44,685 Hits in 8.6 sec

Automatic differentiation in ML: Where we are and where we should be going [article]

Bart van Merriënboer, Olivier Breuleux, Arnaud Bergeron, Pascal Lamblin
2019 arXiv   pre-print
We review the current state of automatic differentiation (AD) for array programming in machine learning (ML), including the different approaches such as operator overloading (OO) and source transformation  ...  Unlike existing dataflow programming representations in ML frameworks, our IR naturally supports function calls, higher-order functions and recursion, making ML models easier to implement.  ...  Early discussions and brainstorming with Olexa Bilaniuk also helped determining the scope and direction of the project.  ... 
arXiv:1810.11530v2 fatcat:g2chgpagsvhn5daeka26diuwle

From Cleaning before ML to Cleaning for ML

Felix Neutatz, Binger Chen, Ziawasch Abedjan, Eugene Wu
2021 IEEE Data Engineering Bulletin  
Traditional data cleaning focuses on quality issues of a dataset in isolation of the application using the data-Cleaning Before ML-which can be inefficient and, counterintuitively, degrade the application  ...  While recent cleaning approaches take into account signals from the ML model, such as the model accuracy, they are still local to a specific model, and do not take into account the entire application's  ...  For instance, the user may specify that the total price in January should be 40 instead of 100, or all values in the output are too low.  ... 
dblp:journals/debu/NeutatzCA021 fatcat:thspsxnq4rdx3psrijelwbg6pq

ML for ML: Learning Cost Semantics by Experiment [chapter]

Ankush Das, Jan Hoffmann
2017 Lecture Notes in Computer Science  
The considered resources in the implementation are heap allocations and execution time.  ...  The derived cost semantics are combined with RAML, a state-of-the-art system for automatically deriving resource bounds for OCaml programs.  ...  S = P j=1 M i=1 T (i,j) − c∈C n (i,j) c T c 2 where T c are the unknowns that need to be learned.  ... 
doi:10.1007/978-3-662-54577-5_11 fatcat:ndiuvlejgfdzlbe26dqo5eakdi

Transferable Graph Optimizers for ML Compilers [article]

Yanqi Zhou, Sudip Roy, Amirali Abdolrashidi, Daniel Wong, Peter Ma, Qiumin Xu, Hanxiao Liu, Phitchaya Mangpo Phothilimthana, Shen Wang, Anna Goldie, Azalia Mirhoseini, James Laudon
2021 arXiv   pre-print
Existing learning based approaches in the literature are sample inefficient, tackle a single optimization problem, and do not generalize to unseen graphs making them infeasible to be deployed in practice  ...  On a diverse set of representative graphs consisting of up to 80,000 nodes, including Inception-v3, Transformer-XL, and WaveNet, GO achieves on average 21% improvement over human experts and 18% improvement  ...  In this paper, we propose an end-to-end deep RL method (GO) for ML compiler graph optimizations where the learned policy is generalizable to new graphs and transferable across multiple tasks.  ... 
arXiv:2010.12438v2 fatcat:ju26bxgmajbgfa4wtwvgrc6k2a

Non-Oriented MLS Gradient Fields

Jiazhou Chen, Gaël Guennebaud, Pascal Barla, Xavier Granier
2013 Computer graphics forum (Print)  
In particular, we show that our novel isotropic linear approximation outperforms its lower-order alternative: surface or image structures are much better preserved, and instabilities are significantly  ...  Thanks to its ease of implementation (on both CPU and GPU) and small performance overhead, we believe our approach will find a widespread use in graphics applications, as demonstrated by the variety of  ...  The models of figures 12 and 18 are courtesy of the Stanford Computer Graphics Laboratory, and the model of figure 14 is courtesy of the AIM@SHAPE Shape Repository.  ... 
doi:10.1111/cgf.12164 fatcat:a56z4bsg4re5lfedz3xz77huui

Kafka-ML: connecting the data stream with ML/AI frameworks [article]

