检索结果-内蒙古大学图书馆

TOPOGRAPH: AN EFFICIENT GRAPH-BASED FRAMEWORK FOR STRICTLY TOPOLOGY PRESERVING IMAGE SEGMENTATION

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lux, Laurin Berger, Alexander H. Weers, Alexander Stucki, Nico Rueckert, Daniel Bauer, Ulrich Paetzold, Johannes C. School of Computation Information and Technology Technical University of Munich Germany Department of Computing Imperial College London United Kingdom Munich Center for Machine Learning Germany Munich Data Science Institute Technical University of Munich Munich Germany

Topological correctness plays a critical role in many image segmentation tasks, yet most networks are trained using pixel-wise loss functions, such as Dice, neglecting topological accuracy. Existing topology-aware methods often lack robust topological guarantees, are limited to specific use cases, or impose high computational costs. In this work, we propose a novel, graph-based framework for topologically accurate image segmentation that is both computationally efficient and generally applicable. Our method constructs a component graph that fully encodes the topological information of both the prediction and ground truth, allowing us to efficiently identify topologically critical regions and aggregate a loss based on local neighborhood information. Furthermore, we introduce a strict topological metric capturing the homotopy equivalence between the union and intersection of prediction-label pairs. We formally prove the topological guarantees of our approach and empirically validate its effectiveness on binary and multi-class datasets. Our loss demonstrates state-of-the-art performance with up to fivefold faster loss computation compared to persistent homology methods. © 2024, CC BY.

关键词： Image segmentation

Tracking the risk of a deployed model and detecting harmful distribution shifts

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Podkopaev, Aleksandr Ramdas, Aaditya Department of Statistics & Data Science Carnegie Mellon University United States Machine Learning Department Carnegie Mellon University United States

When deployed in the real world, machine learning models inevitably encounter changes in the data distribution, and certain—but not all—distribution shifts could result in significant performance degradation. In practice, it may make sense to ignore benign shifts, under which the performance of a deployed model does not degrade substantially, making interventions by a human expert (or model retraining) unnecessary. While several works have developed tests for distribution shifts, these typically either use non-sequential methods, or detect arbitrary shifts (benign or harmful), or both. We argue that a sensible method for firing off a warning has to both (a) detect harmful shifts while ignoring benign ones, and (b) allow continuous monitoring of model performance without increasing the false alarm rate. In this work, we design simple sequential tools for testing if the difference between source (training) and target (test) distributions leads to a significant increase in a risk function of interest, like accuracy or calibration. Recent advances in constructing time-uniform confidence sequences allow efficient aggregation of statistical evidence accumulated during the tracking process. The designed framework is applicable in settings where (some) true labels are revealed after the prediction is performed, or when batches of labels become available in a delayed fashion. We demonstrate the efficacy of the proposed framework through an extensive empirical study on a collection of simulated and real datasets. Copyright © 2021, The Authors. All rights reserved.

关键词： machine learning

A Comprehensive Review of Supervised learning Algorithms in Healthcare Applications

学校读者我要写书评

暂无评论

A Comprehensive Review of Supervised Learning Algorithms in ...

Artificial Intelligence in Education and Industry 4.0 (IDICAIEI), DMIHER International Conference on

作者： Aditya Barhate Abhay Tale Nayan Jikar Prateek Verma Praveen Kumar Prajyot Yesankar Department of Artificial Intelligence and Machine Learning Faculty of Engineering and Technology Datta Meghe Institute of Higher Education and Research Wardha Maharashtra India Department of Artificial Intelligence and Data Science Faculty of Engineering and Technology Datta Meghe Institute of Higher Education and Research Wardha Maharashtra India

ISBN: (数字)9798331528713

ISBN: (纸本)9798331528720

Supervised learning has revolutionized the concept of personalization in treatment with the development of Precision Medicine. This review aims to provide a systematic analysis of the utilization of supervised learning in healthcare and its functions in diagnostics and prognoses, recommendations for treatment, and patient surveillance. Classification algorithms such as Logistic Regression, Support Vector machines, and Decision Trees. Regression techniques such as Linear Regression, Polynomial Regression, and Neural Networks. Ensemble methods such as Bagging, Boosting, and Stacking. Deep learning techniques such as Convolutional Neural Networks, Recurrent Neural Networks, and Long Short-Term Memory are discussed for their efficiency and applicability in various healthcare applications. There are strengths highlighted by the review including enhanced diagnostic capabilities and patient-targeted therapy alongside weaknesses such as quality of data, interpretability as well as compatibility with existing healthcare systems. This review gives a more generalized approach to how supervised learning algorithms help to enhance healthcare results and identifies the likely topic that requires more research in the future.

关键词： Industries Deep learning Ethics Systematics Reviews Precision medicine Supervised learning Medical services Classification algorithms Tuning

Application of machine learning Algorithms for Analyzing Sentiments Using Twitter dataset

学校读者我要写书评

暂无评论

Application of Machine Learning Algorithms for Analyzing Sen...

