Towards an Optimizer for MLbase (Hướng tới một bộ tối ưu cho MLbase) - Ameet Talwalkar

Uni24h·il y a 3 mois·2669 documents

PDF5.3 MB

Téléchargement gratuit

0 téléchargements

0 j’aime

0 je n’aime pas

Génération de l'aperçu...

Suivant de cet auteurMLlib and Distributing the Singular Value Decomposition - Reza Zadeh Précédent dans le coursDistributed Computing with Spark (Tính toán phân tán với Spark) - Reza Zadeh Suivant dans le coursMLlib and Distributing the Singular Value Decomposition - Reza Zadeh

Qu'est-ce que ce document ?

Trình bày về dự án MLbase nhằm đơn giản hóa phát triển và triển khai các pipeline học máy trên nền tảng Apache Spark, cùng những thách thức đối với người dùng cuối và nhà phát triển.

Généré automatiquement · lecture rapide avant téléchargement

Nom du document: Towards an Optimizer for MLbase (Hướng tới một bộ tối ưu cho MLbase) - Ameet Talwalkar
École / Cours: Stanford University · Big Data
Auteur (dans le document): Ameet Talwalkar
Contenu: Tài liệu trình bày vấn đề về sự phức tạp và khó khăn trong việc triển khai học máy quy mô lớn và cho người dùng cuối. MLbase được đề xuất như một giải pháp để đơn giản hóa việc phát triển và triển khai pipeline ML, dựa trên Apache Spark.
Table des matières: Ce document n'a pas de table des matières claire.
Pages: 117 pages
Téléversé par: Uni24h

Contenu du document

Page 1

Towards an OpBmizer for MLbase Ameet Talwalkar Collaborators: Evan Sparks, Michael Franklin, Michael I. Jordan, Tim Kraska UC Berkeley

Page 3

Problem: Scalable implementa.ons Me difficult for ML Developers…ML Contract + Code The Language of Technical Computing ML MATLAB® is a high-level language and interactive environment for numerical com- Key Features putation, visualization, and programming. Using MATLAB, you can analyze data, t )JHIMFWFMMBOHVBHFGPSOVNFSJDBM t #VJMUJOHSBQIJDTGPSWJTVBMJ[JOHEBUB develop algorithms, and create models and DPNQVUBUJPO WJTVBMJ[BUJPO BOEBQ- BOEUPPMTGPSDSFBUJOHDVTUPNQMPUT applications. The language, tools, and built- QMJDBUJPOEFWFMPQNFOU t %FWFMPQNFOUUPPMTGPSJNQSPWJOH in math functions enable you to explore t *OUFSBDUJWFFOWJSPONFOUGPSJUFSBUJWF DPEFRVBMJUZBOENBJOUBJOBCJMJUZ multiple approaches and reach a solution FYQMPSBUJPO EFTJHO BOEQSPCMFN BOENBYJNJ[JOHQFSGPSNBODF faster than with spreadsheets or traditional TPMWJOH t 5PPMTGPSCVJMEJOHBQQMJDBUJPOTXJUI programming languages, such as C/C++ or t .BUIFNBUJDBMGVODUJPOTGPSMJOFBS DVTUPNHSBQIJDBMJOUFSGBDFT Java™.You can use MATLAB for a range of applications, including signal processing and - BMHFCSBGJMUFSJOHJOUFHSBUJPOPQUJNJ[BUJPOTUBUJTUJDTBOETPMWJOH'PVSJFSOVNFSJDBMPSEJOBSZBOBMZTJT t 'VODUJPOTCBTFEQMJDBUJPOTBMHPSJUINTGPSBOEJOUFHSBUJOHMBOHVBHFTXJUIFYUFSOBM."5-"#TVDIBTBQ$- S EJGGFSFOUJBMFRVBUJPOT communications, image and video process- +BWB /&5 BOE.JDSPTPGU®&YDFM® ing, control systems, test and measurement, computational finance, and computational biology. More than a million engineers and scientists in industry and academia use MATLAB, the language of technical computing. MATLAB Overview 2:04 Analyzing and visualizing data using the MATLAB desktop. The MATLAB environment also lets you write programs and develop algorithms and applications.

Page 5

Problem: Scalable implementa.ons Me difficult for ML Developers…ML Contract + Code ML S CHALLENGE: Can we simplify distributed ML development?

Page 9

Problem: ML is difficult for End Users… Too many Too many algorithms… knobs… Too many ways to preprocess…

Page 10

Problem: ML is difficult for End Users… Too many Too many algorithms… knobs… Too many ways to preprocess… Difficult to debug…

Page 11

Problem: ML is difficult for End Users… Too many Too many algorithms… knobs… Too many ways to preprocess… Difficult to debug… Doesn’t scale…

Affichage des 11 premières pages. Téléchargez le document pour le voir en entier.

