大會實作訓練

透過在 GTC 參與由講師主持的訓練課程,拓展並增進您的實作技能。

NVIDIA 深度學習機構 (DLI) 提供人工智慧、加速運算和加速資料科學的實作訓練課程,協助開發人員、資料科學家及其他專家解決最具挑戰性的難題。這些深度實作坊由各領域專家講授,提供領先業界的技術知識,為個人和組織帶來突破性成果。完成全天實作坊後,便可取得 NVIDIA DLI 認證,證明您具備該主題的技術能力,並有助於加速職涯成長。

3 人以上團報或一人報滿 3 堂課即可享每堂 $149 美元的限時優惠,欲享相關優惠請來信 GTC_Taiwan@nvidia.com 洽詢。

由講師帶領的實作坊

  • 台灣
  • 北美
  • 歐洲
  • 印度
  • 日本
  • 韓國

4 月 14 日(三)| 09:00 - 17:00 台灣時間 (UTC+8)

深度學習基礎理論與實踐(DLIW2471)

學習進行電腦視覺和自然語言處理方面的實作練習,藉此瞭解深度學習的運作方式。將會從零開始訓練深度學習模型、學習工具和技巧,以追求高度準確的成果。也會學習如何運用免費的頂尖預先訓練好的模型,如此可以節省時間,並讓深度學習應用程式即刻運作。

4 月 15 日(四)| 09:00 - 17:00 台灣時間 (UTC+8)

多 GPU 深度學習基本原理(DLIW2472)

本實作課程將教導使用多個 GPU 訓練深度神經網絡的技術,以縮短資料密集型應用程式所需的訓練時間。通過使用深度學習工具、框架和工作流程來進行神經網絡訓練,您將學習實現 Horovod 多 GPU 的概念,以減少編寫高效能的分佈式軟體的複雜性。

4 月 16 日(五)| 09:00 - 17:00 台灣時間 (UTC+8)

運用於智慧影片分析的深度學習技術(DLIW2473)

利用 NVIDIA DeepStream 實行智慧影像分析(IVA ),進行偵測、辨識、追蹤和定義物體的軌跡,並透過訓練和評估深度學習模型,搭配轉移學習技術來提升模型的效率和準確性,讓大規模影像分析變得輕而易舉。

Tues, April 13 | 06:00-14:00
(PDT, UTC-7)

Fundamentals of Deep Learning (DLIW2323)

Learn how deep learning works through hands-on exercises in computer vision and natural language processing. You will train deep learning models from scratch, learning tools and tricks to achieve highly accurate results. You’ll also learn to leverage freely available...

Tues, April 13 | 09:00-17:00
(PDT, UTC-7)

Fundamentals of Accelerated Computing with CUDA C/C++ (DLIW2327)

Learn how to accelerate and optimize existing C/C++ CPU-only applications to leverage the power of GPUs using the most essential CUDA techniques and the Nsight Systems profiler.

Wed, April 14 | 06:00-14:00
(PDT, UTC-7)

Fundamentals of Accelerated Data Science with RAPIDS (DLIW2326)

Learn how to perform multiple analysis tasks on large data sets using RAPIDS, a collection of data science libraries that allows end-to-end GPU acceleration for data science workflows.

Wed, April 14 | 09:00-17:00
(PDT, UTC-7)

Building Transformer-Based Natural Language Processing Applications (DLIW2325)

Learn how to use Transformer-based natural language processing models for text classification tasks, such as categorizing documents. You will also get insight on how to leverage Transformer-based models for named-entity recognition (NER) tasks and...

Thurs, April 15 | 06:00-14:00
(PDT, UTC-7)

Applications of AI for Anomaly Detection (DLIW2328)

Learn to detect anomalies in large data sets to identify network intrusions using supervised and unsupervised machine learning techniques, such as accelerated XGBoost, autoencoders, and generative adversarial networks (GANs).

Thurs, April 15 | 09:00-17:00
(PDT, UTC-7)

Fundamentals of Deep Learning for Multi-GPUs (DLIW2324)

Learn how to use multiple GPUs to train neural networks and effectively parallelize training of deep neural networks using TensorFlow.

Fri, April 16 | 06:00-14:00
(PDT, UTC-7)

Deep Learning for Autonomous Vehicles - Perception (DLIW2329)

Learn how to design, train, and deploy deep neural networks and optimize perception components for autonomous vehicles using the NVIDIA DRIVE™ development platform.

Fri, April 16 | 09:00-17:00
(PDT, UTC-7)

Accelerating CUDA C++ Applications with Multiple GPUs (DLIW2322)

Learn how to write CUDA C++ applications that efficiently and correctly utilize all available GPUs in a single node, dramatically improving the performance of applications and making the most cost-effective use of systems with multiple GPUs.