Sustainable Computing and Smart Systems (ICSCSS), International Conference on

作者： Bhairavi Rewatkar Aditya Barhate Prateek Verma Department of Artificial Intelligence and Data Science Faculty of Engineering and Technology Datta Meghe Institute of Higher Education and Research Wardha Maharashtra India Department of Artificial Intelligence and Machine Learning Faculty of Engineering and Technology Datta Meghe Institute of Higher Education and Research Wardha Maharashtra India

ISBN: (数字)9798350379990

ISBN: (纸本)9798350391558

Sentiment analysis is a Natural Language Processing (NLP) approach used to determine whether a piece of text is negative, positive, or neutral based on the emotions it expresses. Despite advancements, several challenges persist in this field, including handling sarcasm, context ambiguity, and the presence of noisy data. Recent techniques, such as deep learning models and transformer-based approaches, have improved accuracy but still face limitations in computational efficiency and real-time analysis. This article employs machine learning (ML) techniques to analyze text and determine the sentiments of reviews shared on Twitter, aiming to address these challenges. The proposed system extracts attributes from the text, such as the sentiment polarity (whether the sentiment is negative or positive) and the topic or theme of the discussion. A comparison with prior work on sentiment analysis in tweets is presented, along with a comprehensive study of feature extraction and representation methods. The study evaluates the performance of six classifiers: Naïve Bayes, Decision Tree, Logistic Regression, Support Vector machine (SVM), Gradient Boosting, and Random Forest, based on different feature extraction techniques. Using the Bag of Words technique, SVM achieved the highest accuracy of 94.58%, while using term frequency, Random Forest showed the highest accuracy of 94.99%. The developed algorithms for sentiment analysis in tweets are useful for analyzing social feedback, product surveys, and social media comments. Combining neural networks and deep learning improves the accuracy of sentiment analysis and provides opportunities to analyze various types of emotions, enabling proactive responses to public sentiment.

关键词： Support vector machines Deep learning Sentiment analysis Accuracy machine learning algorithms Social networking (online) Reviews

Modeling lens potentials with continuous neural fields in galaxy-scale strong lenses

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Biggio, Luca Vernardos, Georgios Galan, Aymeric Peel, Austin Data Analytics Lab Institute of Machine Learning Department of Computer Science ETHZ Switzerland Observatoire de Sauverny Versoix1290 Switzerland

Strong gravitational lensing is a unique observational tool for studying the dark and luminous mass distribution both within and between galaxies. Given the presence of substructures, current strong lensing observations demand more complex mass models than smooth analytical profiles, such as power-law ellipsoids. In this work, we introduce a continuous neural field to predict the lensing potential at any position throughout the image plane, allowing for a nearly model-independent description of the lensing mass. We apply our method on simulated Hubble Space Telescope imaging data containing different types of perturbations to a smooth mass distribution: a localized dark subhalo, a population of subhalos, and an external shear perturbation. Assuming knowledge of the source surface brightness, we use the continuous neural field to model either the perturbations alone or the full lensing potential. In both cases, the resulting model is able to fit the imaging data, and we are able to accurately recover the properties of both the smooth potential and of the perturbations. Unlike many other deep learning methods, ours explicitly retains lensing physics (i.e., the lens equation) and introduces high flexibility in the model only where required, namely, in the lens potential. Moreover, the neural network does not require pre-training on large sets of labelled data and predicts the potential from the single observed lensing image. Our model is implemented in the fully differentiable lens modeling code Herculens. © 2022, CC BY.

关键词： Gravitation

ECTSum: A New Benchmark dataset For Bullet Point Summarization of Long Earnings Call Transcripts

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Mukherjee, Rajdeep Bohra, Abhinav Banerjee, Akash Sharma, Soumya Hegde, Manjunath Shaikh, Afreen Shrivastava, Shivani Dasgupta, Koustuv Ganguly, Niloy Ghosh, Saptarshi Goyal, Pawan Department of Computer Science and Engineering IIT Kharagpur India Goldman Sachs Data Science and Machine Learning Group India Leibniz University of Hannover Germany

Despite tremendous progress in automatic summarization, state-of-the-art methods are predominantly trained to excel in summarizing short newswire articles, or documents with strong layout biases such as scientific articles or govt reports. Efficient techniques to summarize financial documents, including facts and figures, have largely been unexplored, majorly due to the unavailability of suitable datasets. Here, we present ECTSum, a new dataset with transcripts of earnings calls (ECTs), hosted by public companies, as documents, and short experts-written telegram-style bullet point summaries derived from corresponding Reuters articles. ECTs are long unstructured documents without any prescribed length limit or format. We benchmark ECTSum with state-of-the-art summarizers across various metrics evaluating the content quality and factual consistency of the generated summaries. Finally, we present a simple-yet-effective approach, ECT-BPS, to generate a set of bullet points that precisely capture the important facts discussed in the calls. Copyright © 2022, The Authors. All rights reserved.