Description

Extrait du document

Towards an OpBmizer for MLbase Ameet Talwalkar Collaborators: Evan Sparks, Michael Franklin, Michael I. Jordan, Tim Kraska UC Berkeley Problem: Scalable implementa.ons Me diﬃcult for ML ML Contract Developers… + Code ML S Problem: Scalable implementa.ons Me diﬃcult for ML ML Contract Developers… + Code The Language of Technical Computing MATLAB® is a high-level language and interactive environment for numerical computation, visualization, and programming. Using MATLAB, you can analyze data, develop algorithms, and create models and applications. The language, tools, and builtin math functions enable you to explore multiple approaches and reach a solution faster than with spreadsheets or traditional programming languages, such as C/C++ or Java™. You can use MATLAB for a range of applications, including signal processing and communications, image and video processing, control systems, test and measurement, computational finance, and computational biology. More than a million engineers and scientists in industry and academia use MATLAB, the language of technical computing. ML Key Features t )JHIMFWFMMBOHVBHFGPSOVNFSJDBM DPNQVUBUJPO WJTVBMJ[BUJPO BOEBQQMJDBUJPOEFWFMPQNFOU t OUFSBDUJWFFOWJSPONFOUGPSJUFSBUJWF FYQMPSBUJPO EFTJHO BOEQSPCMFN TPMWJOH t .BUIFNBUJDBMGVODUJPOTGPSMJOFBS BMHFCSB TUBUJTUJDT 'PVSJFSBOBMZTJT GJMUFSJOH PQUJNJ[BUJPO OVNFSJDBM JOUFHSBUJPO BOETPMWJOHPSEJOBSZ EJGGFSFOUJBMFRVBUJPOT MATLAB Overview 2:04 Analyzing and visualizing data using the MATLAB desktop. The MATLAB environment also lets you write programs and develop algorithms and applications. t #VJMUJOHSBQIJDTGPSWJTVBMJ[JOHEBUB BOEUPPMTGPSDSFBUJOHDVTUPNQMPUT t %FWFMPQNFOUUPPMTGPSJNQSPWJOH DPEFRVBMJUZBOENBJOUBJOBCJMJUZ BOENBYJNJ[JOHQFSGPSNBODF t 5PPMTGPSCVJMEJOHBQQMJDBUJPOTXJUI DVTUPNHSBQIJDBMJOUFSGBDFT t 'VODUJPOTGPSJOUFHSBUJOH."5-"# CBTFEBMHPSJUINTXJUIFY

Foire aux questions

Ce document est-il gratuit ?

Oui. « Towards an Optimizer for MLbase (Hướng tới một bộ tối ưu cho MLbase) - Ameet Talwalkar » est gratuit — il suffit de vous connecter et de cliquer sur Télécharger pour obtenir le fichier original.

Combien de pages compte ce document ?

Le document contient 117 pages, pour le cours Big Data. Vous pouvez le prévisualiser en ligne avant de le télécharger.

Puis-je prévisualiser avant de télécharger ?

Oui. Vous pouvez prévisualiser ce document directement sur cette page avec le lecteur en ligne, puis décider de le télécharger ou non.

Commentaires (0)

Se connecter pour rejoindre la discussion

Aucun commentaire pour le moment. Soyez le premier !

Towards an Optimizer for MLbase (Hướng tới một bộ tối ưu cho MLbase) - Ameet Talwalkar

Uni24h

PDF5.3 MB

0 téléchargements

117

Génération de l'aperçu...

Contenu du document

Page 1

Towards an OpBmizer for MLbase Ameet Talwalkar Collaborators: Evan Sparks, Michael Franklin, Michael I. Jordan, Tim Kraska UC Berkeley

Page 3

Page 5

Problem: Scalable implementa.ons Me difficult for ML Developers…ML Contract + Code ML S CHALLENGE: Can we simplify distributed ML development?

Page 9

Problem: ML is difficult for End Users… Too many Too many algorithms… knobs… Too many ways to preprocess…

Page 10

Problem: ML is difficult for End Users… Too many Too many algorithms… knobs… Too many ways to preprocess… Difficult to debug…

Page 11

Problem: ML is difficult for End Users… Too many Too many algorithms… knobs… Too many ways to preprocess… Difficult to debug… Doesn’t scale…

Page 12

Problem: ML is difficult for End Users… Too many Too many algorithms… knobs… Too many ways to preprocess… Difficult to debug… CHALLENGE: Can we automate ML pipeline construcBonDoesn’t scale…?

Page 14

MLbase MLbase aims to simplify development and deployment of ML pipelines Apache Spark Spark: Cluster compuBng system designed for iteraBve computaBon 4

Page 15

MLbase MLbase aims to simplify development and deployment of ML MLlib pipelines Apache Spark Spark: Cluster compuBng system designed for iteraBve computaBon MLlib: Spark’s core ML library 4

Page 16

MLbase MLbase aims to simplify development MLI and deployment of ML MLlib pipelines Apache Spark Spark: Cluster compuBng system designed for iteraBve computaBon MLlib: Spark’s core ML library MLI: API to simplify ML development 4

Page 17

MLbase MLOpt MLbase aims to simplify development MLI and deployment of ML MLlib pipelines Apache Spark Spark: Cluster compuBng system designed for iteraBve computaBon MLlib: Spark’s core ML library MLI: API to simplify ML development MLOpt: DeclaraBve layer that aims to automate ML pipeline construcBon via search over feature extractors and models 4

Page 18

MLbase MLOpt MLbase aims to simplify development MLI and deployment of ML MLlib MLOpt and MLI arepipelines Apache Spark Spark: Cluster compuBng system designed for iteraBve computaBonexperimental testbeds MLlib: Spark’s core ML library MLI: API to simplify ML development MLOpt: DeclaraBve layer that aims to automate ML pipeline construcBon via search over feature extractors and models 4

Page 20

MLlib + Scalable and fast + Simple development environment + Part of Spark’s robust ecosystem MLlib GraphX Spark SparkSQL (machine (graph) Streaming learning) Apache Spark 6

Affichage des 20 premières pages. Téléchargez le document pour le voir en entier.