人工智能（AI）終于從大學(xué)實驗室的圈子走出而成為軟件開發(fā)人員工具集的重要組成部分，也成為主流公司的關(guān)注點。??相對于學(xué)術(shù)會議上展示的研究成果，業(yè)界對于應(yīng)用人工智能的信息有著急劇增長且緊迫的需求。

所以我們組織、創(chuàng)辦了這個人工智能大會。

人工智能大會是最前沿科學(xué)研究與商業(yè)實現(xiàn)交融與碰撞的地方。它是一個供大家深入探討人工智能技巧和技術(shù)的殿堂，特別專注于人工智能技術(shù)如何應(yīng)用于真實世界中的應(yīng)用。在這里你將可以仔細剖析案例，深入鉆研最新研究成果，學(xué)習(xí)如何在自己項目中實現(xiàn)人工智能，分享在智能工程和應(yīng)用中正在出現(xiàn)的最佳實踐，揭示人工智能的局限及未被發(fā)掘的機遇，并參與討論人工智能將會如何改變商業(yè)世界的版圖。

2017年6月29日O'Reilly與Intel Nervana宣布建立合作伙伴關(guān)系，共同舉辦人工智能大會。

?

探索在業(yè)務(wù)中應(yīng)用人工智能的機會

本次人工智能大會將硅谷帶到中國
全球兩大人工智能中心相聚在一起——本次人工智能北京大會是世界范圍內(nèi)領(lǐng)先的應(yīng)用人工智能創(chuàng)新者史無前例的大聚會

多數(shù)會議議題為中文議題，部分有選擇的議題采用英文。

?

人工智能大會：將人工智能在工作中用起來

本次大會的獨特之處在于將重點放在應(yīng)用人工智能——彌合人工智能研究領(lǐng)域與產(chǎn)業(yè)商業(yè)應(yīng)用之間的差距。

只有本次北京人工智能大會才將硅谷和中國融合在一起，創(chuàng)造一次全球人工智能專家難得的相聚，并為日程安排帶來非同尋常的技術(shù)深度及實用的業(yè)務(wù)內(nèi)容。講師為來自各公司人工智能專家，包括百度、谷歌、eBay、Bonsai、Uber、微軟、阿里巴巴、亞馬遜、SAS、Unity、SalesForce、IBM、伯克利、斯坦福及牛津大學(xué)——僅為部分公司。

無論你的關(guān)注點在哪里都將在本次人工智能大會上找到：

企業(yè)中的人工智能：執(zhí)行簡報，案例研究及用例，行業(yè)特定應(yīng)用
人工智能對商業(yè)及社會的影響：自動化，安全，規(guī)范
實施人工智能項目：應(yīng)用，工具，架構(gòu)，安全
與人工智能交互：設(shè)計，指標，產(chǎn)品管理，機器人
模型及方法：增強及機器學(xué)習(xí)，TensorFlow，深度學(xué)習(xí)，GAN，自然語言處理及理解，語音識別，計算機視覺
繼在紐約和舊金山成功舉辦（均售罄）后人工智能大會于2018年4月10-13日蒞臨北京。歡迎加入，學(xué)習(xí)如何在工作中應(yīng)用人工智能。

?

為什么應(yīng)該參加

如果你希望理解人工智能將會如何影響商業(yè)世界的版圖，或是正在參與深度學(xué)習(xí)或人工智能的工作（或計劃參與）—那你應(yīng)該參加本次人工智能大會，你將能夠：

成為首批先鋒，理解如何利用這一巨變以及如何在顛覆性結(jié)局下生存。

找到跨行業(yè)跨領(lǐng)域利用已有人工智能資源的新方法。

學(xué)習(xí)如何將人工智能從學(xué)術(shù)項目引入到真正的商業(yè)應(yīng)用。

發(fā)現(xiàn)學(xué)習(xí)、招聘和就業(yè)的機會。

與其他創(chuàng)新人士和思想領(lǐng)袖面對面交流。

你會遇到誰

技術(shù)人員

算法工程師、算法科學(xué)家

CxO（x可以是分析、數(shù)據(jù)、信息、創(chuàng)新或技術(shù)）

數(shù)據(jù)科學(xué)家、數(shù)據(jù)工程師

研究科學(xué)家

軟件工程師

商業(yè)與戰(zhàn)略人士

業(yè)務(wù)分析師

業(yè)務(wù)經(jīng)理、戰(zhàn)略制定人士和決策人士

首席創(chuàng)新官、創(chuàng)新經(jīng)理、創(chuàng)新主管

首席投資官／經(jīng)理

產(chǎn)品市場推廣經(jīng)理、產(chǎn)品經(jīng)理和項目經(jīng)理

人工智能大會體驗

人工智能大會是那些應(yīng)用人工智能領(lǐng)域最聰明大腦聚集在一起的獨特機遇，共同經(jīng)歷日程密集、令人興奮的四天時間。你將會發(fā)現(xiàn)：

