Slow emerging topic detection is a task between event detection, where we aggregate behaviors of different words on short period of time, and language evolution, where we monitor their long term evolution. In this work, we tackle the problem of early detection of slowly emerging new topics. To this end, we gather evidence of weak signals at the word level. We propose to monitor the behavior of words representation in an embedding space and use one of its geometrical properties to characterize the emergence of topics. As evaluation is typically hard for this kind of task, we present a framework for quantitative evaluation. We show positive results that outperform state-of-the-art methods on two public datasets of press and scientific articles.
相關內容
Markov chains with variable length are useful parsimonious stochastic models able to generate most stationary sequence of discrete symbols. The idea is to identify the suffixes of the past, called contexts, that are relevant to predict the future symbol. Sometimes a single state is a context, and looking at the past and finding this specific state makes the further past irrelevant. States with such property are called renewal states and they can be used to split the chain into independent and identically distributed blocks. In order to identify renewal states for chains with variable length, we propose the use of Intrinsic Bayes Factor to evaluate the hypothesis that some particular state is a renewal state. In this case, the difficulty lies in integrating the marginal posterior distribution for the random context trees for general prior distribution on the space of context trees, with Dirichlet prior for the transition probabilities, and Monte Carlo methods are applied. To show the strength of our method, we analyzed artificial datasets generated from different binary models models and one example coming from the field of Linguistics.
Distributed data processing frameworks (e.g., Hadoop, Spark, and Flink) are widely used to distribute data among computing nodes of a cloud. Recently, there have been increasing efforts aimed at evaluating the performance of distributed data processing frameworks hosted in private and public clouds. However, there is a paucity of research on evaluating the performance of these frameworks hosted in a hybrid cloud, which is an emerging cloud model that integrates private and public clouds to use the best of both worlds. Therefore, in this paper, we evaluate the performance of Hadoop, Spark, and Flink in a hybrid cloud in terms of execution time, resource utilization, horizontal scalability, vertical scalability, and cost. For this study, our hybrid cloud consists of OpenStack (private cloud) and MS Azure (public cloud). We use both batch and iterative workloads for the evaluation. Our results show that in a hybrid cloud (i) the execution time increases as more nodes are borrowed by the private cloud from the public cloud, (ii) Flink outperforms Spark, which in turn outperforms Hadoop in terms of execution time, (iii) Hadoop transfers the largest amount of data among the nodes during the workload execution while Spark transfers the least amount of data, (iv) all three frameworks horizontally scale better as compared to vertical scaling, and (v) Spark is found to be least expensive in terms of $ cost for data processing while Hadoop is found the most expensive.
State-of-the-art machine learning models are routinely trained on large-scale distributed clusters. Crucially, such systems can be compromised when some of the computing devices exhibit abnormal (Byzantine) behavior and return arbitrary results to the parameter server (PS). This behavior may be attributed to a plethora of reasons, including system failures and orchestrated attacks. Existing work suggests robust aggregation and/or computational redundancy to alleviate the effect of distorted gradients. However, most of these schemes are ineffective when an adversary knows the task assignment and can choose the attacked workers judiciously to induce maximal damage. Our proposed method Aspis assigns gradient computations to worker nodes using a subset-based assignment which allows for multiple consistency checks on the behavior of a worker node. Examination of the calculated gradients and post-processing (clique-finding in an appropriately constructed graph) by the central node allows for efficient detection and subsequent exclusion of adversaries from the training process. We prove the Byzantine resilience and detection guarantees of Aspis under weak and strong attacks and extensively evaluate the system on various large-scale training scenarios. The principal metric for our experiments is the test accuracy, for which we demonstrate a significant improvement of about 30% compared to many state-of-the-art approaches on the CIFAR-10 dataset. The corresponding reduction of the fraction of corrupted gradients ranges from 16% to 99%.
Inferring the scene illumination from a single image is an essential yet challenging task in computer vision and computer graphics. Existing works estimate lighting by regressing representative illumination parameters or generating illumination maps directly. However, these methods often suffer from poor accuracy and generalization. This paper presents Geometric Mover's Light (GMLight), a lighting estimation framework that employs a regression network and a generative projector for effective illumination estimation. We parameterize illumination scenes in terms of the geometric light distribution, light intensity, ambient term, and auxiliary depth, which can be estimated by a regression network. Inspired by the earth mover's distance, we design a novel geometric mover's loss to guide the accurate regression of light distribution parameters. With the estimated light parameters, the generative projector synthesizes panoramic illumination maps with realistic appearance and high-frequency details. Extensive experiments show that GMLight achieves accurate illumination estimation and superior fidelity in relighting for 3D object insertion. The codes are available at \href{//github.com/fnzhan/Illumination-Estimation}{//github.com/fnzhan/Illumination-Estimation}.