Mon, April 12 | 09:00-17:00
(CEST, UTC+2)

Fundamentals of Deep Learning (DLIW2315)

Learn how deep learning works through hands-on exercises in computer vision and natural language processing. You will train deep learning models from scratch, learning tools and tricks to achieve highly accurate results. You’ll also learn to leverage freely available...

Tues, April 13 | 09:00-17:00
(CEST, UTC+2)

Applications of AI for Anomaly Detection (DLIW2317)

Learn to detect anomalies in large data sets to identify network intrusions using supervised and unsupervised machine learning techniques, such as accelerated XGBoost, autoencoders, and generative adversarial networks (GANs).

Tues, April 13 | 09:00-17:00
(CEST, UTC+2)

Fundamentals of Accelerated Computing with CUDA C/C++ (DLIW2314)

Learn how to accelerate and optimize existing C/C++ CPU-only applications to leverage the power of GPUs using the most essential CUDA techniques and the Nsight Systems profiler.

Wed, April 14 | 09:00-17:00
(CEST, UTC+2)

Building Transformer-Based Natural Language Processing Applications (DLIW2319)

Learn how to use Transformer-based natural language processing models for text classification tasks, such as categorizing documents. You will also get insight on how to leverage Transformer-based models for named-entity recognition (NER) tasks and...

Wed, April 14 | 09:00-17:00
(CEST, UTC+2)

Deep Learning for Autonomous Vehicles - Perception (DLIW2318)

Learn how to design, train, and deploy deep neural networks and optimize perception components for autonomous vehicles using the NVIDIA DRIVE™ development platform.

Thurs, April 15 | 09:00-17:00
(CEST, UTC+2)

Accelerating CUDA C++ Applications with Multiple GPUs (DLIW2321)

Learn how to write CUDA C++ applications that efficiently and correctly utilize all available GPUs in a single node, dramatically improving the performance of applications and making the most cost-effective use of systems with multiple GPUs.

Thurs, April 15 | 09:00-17:00
(CEST, UTC+2)

Fundamentals of Accelerated Data Science with RAPIDS (DLIW2320)

Learn how to perform multiple analysis tasks on large data sets using RAPIDS, a collection of data science libraries that allows end-to-end GPU acceleration for data science workflows.

Wed, April 14 | 09:00-17:00
(CEST, UTC+2)

Deep Learning for Autonomous Vehicles - Perception (DLIW2318)

Learn how to design, train, and deploy deep neural networks and optimize perception components for autonomous vehicles using the NVIDIA DRIVE™ development platform.

Thurs, April 15 | 09:00-17:00
(CEST, UTC+2)

Accelerating CUDA C++ Applications with Multiple GPUs (DLIW2321)

Learn how to write CUDA C++ applications that efficiently and correctly utilize all available GPUs in a single node, dramatically improving the performance of applications and making the most cost-effective use of systems with multiple GPUs.

Thurs, April 15 | 09:00-17:00
(CEST, UTC+2)

Fundamentals of Accelerated Data Science with RAPIDS (DLIW2320)

Learn how to perform multiple analysis tasks on large data sets using RAPIDS, a collection of data science libraries that allows end-to-end GPU acceleration for data science workflows.

Mon, April 12 | 09:00-17:00
(IST, UTC+5:30)

Fundamentals of Deep Learning (DLIW2474)

Learn how deep learning works through hands-on exercises in computer vision and natural language processing. You will train deep learning models from scratch, learning tools and tricks to achieve highly accurate results. You’ll also learn to leverage freely available...

Fri, April 16 | 09:00-17:00
(IST, UTC+5:30)

Building Transformer-Based Natural Language Processing Applications (DLIW2475)

Learn how to use Transformer-based natural language processing models for text classification tasks, such as categorizing documents. You will also get insight on how to leverage Transformer-based models for named-entity recognition (NER) tasks and...

Mon, April 12 | 09:00-17:00
(JST, UTC+9)

Fundamentals of Deep Learning (DLIW2476)

Learn how deep learning works through hands-on exercises in computer vision and natural language processing. You will train deep learning models from scratch, learning tools and tricks to achieve highly accurate results. You’ll also learn to leverage freely available...

Wed, April 14 | 09:00-17:00
(JST, UTC+9)

Fundamentals of Deep Learning for Multi-GPUs (DLIW2477)

Learn how to use multiple GPUs to train neural networks and effectively parallelize training of deep neural networks using TensorFlow.

Wed, April 14 | 09:00-17:00
(KST, UTC+9)

Fundamentals of Deep Learning (DLIW2478)

Learn how deep learning works through hands-on exercises in computer vision and natural language processing. You will train deep learning models from scratch, learning tools and tricks to achieve highly accurate results. You’ll also learn to leverage freely available...