Cristian Martín, Peter Langendoerfer, Pouya Soltani Zarrin, Manuel Díaz, Bartolomé Rubio
2020 arXiv   pre-print
In this paper, we proposed Kafka-ML, an open-source framework that enables the management of TensorFlow ML/AI pipelines through data streams (Apache Kafka).  ...  Kafka-ML provides an accessible and user-friendly Web User Interface where users can easily define ML models, to then train, evaluate and deploy them for inference.  ...  Moreover, high-availability, load-balancing and fault-tolerance may be required in ML/AI mission-critical applications and should be provided in a transparent way to users.  ... 
arXiv:2006.04105v2 fatcat:mbfukjzkvbfjjod47k4kykarbq

Programming by Examples: PL Meets ML [chapter]

Sumit Gulwani, Prateek Jain
2017 Lecture Notes in Computer Science  
There are three key components in a PBE system. (i) A search algorithm that can efficiently search for programs that are consistent with the examples provided by the user.  ...  PBE systems are already revolutionizing the application domain of data wrangling and are set to significantly impact several other domains including code refactoring.  ...  in this article related to using ML techniques for search and ranking.  ... 
doi:10.1007/978-3-319-71237-6_1 fatcat:nou2fnkpt5elfj3ohaunnfmy7y

Intriguing Properties of Adversarial ML Attacks in the Problem Space [article]

Fabio Pierazzi, Feargus Pendlebury, Jacopo Cortellazzi, Lorenzo Cavallaro
2020 arXiv   pre-print
Our results demonstrate that "adversarial-malware as a service" is a realistic threat, as we automatically generate thousands of realistic and inconspicuous adversarial applications at scale, where on  ...  Recent research efforts on adversarial ML have investigated problem-space attacks, focusing on the generation of real evasive objects in domains where, unlike images, there is no clear inverse mapping  ...  ACKNOWLEDGEMENTS We thank the anonymous reviewers and our shepherd, Nicolas Papernot, for their constructive feedback, as well as Battista Biggio, Konrad Rieck, and Erwin Quiring for feedback on early  ... 
arXiv:1911.02142v2 fatcat:fioc4k5eczf2toexvneuetxnhi


Claudio V. Russo, Dimitrios Vytiniotis
2009 Proceedings of the 2009 ACM SIGPLAN workshop on ML - ML '09  
Quantified types co-exist with ordinary ML schemes, which are in turn implicitly introduced and eliminated at let-bindings and use sites, respectively.  ...  This paper suggests a modest extension of ML with System F types: the heart of the idea is to extend the language of types with unary universal and existential quantifiers.  ...  Acknowledgements Thanks to the ML'09 anonymous reviewers for their valuable suggestions on related work, Philip Wadler for reminding us of O'Toole and Gifford's work (16) , and Simon Peyton Jones for  ... 
doi:10.1145/1596627.1596630 fatcat:ruxpndc3nvfetlkyspuw4wtxca

Ease.ML/Snoopy: Towards Automatic Feasibility Studies for ML via Quantitative Understanding of "Data Quality for ML" [article]

Cedric Renggli, Luka Rimanic, Luka Kolar, Wentao Wu, Ce Zhang
2022 arXiv   pre-print
In this paper, we present Snoopy, with the goal of supporting data scientists and machine learning engineers performing a systematic and theoretically founded feasibility study before building ML applications  ...  In our experience of working with domain experts who are using today's AutoML systems, a common problem we encountered is what we call "unrealistic expectations" -- when users are facing a very challenging  ...  However, in order to apply a BER estimator in a system for a feasibility study in real-world datasets, we need to be able to go beyond uniform noise.  ... 
arXiv:2010.08410v3 fatcat:yfxpd6o5efd6finrk2rhmvv3iq