極具啟發(fā)性的主題演講與實用且信息量豐富的議題，探索最新人工智能進展、案例研究和最佳實踐

教學(xué)輔導(dǎo)課與培訓(xùn)課程，專門設(shè)計來深度考察工具、算法和應(yīng)用

與迅速發(fā)展的人工智能領(lǐng)域中數(shù)百名工程師、研究人員、開發(fā)人員、項目與產(chǎn)品經(jīng)理、創(chuàng)新主管以及其他深度學(xué)習(xí)專家建立關(guān)系的好機會

一個明亮的“大廳走廊”方便與會者、講師和記者就重要話題進行辯論與討論

A practical guide towards explainability and bias evaluation in machine learning

紫金大廳B（Grand Hall B)

與人工智能交互 (Interacting with AI)?,?實施人工智能 (Implementing AI)?,?模型與方法 (Models & Methods)

Alejandro Saucedo (The Institute for Ethical Ai & Machine Learning)

Undesired bias in machine learning has become a worrying topic due to the numerous high profile incidents. In this talk we demystify machine learning bias through a hands-on example. We'll be tasked to automate the loan approval process for a company, and introduce key tools and techniques from latest research that allow us to assess and mitigate undesired bias in our machine learning models.

Design Thinking for AI

紫金大廳B（Grand Hall B)

人工智能對商業(yè)及社會的影響 (Impact of AI on Business & Society)

Chris Butler (Philosophie)

Purpose, a well-defined problem, and trust from people are important factors to any system, especially those that employ AI. Chris Butler leads you through exercises that borrow from the principles of design thinking to help you create more impactful solutions and better team alignment.

基于深度學(xué)習(xí)的時間序列預(yù)測 (Deep Learning for Time Series Forecasting）

報告廳（Auditorium)

模型與方法 (Models & Methods)

Yijing Chen (Microsoft)

???????

Dmitry Pechyoni (Microsoft)

Angus Taylor (Microsoft)

Vanja Paunic (Microsoft)

Henry Zeng (Microsoft)

Almost every business today uses forecasting to make better decisions and allocate resources more effectively. Deep learning has achieved a lot of success in computer vision, text and speech processing, but has only recently been applied to time series forecasting. In this tutorial we show how and when to apply deep neural networks to time series forecasting. The tutorial will be in CHN and EN.

通過自動化機器學(xué)習(xí)民主化和加速AI落地 (Democratizing and Accelerating AI through Automated Machine Learning)

報告廳（Auditorium)

實施人工智能 (Implementing AI)?,?模型與方法 (Models & Methods)

Sujatha Sagiraju (Microsoft),?Henry Zeng (Microsoft)

Intelligent experiences powered by AI can seem like magic to users. Developing them, however, is pretty cumbersome involving a series of sequential and interconnected decisions along the way that are pretty time consuming. What if there was an automated service that identifies the best machine learning pipelines for a given problem/data? Automated machine learning does exactly that!

Analytics Zoo: Distributed Tensorflow and Keras on Apache Spark

多功能廳8A+B（Function Room 8A+B)

實施人工智能 (Implementing AI)

Zhichao Li (Intel)

In this tutorial, we will show how to build and productionize deep learning applications for Big Data using "Analytics Zoo":https://github.com/intel-analytics/analytics-zoo (a unified analytics + AI platform that seamlessly unites Spark, TensorFlow, Keras and BigDL programs into an integrated pipeline) using real-world use cases (such as JD.com, MLSListings, World Bank, Baosight, Midea/KUKA, etc.)

Building reinforcement learning models and AI applications with Ray

多功能廳8A+B（Function Room 8A+B)

英文講話 (Presented in English)

Richard Liaw (UC Berkeley RISELab)

Ray is a general purpose framework for programming your cluster. We will lead a deep dive into Ray, walking you through its API and system architecture and sharing application examples, including several state-of-the-art AI algorithms.