Thurs, April 15 | 09:00-17:00
(KST, UTC+9)

Fundamentals of Accelerated Computing with CUDA C/C++ (DLIW2479)

Learn how to accelerate and optimize existing C/C++ CPU-only applications to leverage the power of GPUs using the most essential CUDA techniques and the Nsight Systems profiler.

預覽 DLI 體驗

 

NVIDIA 開發者計畫

獲得在任何 NVIDIA 技術平台上開發各式應用所需的最新工具和實作訓練。

gtc21-spring-web-topic-industry-page-join-slide-2-3c33-d
gtc21-spring-web-topic-industry-page-join-slide-3-3c33-d

加速新創發展

探索 GTC 新創議程,瞭解 NVIDIA Inception 計畫如何協助新創成長、提供世界一流的實作訓練,以及技術支援。

NVIDIA 開發人員計畫

獲得在任何 NVIDIA 技術平台上開發應用程序所需的最新工具和培訓。

加速您的新創公司

探索 GTC 新創頁面,以了解 NVIDIA Inception 如何透過市場進入支援、世界一流的訓練,以及技術支援來協助推動新創公司發展。

打造智慧推薦系統

深度學習技術的推薦系統,正是提供個人化線上體驗的秘方,也是零售、娛樂、醫療照護、金融與其他產業的強大決策支援工具。本次實作坊的內容,涵蓋了建立高效率推薦系統所需的基本工具和技術,也會說明如何部署 GPU 加速的解決方案以即時提供推薦。

參加本實作坊可學會:

  • 使用開放原始碼 cuDF 函式庫和 Apache Arrow 建立以內容為基礎的推薦系統
  • 使用交替最小平方 (ALS) 和 CuPy 建立協作式篩選推薦系統
  • 使用 TensorFlow 2 設計出兼具廣度和深度的神經網路,打造混合式推薦系統
  • 使用大型的稀疏資料集,達到最佳訓練和推論效能
  • 將推薦模型部署為具備高效能的網路服務
打造智慧推薦系統
打造以轉譯器為基礎的自然語言處理應用程式

打造以轉譯器為基礎的自然語言處理應用程式

探索如何將以轉譯器為基礎的自然語言處理模型運用於文字分類工作,例如分類文件。你將會學到如何利用以轉譯器為基礎的模型來處理命名實體識別 (NER) 工作,以及如何分析各種模型功能、限制和特性,以便依據指標、領域獨特性和可用資源,判斷何種模型最適合特定使用案例。

參加本實作坊可學會:

  • 瞭解文字內嵌如何在 NLP 工作中迅速進化,例如 Word2Vec、遞歸神經網路 (RNN) 型內嵌及轉譯器
  • 瞭解轉譯器架構功能 (尤其是自我注意力機制) 是如何用於建立語言模型,而無需 RNN
  • 使用自我監督技術提升 BERT、Megatron 和其他變體中的轉譯器架構,以達到優異的 NLP 成果 
  • 利用預先訓練的現代 NLP 模型處理多項工作,例如文字分類、NER 和問答系統
  • 管理推論挑戰,並部署即時應用程式的微調模型

運用 CUDA Python 加速運算的基本原理

本課程帶領你探索如何及時運用特殊化類型的 Python 函數編譯器 Numba,以在大型平行 NVIDIA GPU 上加速執行 Python 程式。

你將學會:  

  • 透過 NumPy 通用函數 (ufuncs) 使用 Numba 編譯 NVIDIA(R) CUDA(R) 核心
  • 使用 Numba 建立並啟動自訂 CUDA 核心
  • 應用關鍵 GPU 記憶體管理技術 
完成本課程後,你將能夠使用 Numba 編譯並啟動 CUDA 核心,藉以加速 NVIDIA GPU 上的 Python 應用程式。
運用 CUDA Python 加速運算的基本原理
適用於預測性維護的人工智慧應用

適用於預測性維護的人工智慧應用

瞭解如何辨識時間序列資料中的異常和故障狀況、預估對應零件的剩餘使用年限,並利用這份資訊對應異常狀況與故障條件。  

你將學會:

  • 運用預測性維護來管理故障狀況,並避免代價高昂的意外停機時間 >為了避免變成代價高昂的故障,在找出異常狀況時辨識主要挑戰
  • 使用時間序列資料預測利用 XGBoost 機器學習分類模型的結果
  • 透過長短期記憶體 (LSTM) 模型預測裝置故障,進而應用於預測性維護程序
  • 使用先前步驟的時間序列順序,嘗試以自動編碼器偵測異常狀況 完成後,你將瞭解如何使用人工智慧預測設備狀況,並預估執行維護的時間。