The history of Standard ML

David MacQueen, Robert Harper, John Reppy
2020 Proceedings of the ACM on Programming Languages (PACMPL)  
The use of parametric polymorphism in its type system, together with the automatic inference of such types, has influenced a wide variety of modern languages (where polymorphism is often referred to as  ...  properties, and as a guide to łprincipledž language design. 5 LCF/ML Ð The Original ML Embedded in the LCF Theorem Prover We start with a brief look at some programming languages that influenced the design  ...  ., using an integer where a real is expected) should be easy to support.  ... 
doi:10.1145/3386336 fatcat:2hrtsaf5azfe3htngsqvxv3kre

ML-based Visualization Recommendation: Learning to Recommend Visualizations from Data [article]

Xin Qian, Ryan A. Rossi, Fan Du, Sungchul Kim, Eunyee Koh, Sana Malik, Tak Yeon Lee, Joel Chan
2020 arXiv   pre-print
Visualization recommendation seeks to generate, score, and recommend to users useful visualizations automatically, and are fundamentally important for exploring and gaining insights into a new or existing  ...  In this work, we propose the first end-to-end ML-based visualization recommendation system that takes as input a large corpus of datasets and visualizations, learns a model based on this data.  ...  model M should be able to assign a high score to this visualization and consider it as effective.  ... 
arXiv:2009.12316v1 fatcat:k3rr5muny5hn3kvrehglnplgra

Democratizing Data Science through Interactive Curation of ML Pipelines

Zeyuan Shang, Emanuel Zgraggen, Benedetto Buratti, Ferdinand Kossmann, Philipp Eichmann, Yeounoh Chung, Carsten Binnig, Eli Upfal, Tim Kraska
2019 Proceedings of the 2019 International Conference on Management of Data - SIGMOD '19  
Domain experts are often overwhelmed by such complexity, de-facto inhibiting a wider adoption of ML techniques in other fields.  ...  In Machine Learning, high-quality results are only attainable via mindful data preprocessing, hyperparameter tuning and model selection.  ...  The view of some "purist" is that the input of a NN should be the raw data and that the model -if correctly tuned, for example, by an automatic NN architecture search -should do all the rest.  ... 
doi:10.1145/3299869.3319863 dblp:conf/sigmod/ShangZBKECBUK19 fatcat:c7pkyoxqvrbvnb4ztl3vuvweim

Looper: An end-to-end ML platform for product decisions [article]

Igor L. Markov, Hanson Wang, Nitya Kasturi, Shaun Singh, Sze Wai Yuen, Mia Garrard, Sarah Tran, Yin Huang, Zehui Wang, Igor Glotov, Tanvi Gupta, Boshuang Huang (+7 others)
2022 arXiv   pre-print
To address shortcomings of prior platforms, we introduce general principles for and the architecture of an ML platform, Looper, with simple APIs for decision-making and feedback collection.  ...  We sum up experiences of platform adopters and describe their learning curve.  ...  As in medical trials, (1) we need evidence of a positive effect, (2) side-effects should be tolerable, and (3) we should not overlook evidence of side-effects.  ... 
arXiv:2110.07554v8 fatcat:idoqe2xo2jbbbicqvswhc37zxu

Intriguing Properties of Adversarial ML Attacks in the Problem Space

Fabio Pierazzi, Feargus Pendlebury, Jacopo Cortellazzi, Lorenzo Cavallaro
2020 2020 IEEE Symposium on Security and Privacy (SP)  
And where the final published version is provided on the Research Portal, if citing you are again advised to check the publisher's website for any subsequent corrections.  ...  If citing, it is advised that you check and use the publisher's definitive version for pagination, volume/issue, and date of publication details.  ...  ACKNOWLEDGEMENTS We thank the anonymous reviewers and our shepherd, Nicolas Papernot, for their constructive feedback, as well as Battista Biggio, Konrad Rieck, and Erwin Quiring for feedback on early  ... 
doi:10.1109/sp40000.2020.00073 dblp:conf/sp/PierazziPCC20 fatcat:mk34n5mqwndexh6irwqmi6fopa
« Previous Showing results 1 — 15 out of 44,685 results