Session

多功能廳2（Function Room 2)

Session

紫金大廳B（Grand Hall B)

Session

紫金大廳B（Grand Hall B)

Using deep learning and time-series forecasting to reduce transit delays

紫金大廳B（Grand Hall B)

與人工智能交互 (Interacting with AI)?,?實施人工智能 (Implementing AI)?,?英文講話 (Presented in English)

Mark Ryan (IBM),

Alina Li Zhang (Nobul)

Toronto is unique among North American cities for having a legacy streetcar network as an integral part of its transit system. This means streetcar delays are a major contributor to gridlock in the city. Using deep learning and time-series forecasting, we'll show how streetcar delays can be predicted... and prevented.

Hacking Humans Made Easy: Signal Processing + AI + Video

紫金大廳B（Grand Hall B)

人工智能對商業(yè)及社會的影響 (Impact of AI on Business & Society)?,?英文講話 (Presented in English)

David Maman (Binah.ai)

Zero-day attacks. IoT-based botnets. Cybercriminal AI v. cyberdefender AI. While these won’t be going away, they aren’t the biggest worry we have in cybercrime. Hacking humans is. The combination of mere minutes of video, signal processing, remote heart rate monitoring, AI, machine learning, and data science can identify a person’s health vulnerabilities, which evildoers can make worse.

AI技術(shù)在外賣個性化場景中的落地與思考

報告廳（Auditorium)

實施人工智能 (Implementing AI)

劉先生 (美團)

該議題的內(nèi)容包括： 1.外賣個性化場景：個性化搜索，個性化推薦 2.個性化產(chǎn)品形態(tài)包括：商家、商品、套餐等 3.外賣個性化中應(yīng)用的AI技術(shù)包括：NLP，DNN，圖像技術(shù)，強化學(xué)習(xí) 4.針對外賣業(yè)務(wù)的特點，介紹個性化場景中，幾項重點AI技術(shù)的落地、挑戰(zhàn)與思考

ONNX:開放和互操作平臺讓AI無處不在(AI everywhere: Open and interoperable platform for AI with ONNX)

報告廳（Auditorium)

實施人工智能 (Implementing AI)?,?模型與方法 (Models & Methods)

Prasanth Pulavarthi (Microsoft),?Henry Zeng (Microsoft)

An open and interoperable ecosystem enables you to choose the framework that's right for you, train at scale, and deploy to cloud and edge. ONNX provides a common format supported by many popular frameworks and hardware accelerators. This session provides an introduction to ONNX and its core concepts. The session will be delivered in English and Chinese jointly.

Exciting new features in TensorFlow 2.0

報告廳（Auditorium)

實施人工智能 (Implementing AI)

Tiezhen Wang (Google)

TensorFlow 2.0 is a major milestone with a focus on ease of use. This talk will give a in depth introduction to the new exciting features and best practices. Topics such as distributed strategies and edge deployment (TensorFlow Lite and TensorFlow.js) will also be covered.

自動機器學(xué)習(xí)（automated machine learning）技術(shù)的實踐與應(yīng)用

報告廳（Auditorium)

實施人工智能 (Implementing AI)?,?模型與方法 (Models & Methods)

Hui Xue (微軟亞洲研究院)

人工智能在過去的幾年里飛速發(fā)展，但是機器學(xué)習(xí)的實踐和應(yīng)用需要消耗一定的人力和時間。例如，如何去做特征選擇，如何設(shè)計一個適合該任務(wù)的神經(jīng)網(wǎng)絡(luò)模型等等。而自動機器學(xué)習(xí)技術(shù)，可以幫助開發(fā)者和機器學(xué)習(xí)實戰(zhàn)者，縮短開發(fā)周期，提高效率。我們的介紹主要包括：自動機器學(xué)習(xí)技術(shù)的進展；我們開源的自動機器學(xué)習(xí)開源庫Neural Network Intelligence; 如何利用自動機器學(xué)習(xí)的技術(shù)，在產(chǎn)品和應(yīng)用上提高效率，節(jié)省所需的時間和縮短周期。我們會在最后一部分，分享一些利用自動特征選擇，自動參數(shù)調(diào)整以及模型架構(gòu)搜索上的成功案例。

打造A.I.閉環(huán) 引領(lǐng)產(chǎn)業(yè)變革

報告廳（Auditorium)

人工智能對商業(yè)及社會的影響 (Impact of AI on Business & Society)

溫浩 (云從科技)

AI企業(yè)發(fā)展應(yīng)該是一個從學(xué)術(shù)研究、行業(yè)驗證、商業(yè)落地、行業(yè)平臺到智能生態(tài)的一層層深入過程，這也是人工智能企業(yè)理想的發(fā)展階段。 云從科技計劃打造核心技術(shù)閉環(huán)，讓計算機更好地服務(wù)人類。并將全面降低人工智能準入門檻，讓“AI普惠”成為可能。

The Unreasonable Effectiveness of Transfer Learning on NLP

多功能廳2（Function Room 2)

模型與方法 (Models & Methods)?,?英文講話 (Presented in English)

David Low (Pand.ai)

Transfer Learning has been proven to be a tremendous success in the Computer Vision field as a result of ImageNet competition. In the past months, the Natural Language Processing field has witnessed several breakthroughs with transfer learning, namely ELMo, Transformer, ULMFit and BERT. In this talk, David will be showcasing the use of transfer learning on NLP application with SOTA accuracy.