关键词： Benchmarking

Instance-Dependent Noisy Label learning via Graphical Modelling

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Garg, Arpit Nguyen, Cuong Felix, Rafael Do, Thanh-Toan Carneiro, Gustavo Australian Institute for Machine Learning University of Adelaide Australia Department of Data Science and AI Faculty of Information Technology Monash University Australia

Noisy labels are unavoidable yet troublesome in the ecosystem of deep learning because models can easily overfit them. There are many types of label noise, such as symmetric, asymmetric and instance-dependent noise (IDN), with IDN being the only type that depends on image information. Such dependence on image information makes IDN a critical type of label noise to study, given that la-belling mistakes are caused in large part by insufficient or ambiguous information about the visual classes present in images. Aiming to provide an effective technique to address IDN, we present a new graphical modelling approach called InstanceGM, that combines discriminative and generative models. The main contributions of InstanceGM are: i) the use of the continuous Bernoulli distribution to train the generative model, offering significant training advantages, and ii) the exploration of a state-of-the-art noisy-label discriminative classifier to generate clean labels from instance-dependent noisy-label samples. InstanceGM is competitive with current noisy-label learning approaches, particularly in IDN benchmarks using synthetic and real-world datasets, where our method shows better accuracy than the competitors in most experiments. © 2022, CC BY.

关键词： Deep learning

Partial counterfactual identification and uplift modeling: theoretical results and real-world assessment

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Verhelst, Théo Mercier, Denis Shrestha, Jeevan Bontempi, Gianluca Machine Learning Group Department of Computer Science Université Libre de Bruxelles Brussels Belgium Data Science Team Orange Belgium Brussels Belgium

Counterfactuals are central in causal human reasoning and the scientific discovery process. The uplift, also called conditional average treatment effect, measures the causal effect of some action, or treatment, on the outcome of an individual. This paper discusses how it is possible to derive bounds on the probability of counterfactual statements based on uplift terms. First, we derive some original bounds on the probability of counterfactuals and we show that tightness of such bounds depends on the information of the feature set on the uplift term. Then, we propose a point estimator based on the assumption of conditional independence between the counterfactual outcomes. The quality of the bounds and the point estimators are assessed on synthetic data and a large real-world customer data set provided by a telecom company, showing significant improvement over the state of the art. © 2022, CC BY-NC-ND.

关键词：

Position: Bayesian deep learning is needed in the age of large-scale AI 24

学校读者我要写书评

暂无评论

Position: Bayesian deep learning is needed in the age of lar...

Proceedings of the 41st International Conference on machine learning

作者： Theodore Papamarkou Maria Skoularidou Konstantina Palla Laurence Aitchison Julyan Arbel David Dunson Maurizio Filippone Vincent Fortuin Philipp Hennig José Miguel Hernández-Lobato Aliaksandr Hubin Alexander Immer Theofanis Karaletsos Mohammad Emtiyaz Khan Agustinus Kristiadi Yingzhen Li Stephan Mandt Christopher Nemeth Michael A. Osborne Tim G. J. Rudner David Rügamer Yee Whye Teh Max Welling Andrew Gordon Wilson Ruqi Zhang Department of Mathematics The University of Manchester Manchester UK Eric and Wendy Schmidt Center Broad Institute of MIT and Harvard Cambridge Spotify London UK Computational Neuroscience Unit University of Bristol Bristol UK Centre Inria de l'Université Grenoble Alpes Grenoble France Department of Statistical Science Duke University Statistics Program KAUST Saudi Arabia Helmholtz AI Munich Germany and Department of Computer Science Technical University of Munich Munich Germany and Munich Center for Machine Learning Munich Germany Tübingen AI Center University of Tübingen Tübingen Germany Department of Engineering University of Cambridge Cambridge UK Department of Mathematics University of Oslo Oslo Norway and Bioinformatics and Applied Statistics Norwegian University of Life Sciences Ås Norway Department of Computer Science ETH Zurich Switzerland Chan Zuckerberg Initiative California Center for Advanced Intelligence Project RIKEN Tokyo Japan Vector Institute Toronto Canada Department of Computing Imperial College London London UK Department of Computer Science UC Irvine Irvine Department of Mathematics and Statistics Lancaster University Lancaster UK Department of Engineering Science University of Oxford Oxford UK Center for Data Science New York University New York Munich Center for Machine Learning Munich Germany and Department of Statistics LMU Munich Munich Germany DeepMind London UK and Department of Statistics University of Oxford Oxford UK Informatics Institute University of Amsterdam Amsterdam Netherlands Courant Institute of Mathematical Sciences and Center for Data Science Computer Science Department New York University New York Department of Computer Science Purdue University West Lafayette

In the current landscape of deep learning research, there is a predominant emphasis on achieving high predictive accuracy in supervised tasks involving large image and language datasets. However, a broader perspective reveals a multitude of overlooked metrics, tasks, and data types, such as uncertainty, active and continual learning, and scientific data, that demand attention. Bayesian deep learning (BDL) constitutes a promising avenue, offering advantages across these diverse settings. This paper posits that BDL can elevate the capabilities of deep learning. It revisits the strengths of BDL, acknowledges existing challenges, and highlights some exciting research avenues aimed at addressing these obstacles. Looking ahead, the discussion focuses on possible ways to combine large-scale foundation models with BDL to unlock their full potential.

关键词：