Oracle将利用AI，自主云平台中的机器学习

甲骨文周二在纽约的Oracle CloudWorld上展示了Oracle Cloud Platform在人工智能和机器学习方面的进步。

用于应用程序开发，移动和机器人，集成，分析，安全性和系统管理的自主功能计划于今年上半年推出。

Oracle宣布了一系列新的平台即服务功能和特性，其中包括：

手机和机器人-

• 自学聊天机器人，观察交互模式和偏好，以自动执行经常执行的最终用户操作，从而腾出时间来完成更高生产率的任务；
• 使用机器学习从用户对话中学习，实现流畅的上下文对话的无监督智能机器人；和
• 实时将API调用自动自动缓存到最近的数据中心，以根据最终用户的位置实现最低延迟的响应。

安全与管理-

• 机器学习驱动的用户和实体行为分析，可自动隔离和消除可疑和恶意用户；
• 预防性控制措施，可拦截结构化和非结构化数据存储库之间的数据泄漏；和

· 跨日志，性能，用户体验和配置数据的统一数据存储库，并带有应用的AI和ML，可自动设置和管理性能以及安全监控元数据。

Oracle Digital Assistant

Oracle还展示了Oracle Digital Assistant，它将在CRM，ERP，HCM，自定义应用程序和商业智能数据之间提供集中式通信。

助手将使用AI关联数据并自动执行用户行为。

其功能包括：

· 与基于语音的设备和软件集成，例如Amazon Echo（Alexa），Apple Siri，Google Home and Speech，Harman Kardon（Cortana）和Microsoft Cortana；

· 基于深度神经网络的机器学习算法可处理来自基于语音的设备的消息，以了解最终用户的输入并采取行动；

· 智能路由到Oracle Bots，具有处理最终用户输入的知识；和

· 助手可用于自学习的用户行为，上下文，首选项和例程的深入了解，以便代表用户对所有数据集进行推荐和自动化。

Constellation Research 首席分析师Holger Mueller表示，Assistant的交叉应用能力“是与众不同的” 。

但是，他对《电子商务时报》表示“我很怀疑，因为甲骨文尚未宣布任何神经网络。”

Enderle Group的首席分析师Rob Enderle说，“如果它真的有效的话，助手将引人注目。”

也就是说，“该领域的主要硬件厂商已经拥有自己的后端”，他告诉《电子商务时报》。

恩德勒指出，多年来，亚马逊，谷歌，微软和IBM一直不是甲骨文最大的拥护者。他说：“我认为诺言远远超出了现实。”

自治数据库

甲骨文还宣布了即将推出的自治数据库服务。使用高级AI和ML的自治数据库于10月首次发布。

今年将发布多种自治数据库服务，每种服务都针对特定的工作负载：

· Oracle自主数据仓库云服务用于分析；

· 适用于事务和混合工作负载的Oracle自治数据库OLTP；和

· Oracle Autonomous NoSQL数据库，用于快速大规模读写。

腿长还是高个子？

星座研究公司的穆勒（Mueller）指出：“自主Oracle确实领先于所有人。” “此公告现在使自治数据库更加完善，后者在安全性和旧代码风险方面存在差距。”

但是，总的来说，甲骨文的声明“读起来像个废话，” Enderle说，“人们忘记了甲骨文以宣称自己具备的功能早于真正拥有功能而闻名。”

他补充说，实现其自主云承诺“将需要Oracle尚未在尚未构建的数据中心中展示的强大功能。”

此外，Oracle的政策是锁定客户，“这些努力对互操作性有很高的要求，” Enderle说。而且，亚马逊，谷歌，微软和IBM“拥有可以做[相同的事情，但是]它们包装得不好的服务”。

不过，如果成功执行，他承认Oracle的计划将“极大地促进”其在目标领域的现有努力。

原文题目：Oracle to Leverage AI, Machine Learning in Autonomous Cloud Platform

原文：Oracle on Tuesday demonstrated artificial intelligence and machine learning advances in the Oracle Cloud Platform at Oracle CloudWorld in New York.

Autonomous capabilities for application development, mobile and bots, integration, analytics, security and system management are scheduled for availability in the first half of this year.

Oracle announced a slew of new Platform as a Service features and capabilities, including the following:

Mobile and bots --

• Self-learning chatbots that observe interaction patterns and preferences to automate frequently performed end-user actions, freeing up time for higher-productivity tasks;
• Unsupervised smart bots that use machine learning to learn from user conversations, enabling fluid, contextual conversations; and
• Automated caching of API calls to the nearest data center in real time for lowest-latency responses based on the end user's location.

Security and management --

• Machine learning-driven user and entity behavior analytics to isolate and eliminate suspicious and malicious users automatically;
• Preventative controls to intercept data leaks across both structured and unstructured data repositories; and

· A unified data repository across log, performance, user experience, and configuration data with applied AI and ML to automate setting and managing performance and security monitoring metadata.

The Oracle Digital Assistant

Oracle also demonstrated an Oracle Digital Assistant, which will provide centralized communications across CRM, ERP, HCM, custom applications and business intelligence data.

The assistant will use AI to correlate data and automate user behavior.

Among its capabilities:

· Integration with speech-based devices and software like Amazon Echo (Alexa), Apple Siri, Google Home and Speech, Harman Kardon (Cortana) and Microsoft Cortana;

· Deep neural net-based machine learning algorithms to process messages from voice-based devices to understand end-user input and take action;

· Intelligent routing to Oracle Bots with the knowledge to process end users input; and

· Deep insights into user behavior, context, preferences and routines the Assistant uses to self-learn, in order to recommend and automate across all data sets on behalf of the user.

The Assistant's cross-application capability "is the differentiator," said Holger Mueller, principal analyst at Constellation Research.

However, "I'm skeptical," he told the E-Commerce Times, "as no neural network has been announced by Oracle."

The Assistant "would be compelling if it actually works," said Rob Enderle, principal analyst at the Enderle Group.

That said, "the major hardware players in the space already have their own back ends," he told the E-Commerce Times.

Amazon, Google, Microsoft and IBM "haven't exactly been Oracle's greatest fans over the years," Enderle pointed out. "I think the promise is massively ahead of reality here."

The Autonomous Database

Oracle also announced upcoming Autonomous Database services. The Autonomous Database, which uses advanced AI and ML, was first announced in October.

Multiple Autonomous Database services, each tuned to a specific workload, will be released this year:

· Oracle Autonomous Data Warehouse Cloud Service for analytics;

· Oracle Autonomous Database OLTP for transactional and mixed workloads; and

· Oracle Autonomous NoSQL Database for fast massive-scale reads and writes.

Leg Up or Tall Story?

"An autonomous Oracle is truly ahead of everybody," observed Constellation Research's Mueller. "This announcement now fleshes out the autonomous database, which had gaps in security and the risk of old code."

However, overall, Oracle's announcement "reads like a whopper," Enderle said."Folks forget that Oracle is famous for making claims they have capabilities long before they actually have them."

Delivering on its autonomous cloud promise "would require massive capabilities Oracle has yet to demonstrate in data centers that haven't been built yet," he added.

Further, Oracle's policy is to lock in customers, and "these efforts have high requirements for interoperability," Enderle said. Also, Amazon, Google, Microsoft and IBM "have services that can do [the same thing but] they just aren't packaged as nicely."

Still, if successfully executed, he acknowledged that Oracle's plans "would be a massive boost" to its existing efforts in the targeted areas.

原文作者：Richard Adhikari

原文链接：https://www.technewsworld.com/story/85134.html

翻译人：田冠宇