The future of machine learning is decentralized

多功能廳2（Function Room 2)

英文講話 (Presented in English)?,?隱私、道德與規(guī)范 (Privacy, Ethics, & Compliance)

Alex Ingerman (Google)

Federated Learning is the approach of training ML models across a fleet of participating devices, without collecting their data in a central location. Alex Ingerman introduces Federated Learning, compares the traditional and federated ML workflows, and explores the current and upcoming use cases for decentralized machine learning, with examples from Google's deployment of this technology.

Session

多功能廳2（Function Room 2)

Efficient Deep Learning for the Edge

紫金大廳B（Grand Hall B)

模型與方法 (Models & Methods)?,?英文講話 (Presented in English)

Bichen Wu (UC Berkeley)

The success of deep neural networks is attributed to three factors: stronger computing capacity, more complex neural networks, and more data. These factors, however, are usually not available with the edge applications as autonomous driving, AR/VR, IoT, and so on. In this talk we discuss how we apply AutoML, SW/HW codesign, domain adaptation to solve these problems.

Session

多功能廳2（Function Room 2)

Session

多功能廳5A+B（Function Room 5A+B)

A Humane AI Solution to Improve Debt Collection

多功能廳5A+B（Function Room 5A+B)

實施人工智能 (Implementing AI)

Ying Liu (Abakus 鯨算科技(Wecash閃銀）)

AI debt collection platform of Abakus provides a friendly and humane product solution which is designed for people who work in the live agents of the organization in the frontline. The agent training of the organization could be enhanced more smoothly with an AI friendly culture. It has been proved in our experiment that the performance of the collection assistants has been highly improved.

線上財富管理領(lǐng)域中的AI應(yīng)用

多功能廳5A+B（Function Room 5A+B)

執(zhí)行簡報/最佳實踐 (Executive Briefing/Best Practices)

楊博理 (宜信大數(shù)據(jù)創(chuàng)新中心)

AI技術(shù)是線上財富管理領(lǐng)域中不可或缺的一環(huán)。在這個演講中，我會將財富管理進一步細分為投資和實現(xiàn)財務(wù)目標兩個方面，并分別講解AI技術(shù)在這兩個細分層面上的應(yīng)用問題。對于投資而言，一些具備強金融邏輯的變量可能更適合使用機器學(xué)習(xí)進行預(yù)測。而在資產(chǎn)價格的預(yù)測上，可以嘗試使用AI和大數(shù)據(jù)技術(shù)獲取更多的有價值信息。對于實現(xiàn)財務(wù)目標而言，基于NLP技術(shù)的語義理解、引導(dǎo)式對話是理解用戶的關(guān)鍵，基于AI和大數(shù)據(jù)的KYC也是判斷用戶狀態(tài)的有效工具，而一個融合了財務(wù)規(guī)劃、投資和精算知識的專家系統(tǒng)則是定制級規(guī)劃的核心。

Real-time product recommendations leveraging deep learning on Apache Spark in Office Depot

多功能廳5A+B（Function Room 5A+B)

實施人工智能 (Implementing AI)?,?模型與方法 (Models & Methods)

Guoqiong Song (Intel)

Luyang Wang (Office Depot)

Jennie Wang (Intel)

?Jing (Nicole) Kong (Office Depot)

To show case how to build efficient recommender systems for e-commerce industry using deep learning technologies

Session

多功能廳5A+B（Function Room 5A+B)

Session

多功能廳6A+B (Function Room 6A+B)

Forecasting Customer Activities with Dilated Convolution Neural Networks: Use Case and Best Practices

多功能廳6A+B (Function Room 6A+B)

模型與方法 (Models & Methods)

Tao Lu (Microsoft)

Chenhui Hu (Microsoft)

Forecasting customer activities is one of the most important and common business problems. In Microsoft Azure Identity team, we forecast customer behavior based on billions of user activities. We will share how we improve 25% of forecasting accuracy with dilated convolutional neural networks and reduce 80% of the time in development with the best practices of time series forecasting.