The dominating NLP paradigm of training a strong neural predictor to perform one task on a specific dataset has led to state-of-the-art performance in a variety of applications (eg. sentiment classification, span-prediction based question answering or machine translation). However, it builds upon the assumption that the data distribution is stationary, ie. that the data is sampled from a fixed distribution both at training and test time. This way of training is inconsistent with how we as humans are able to learn from and operate within a constantly changing stream of information. Moreover, it is ill-adapted to real-world use cases where the data distribution is expected to shift over the course of a model's lifetime. The first goal of this thesis is to characterize the different forms this shift can take in the context of natural language processing, and propose benchmarks and evaluation metrics to measure its effect on current deep learning architectures. We then proceed to take steps to mitigate the effect of distributional shift on NLP models. To this end, we develop methods based on parametric reformulations of the distributionally robust optimization framework. Empirically, we demonstrate that these approaches yield more robust models as demonstrated on a selection of realistic problems. In the third and final part of this thesis, we explore ways of efficiently adapting existing models to new domains or tasks. Our contribution to this topic takes inspiration from information geometry to derive a new gradient update rule which alleviate catastrophic forgetting issues during adaptation.
Evaluating the quality of learned representations without relying on a downstream task remains one of the challenges in representation learning. In this work, we present Geometric Component Analysis (GeomCA) algorithm that evaluates representation spaces based on their geometric and topological properties. GeomCA can be applied to representations of any dimension, independently of the model that generated them. We demonstrate its applicability by analyzing representations obtained from a variety of scenarios, such as contrastive learning models, generative models and supervised learning models.
Transformer architectures show significant promise for natural language processing. Given that a single pretrained model can be fine-tuned to perform well on many different tasks, these networks appear to extract generally useful linguistic features. A natural question is how such networks represent this information internally. This paper describes qualitative and quantitative investigations of one particularly effective model, BERT. At a high level, linguistic features seem to be represented in separate semantic and syntactic subspaces. We find evidence of a fine-grained geometric representation of word senses. We also present empirical descriptions of syntactic representations in both attention matrices and individual word embeddings, as well as a mathematical argument to explain the geometry of these representations.
Latest deep learning methods for object detection provide remarkable performance, but have limits when used in robotic applications. One of the most relevant issues is the long training time, which is due to the large size and imbalance of the associated training sets, characterized by few positive and a large number of negative examples (i.e. background). Proposed approaches are based on end-to-end learning by back-propagation [22] or kernel methods trained with Hard Negatives Mining on top of deep features [8]. These solutions are effective, but prohibitively slow for on-line applications. In this paper we propose a novel pipeline for object detection that overcomes this problem and provides comparable performance, with a 60x training speedup. Our pipeline combines (i) the Region Proposal Network and the deep feature extractor from [22] to efficiently select candidate RoIs and encode them into powerful representations, with (ii) the FALKON [23] algorithm, a novel kernel-based method that allows fast training on large scale problems (millions of points). We address the size and imbalance of training data by exploiting the stochastic subsampling intrinsic into the method and a novel, fast, bootstrapping approach. We assess the effectiveness of the approach on a standard Computer Vision dataset (PASCAL VOC 2007 [5]) and demonstrate its applicability to a real robotic scenario with the iCubWorld Transformations [18] dataset.
We introduce an algorithmic method for population anomaly detection based on gaussianization through an adversarial autoencoder. This method is applicable to detection of `soft' anomalies in arbitrarily distributed highly-dimensional data. A soft, or population, anomaly is characterized by a shift in the distribution of the data set, where certain elements appear with higher probability than anticipated. Such anomalies must be detected by considering a sufficiently large sample set rather than a single sample. Applications include, but not limited to, payment fraud trends, data exfiltration, disease clusters and epidemics, and social unrests. We evaluate the method on several domains and obtain both quantitative results and qualitative insights.
Neural word embeddings have been widely used in biomedical Natural Language Processing (NLP) applications since they provide vector representations of words that capture the semantic properties of words and the linguistic relationship between words. Many biomedical applications use different textual sources to train word embeddings and apply these word embeddings to downstream biomedical applications. However, there has been little work on comprehensively evaluating the word embeddings trained from these resources. In this study, we provide a comprehensive empirical evaluation of word embeddings trained from four different resources, namely clinical notes, biomedical publications, Wikepedia, and news. We perform the evaluation qualitatively and quantitatively. In qualitative evaluation, we manually inspect five most similar medical words to a given set of target medical words, and then analyze word embeddings through the visualization of those word embeddings. Quantitative evaluation falls into two categories: extrinsic and intrinsic evaluation. Based on the evaluation results, we can draw the following conclusions. First, EHR and PubMed can capture the semantics of medical terms better than GloVe and Google News and find more relevant similar medical terms. Second, the medical semantic similarity captured by the word embeddings trained on EHR and PubMed are closer to human experts' judgments, compared to these trained on GloVe and Google News. Third, there does not exist a consistent global ranking of word embedding quality for downstream biomedical NLP applications. However, adding word embeddings as extra features will improve results on most downstream tasks. Finally, word embeddings trained from a similar domain corpus do not necessarily have better performance than other word embeddings for any downstream biomedical tasks.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191