABSTRACT FROM AUTHOR]

Recently network embedding has gained increasing attention due to its advantages in facilitating network computation tasks such as link prediction, node classification and node clustering. The objective of network embedding is to represent network nodes in a low-dimensional vector space while retaining as much information as possible from the original network including structural, relational, and semantic information. However, asymmetric nature of directed networks poses many challenges as how to best preserve edge directions during the embedding process. Here, we propose a novel deep asymmetric attributed network embedding model based on the convolutional graph neural network, called AAGCN. The main idea is to maximally preserve the asymmetric proximity and asymmetric similarity of directed attributed networks. AAGCN introduces two neighbourhood feature aggregation schemes to separately aggregate the features of a node with the features of its in- and out- neighbours. Then, it learns two embedding vectors for each node, one source embedding vector and one target embedding vector. The final representations are the results of concatenating source and target embedding vectors. We test the performance of AAGCN on four real-world networks for network reconstruction, link prediction, node classification and visualization downstream tasks and investigate the impact of hyperparameters of the proposed method on the performance of the tasks. The experimental results show the superiority of AAGCN against state-of-the-art embedding methods. [ABSTRACT FROM AUTHOR]