Confidence Estimation for Deep Neural Networks

多功能廳6A+B (Function Room 6A+B)

模型與方法 (Models & Methods)

Jialin Jiao (Pony.ai)

While deep learning has been in the center of AI with unprecedentedly great results, predictions of deep neural networks usually do not come with a reliable and well-calibrated confidence score. Wrong but confident predictions place great threads to critical real-life applications, e.g. self-driving car. This talk is a tutorial/comparison of confidence estimation methods for deep neural networks.

Session

多功能廳6A+B (Function Room 6A+B)

Session

多功能廳6A+B (Function Room 6A+B)

快速社交 (Speed Networking)

本次人工智能會議上午8:00-8:30可以和希望社交的與會來賓見面。我們將在周五主題演講之前搞一個非正式快速社交活動。一定記得帶名片參加活動。

Best practice of building data science platform in Rakuten

多功能廳2（Function Room 2)

與人工智能交互 (Interacting with AI)?,?實施人工智能 (Implementing AI)

安敖日奇朗 (Rakuten, Inc.)

?TzuLin Chin (Rakuten, Inc.)

Data Science Platform is a suite of tools for exploring data, training models, and running GPU/CPU compute jobs in an isolated container environment. It provides one click machine learning environment creation, powerful job scheduler and flexible "function as a service" component. It runs on Kubernetes and supports both on-premises and cloud environment, as well as hybrid mode.

Session with Joe Spisak

紫金大廳B（Grand Hall B)

Joseph Spisak (Facebook)

Session with Joe Spisak

Session

紫金大廳B（Grand Hall B)

ML Ops and Kubeflow Pipeline

紫金大廳B（Grand Hall B)

實施人工智能 (Implementing AI)?,?模型與方法 (Models & Methods)

Kazunori Sato (Google)

Creating an ML model is just a starting point. To bring the technology into production service, you need to solve various real-world issues such as: building a data pipeline for continuous training, automated validation of the model, version control of the model, scalable serving infra, and ongoing operation of the ML infra with monitoring and alerting.

AI at ING: the why, how, and what of a data-driven enterprise

紫金大廳B（Grand Hall B)

執(zhí)行簡報/最佳實踐 (Executive Briefing/Best Practices)?,?文化與組織 (Culture & Organization)?,?案例研究 (Case Studies)

Bas Geerdink (ING)

AI is at the core of ING’s business. We are a data-driven enterprise, with ‘a(chǎn)nalytics skills’ as a top strategic priority. We are investing in AI, big data, and analytics to improve business processes such as balance forecasting, fraud detection and customer relation management. In this talk, Bas will give an overview of the use cases and technology to inspire the audience!

Analytics Zoo: Distributed TensorFlow in Production on Apache Spark

報告廳（Auditorium)

實施人工智能 (Implementing AI)

Yang Wang (Intel)

We will introduce Analytics Zoo, a unified analytics + AI platform for distributed TensorFlow, Keras and BigDL on Apache Spark, designed for production environment. It enables easy deployment, high performance and efficient model serving for deep learning applications.

Sparkling: 基于Apache Spark進行一站式機器學(xué)習(xí)

報告廳（Auditorium)

實施人工智能 (Implementing AI)

Yiheng Wang (Tencent)

機器學(xué)習(xí)項目在企業(yè)中實際落地往往涉及到復(fù)雜工作流構(gòu)建和數(shù)據(jù)管理，以及多種工具的整合。而且隨著數(shù)據(jù)規(guī)模的增加，團隊規(guī)模的擴大，這一任務(wù)更具挑戰(zhàn)性。Apache Spark是業(yè)界流行的大數(shù)據(jù)框架，被廣泛的應(yīng)用在海量數(shù)據(jù)的分析處理。本議題將介紹我們在騰訊云上如何基于Apache Spark為客戶建立一個一站式機器學(xué)習(xí)平臺的相關(guān)工作。主要內(nèi)容包括多種數(shù)據(jù)源的接入，構(gòu)建復(fù)雜數(shù)據(jù)管線，利用數(shù)據(jù)可視化理解數(shù)據(jù)，通過可插拔的機制使用各種流行的機器學(xué)習(xí)框架，以及部署和監(jiān)控模型。我們也會分享在這一過程中遇到的問題和挑戰(zhàn)。聽眾也可以了解到，通過這種和大數(shù)據(jù)緊密結(jié)合的一站式機器學(xué)習(xí)，用戶可以怎樣更加高效的建立和管理他們的機器學(xué)習(xí)項目，從而加速了機器學(xué)習(xí)在業(yè)務(wù)中的落地。

AVA: a Cloud-Native Deep Learning Platform at Qiniu

報告廳（Auditorium)

實施人工智能 (Implementing AI)

Chaoguang Li (Qiniu)

Bin Fan (Alluxio)

Atlab Lab at Qiniu Cloud focuses on deep learning for computer vision. Our team has built a high-performance and cost-effective training platform based on Cloud for deep learning, called AVA, which deeply integrates open source software stack including Tensorflow, Caffe, Alluxio and KODO our own cloud object storage.

Query the planet: Geospatial big data analytics at Uber

報告廳（Auditorium)

實施人工智能 (Implementing AI)

Zhenxiao Luo (Uber)

One of the distinct challenges for Uber is analyzing geospatial big data. Locations and trips provide insights that can improve business decisions and better serve users. Geospatial data analysis is particularly challenging, especially in a big data scenario. For these analytical requests, we must achieve efficiency, usability, and scalability in order to meet user needs and business requirements.

保險中的機器學(xué)習(xí)實踐

報告廳（Auditorium)

實施人工智能 (Implementing AI)

鞠芳 (中國人壽研發(fā)中心)

分析保險行業(yè)人工智能發(fā)展情況及現(xiàn)有數(shù)據(jù)特性，評估機器學(xué)習(xí)模型構(gòu)建的主流工具、語言、算法?？偨Y(jié)基于機器學(xué)習(xí)技術(shù)，實現(xiàn)一個保險業(yè)人工智能場景的全流程——從場景研討、數(shù)據(jù)加工提取到模型構(gòu)建、模型效果評估、模型落地實施。以一個真實的機器學(xué)習(xí)模型項目為例，介紹整個方法論不同環(huán)節(jié)中各方人員的參與工作內(nèi)容和比例，探討特征穩(wěn)定性、樣本不均衡、參數(shù)選擇、模型可解釋性等環(huán)節(jié)的難點及嘗試方案。為金融或者其他行業(yè)的機器學(xué)習(xí)項目落地提供參考和指導(dǎo)。

Achieving Salesforce-Scale Machine Learning in Production

多功能廳2（Function Room 2)

執(zhí)行簡報/最佳實踐 (Executive Briefing/Best Practices)

Sarah Aerni (Salesforce Einstein)

At Salesforce Einstein data science is an agile partner to over 100,000 customers. How do we achieve this scale? We share lessons learned in business, technology and process along the way. Via use cases, oft-missed foundational elements for deployment, and the evaluations that must happen along the way, we will share how to achieve and sustain models in production, and where to go from there.

AI in the Newsroom: Helping Authors Do What They Do Best

多功能廳2（Function Room 2)

與人工智能交互 (Interacting with AI)

Boris Yakubchik (Forbes)

Forbes has been working on a custom publishing platform and we have integrated machine learning into it from the ground up. We provide an assistant to our writers that maintains a live connection and helps them as they work. In this talk we’ll share our system design, discuss choices we had to make along the way, and walk through several examples.

Session

多功能廳2（Function Room 2)

Session

紫金大廳B（Grand Hall B)

Using ML for personalizing Food Recommendations

多功能廳2（Function Room 2)

模型與方法 (Models & Methods)

Maulik Soneji (Go-jek)

Jewel James (Go-jek)

The story of how we prototyped the search framework that personalizes the restaurant search results by using ML to learn what constitutes a relevant restaurant given a user's purchasing history

自動駕駛技術(shù)是如何應(yīng)用于新潮傳媒、新零售行業(yè)

多功能廳5A+B（Function Room 5A+B)

實施人工智能 (Implementing AI)

Li Yuan (Perceptin 深圳普思英察科技有限公司)

如何令自動駕駛技術(shù)落地并結(jié)合新潮傳媒以及新零售業(yè)務(wù)，相關(guān)的技術(shù)是如何實現(xiàn)，商業(yè)模式是什么以及如何通過人工只能技術(shù)提升行業(yè)的效率。

How China Telecom combats financial frauds with Adversarial AutoEncoder?

多功能廳5A+B（Function Room 5A+B)

案例研究 (Case Studies)

Weisheng Xie (China Telecom BestPay Co., Ltd)

We exploit the good representation capability of AAE (Adversarial AutoEncoder) in our risk factors modeling in fighting a special kind of financial frauds. It's one step of our long stack of unsupervised tasks, yet it's proved to be efficient and effective in our practice.

人工智能病理影像輔助診斷系統(tǒng)——從方法到落地

多功能廳5A+B（Function Room 5A+B)

與人工智能交互 (Interacting with AI)?,?實施人工智能 (Implementing AI)

王書浩 (透徹影像)

病理學(xué)是醫(yī)學(xué)診斷的“金標準”，病理報告對于臨床醫(yī)生提供進一步治療策略至關(guān)重要。一位能夠獨立發(fā)病理報告的病理醫(yī)師需要10年以上的培養(yǎng)周期，我國目前共有約1萬名注冊在案的病理醫(yī)師，根據(jù)WHO的要求，人才缺口為4-9萬人。使用人工智能來輔助病理醫(yī)師對樣本進行診斷，不僅能夠大幅提高醫(yī)師的診斷效率，而且可以減少漏診，提高診斷準確率。數(shù)字化的病理影像能夠觀察到組織的細胞形態(tài)，在最高倍數(shù)字掃描時，文件尺寸達到GB量級，需要從人工智能和系統(tǒng)工程的層面去應(yīng)對這些挑戰(zhàn)。在這個演講中，我們將從人工智能系統(tǒng)的構(gòu)建方法入手，介紹透徹影像與中國人民解放軍總醫(yī)院在消化道病理影像輔助系統(tǒng)研發(fā)過程中的技術(shù)細節(jié)。同時，我們將分享診斷系統(tǒng)從部署到落地使用的一些經(jīng)驗。

Session

多功能廳5A+B（Function Room 5A+B)

基于目標檢測的智能化成礦異常信息提取

多功能廳5A+B（Function Room 5A+B)

實施人工智能 (Implementing AI)

李蒼柏 (中國地質(zhì)科學(xué)院礦產(chǎn)資源研究所)

礦床所在的位置往往伴隨著地質(zhì)、地球物理、地球化學(xué)、遙感異常，因此，這些異常所在的位置也往往伴隨著礦床的存在。所以，在找礦工作當(dāng)中，一個重要的過程便是在地、物、化、遙數(shù)據(jù)中尋找異常，并將其整合，得出該區(qū)域成礦的概率，從而推斷出靶區(qū)所在的位置。但傳統(tǒng)方法并未考慮空間中點與點之間的相關(guān)關(guān)系。而卷積神經(jīng)網(wǎng)絡(luò)中的卷積和池化方法，充分考慮了點與點之間的相關(guān)關(guān)系。但單純使用卷積神經(jīng)網(wǎng)絡(luò)只能進行特征提取，不能圈定異常所在的區(qū)域。因此，特將目標檢測的相關(guān)算法引入其中，從而圈定異常所在的區(qū)域。

Session

多功能廳6A+B (Function Room 6A+B)

在邊緣實現(xiàn)深度學(xué)習(xí)

多功能廳6A+B (Function Room 6A+B)

模型與方法 (Models & Methods)

陳玉榮 (Intel)

深度學(xué)習(xí)在許多領(lǐng)域尤其是視覺識別/理解方面取得了巨大突破，但它在訓(xùn)練和部署方面都存在一些挑戰(zhàn)。本講座將介紹我們通過高效CNN算法設(shè)計、領(lǐng)先DNN模型壓縮技術(shù)和創(chuàng)新部署時DNN網(wǎng)絡(luò)結(jié)構(gòu)優(yōu)化來解決深度學(xué)習(xí)部署挑戰(zhàn)的前沿研究成果。

非監(jiān)督學(xué)習(xí)在大規(guī)模圖譜上的案例應(yīng)用和開源算法剖析

多功能廳6A+B (Function Room 6A+B)

模型與方法 (Models & Methods)

Mingxi Wu (TigerGraph)

圖數(shù)據(jù)上的非監(jiān)督學(xué)習(xí)在激活大數(shù)據(jù)的經(jīng)濟價值上有著廣泛和不可替代的作用。 PageRank能夠發(fā)掘重要的實體, 社區(qū)發(fā)掘(community detection)可以找到具有某種特性的群體，緊密度中心性算法(Closeness Centrality)可以自動找到遠離群體的個體。所有這些算法都是非監(jiān)督的學(xué)習(xí)。 我們分享一些具體客戶案例來展示他們的價值，同時分享怎樣在大數(shù)據(jù)上靈活應(yīng)用這些開源算法。

PAI Tensor Accelerator and Optimizer---Yet Another Deep Learning Compiler

多功能廳6A+B (Function Room 6A+B)

與人工智能交互 (Interacting with AI)?,?實施人工智能 (Implementing AI)

楊軍 (阿里巴巴)

龍國平 (Alibaba)

本次演講會介紹阿里計算平臺PAI團隊過去一年多時間里在深度學(xué)習(xí)編譯器領(lǐng)域的技術(shù)工作進展----PAI TAO(Tensor Accelerator and Optimizer)。PAI-TAO采用通用編譯優(yōu)化技術(shù)，來解決PAI平臺所承載的多樣性AI workload面臨的訓(xùn)練及推理需求的性能優(yōu)化問題，在部分workload上獲得了20%到4X不等的顯著加速效果，并且基本作到用戶層全透明，在顯著提升平臺效率性能的同時也有效照顧了用戶的使用慣性。目前PAI-TAO已經(jīng)先后用于支持阿里內(nèi)部搜索、推薦、圖像、文本等多個業(yè)務(wù)場景的日常訓(xùn)練及推理需求。

Low precision inference on Intel Architecture

多功能廳6A+B (Function Room 6A+B)

實施人工智能 (Implementing AI)

Lei Xia (Intel)

Vector Neural Network Instructions or VNNI is the new Intel instruction set for low precision AI inference inside next generation Xeon platform. This lecture is to introduce the features of the VNNI and Intel software tools to support developers to use this new instruction set to accelerate inference with INT8.

快速社交 (Speed Networking)

在本次人工智能大會上與尋求聯(lián)系的與會者會面。會議將在周四主題演講之前舉行一個非正式的快速社交活動。一定要帶上自己的名片來享受社交活動。

所有培訓(xùn)課程安排在周二周三上午9:00至下午17:00。為了保證高水平的動手學(xué)習(xí)和與導(dǎo)師互動機會，全部培訓(xùn)課程都會限制在一定規(guī)模。?白金門票和培訓(xùn)門票不包含周三教學(xué)輔導(dǎo)課。

09:00 - 17:00 Tuesday, April 10 & Wednesday, April 11
用TensorFlow進行深度學(xué)習(xí)
Location: 多功能廳5A+B（Function Room 5A+B)
TensorFlow是一個流行的深度學(xué)習(xí)的工具。我們會介紹TensorFlow的流程圖、學(xué)習(xí)使用它的Python API，并展示它的用處。我們會從簡單的機器學(xué)習(xí)算法開始，然后實現(xiàn)神經(jīng)網(wǎng)絡(luò)。我們還會討論一些真實的深度學(xué)習(xí)的應(yīng)用，包括機器視覺、文本處理和生成型網(wǎng)絡(luò)。

09:00 - 17:00 Tuesday, April 10 & Wednesday, April 11
用Deeplearning4j框架構(gòu)建神經(jīng)網(wǎng)絡(luò)分析時間序列
Location: 多功能廳6A+B (Function Room 6A+B)
Secondary topics: ?深度學(xué)習(xí)（Deep Learning）, 金融服務(wù) (Financial Services)
在分析時間序列或者序列數(shù)據(jù)方面循環(huán)神經(jīng)網(wǎng)絡(luò)（RNN）已經(jīng)被證明是非常有效的，那么在實際的案例中如何才能把循環(huán)神經(jīng)網(wǎng)絡(luò)（RNN）的優(yōu)點發(fā)揮出來吶？這里將演示如何用Deeplearning4j框架構(gòu)建循環(huán)神經(jīng)網(wǎng)絡(luò)（RNN）來解決時間序列的問題。

09:00 - 17:00 Tuesday, April 10 & Wednesday, April 11
人工智能和金融科技：量化金融信用與欺詐風(fēng)險的評估
Location: 多功能廳3A+B（Function Room 3A+B)
您想了解金融企業(yè)是怎樣利用大數(shù)據(jù)和人工智能技術(shù)來畫像個人行為并檢測欺詐用戶的嗎？互聯(lián)網(wǎng)金融幕后的量化分析流程是怎么楊的？個人信用是怎樣通過大數(shù)據(jù)被量化的？在實踐過程中，機器學(xué)習(xí)算法的應(yīng)用存在著哪些需要關(guān)注的方面？怎樣通過圖譜分析來融合多維數(shù)據(jù)，為我們區(qū)分正常用戶和欺詐用戶？ 這套輔導(dǎo)課基于清華大學(xué)交叉信息研究院2017年春天新開設(shè)的一門"量化金融信用與風(fēng)控分析”研究生課。其中會用LendingClub的真實借貸數(shù)據(jù)做為案例，解說一些具體模型的實現(xiàn)。
?? ?