Skip to content
center-gradient-cover-bg
right-gradient-cover-bg
background gradient desk
Blog

FPT.AI – A comprehensive Artificial Intelligence platform

November 19, 2019

Share with:

Table of Contents

Each industrial revolution is driven by technology development, from stream engine to computer and those technologies have been driving human and the whole industry to a new era. Entering the fourth industrial revolution, it is believed that Artificial Intelligence (AI), Blockchain, and Internet of Things (IoT) will be the key enabling technology with AI is the de-factor.

According to Goldman Sachs’ report, AI is the technology that is receiving most investment from both corporate and venture capitalists. And most companies have been choosing it as the strategic investment.

In FPT’s perspective, AI also identified as the most critical technology area, which FPT needs to put investments comprehensively from building human capital, identifying applied areas and build up applications and most importantly to build up an advanced platform that would help us become a globally competitive technology company. The platform that FPT is developing named FPT.AI has been providing key capabilities such as Natural Language Understanding, Speech Recognition, Computer Vision, and the Knowledge Base. With the ambitious objective, FPT.AI aims to become a platform that can assist human in automating decision-making processes.

FPT.AI provides the foundation for machines to learn, abstract and discover knowledge so that organization will be able to bring their documents such as product definition and policies, user manuals, guidelines and mixed with the specific domain knowledge to create a network of concepts and their relations. From this graph, the platform will provide capability for question and answer on the domains and knowledge that the platform has learned. The knowledge base module will act as a brain and provides information to Automated Agents for communicating with human.

Knowledge Graph

Automated Agents

FPT.AI provides capabilities for organizations to build Automated Agents that can help machines to communicate with human in natural ways which is enabled by three modules: Natural Language Understanding, Speech Recognition and Synthesis and especially Dialog Management.

  • Natural Language Understanding has incorporated state of the art techniques to ensure the best understanding of human intentions and information provided. The module supports three languages: Vietnamese, English and Japanese.
  • The largest obstacle to creating good Chatbot is how to keep your conversation be natural and be flexible for supporting multiple chat scenarios in one bot. The new version of Dialog Management will provide the flexible yet easy to configure Dialog Management Framework to bot developers which following the advanced Information State approach. In which, multiple scenarios will be supported, and users can chat and switch context at their need.
  • While Speech Synthesis has been developing for the last five years, Speech Recognition is a new module developed and integrated into FPT.AI which fulfills the key building blocks for building an Automated Agent. FPT.AI Speech components focus only on Vietnamese where users are able to match with state-of-the-art

The key applications for Automated Agents would be Chatbot and Automated Call Center. While chatbot is a typical case, Automated Call Center will be more complicated but would bring great benefit to organizations and customers. In later 2018, there would be two versions of Automated Call Center released. The Voice Gateway which aims to replace Interactive Voice Response, instead of listening to a tree of guidelines from the phone, caller would just need to speak out the topics that they would need support. Then basic Question and Answer capability will also be released based on the technologies above and the Knowledge base components.

Vision

Screenshot from FPT.AI, module Dialog Management

Vision will also be a new component for FPT.AI that will focus on enabling Cognitive RPA (Robotics Process Automation) for organizations. Image Recognition, form recognition, and OCR (Optical Character Recognition) are the main capabilities. The module aims to capture and digitalize all basic papers such as National ID Card, Driver licenses, Invoices and moving toward generic forms.

Throughout two years offering services to the market, FPT.AI has achieved encouraging results in international and domestics markets. Where multiple use cases have been implemented to support users accounts, billing, and sales. More than 1,000 chatbots created on the FPT.AI platform have been enhancing the user experience in many areas such as telecommunications, e-commerce, banking, or public administration, as well as reducing operating fees for businesses. A typical example includes a bot designated by FPT.AI aiming at supporting information related to user accounts, new products, and services for one of the leading telecommunication companies in Singapore. The Chatbot has received and processed over 40,000 user requests in one month or over 4,000 requests per day in peak time.

In addition, new capabilities have been realized to bring up new innovative business models to our partners. FPT.AI’s automated call center has implemented 750,000 calls per month, 15,000 calls at once time. With the help of AI, machines now can perform simple tasks for human, so that we could have more resources on higher-skilled jobs with higher values.

More than 7.5 million requests have been made to the platform and more than 2,000 hours of voices have been processed and generated by the platform that consumed by FPT.AI eco partners is the most obvious demonstration of the benefits and efficiency of FPT.AI.

1.2/5 - (5 votes)
Related Posts
image post
Blog
FPT AI Readerを使用して画像から情報を抽出する方法

FPT AI Readerは、OCR技術を使用して、既存の文書テンプレート(個人書類、請求書)や任意のカスタム文書形式(契約書、証明書など)から画像の内容を正確に認識および抽出するアプリケーションであり、迅速かつ便利に文書をデジタル化します。この記事では、FPT.AIがFPT AI Readerを使用して画像から情報を抽出する方法を説明します。一緒に見てみましょう! FPT AI Readerのアカウント登録と言語選択 FPT.AIの専門家によって研究開発されたFPT AI Readerアプリケーションは、人工知能(AI)とコンピュータビジョン、光学文字認識(OCR)、インテリジェントドキュメント処理(IDP)技術を使用し、画像処理技術と自然言語処理(NLP)を組み合わせて、ユーザーが数秒で正確に文書をデジタル化し、高いセキュリティと簡単な保存を実現します。 FPT.AI Readerアプリケーションを使用するには、FPT IDアカウントが必要です。アカウントをお持ちでない場合は、こちらで登録してください: https://id.fptcloud.com/.その後、こちらにアクセスしてください: https://reader.fpt.ai/ 素晴らしい体験を始めましょう! FPT AI Readerアプリケーションは、英語とベトナム語の2つの言語をサポートしています。画面右上のアイコンをクリックして使用する言語を選択してください。   FPT AI Readerの既存の文書ライブラリを使用して情報を抽出する方法 次に、FPT AI Readerのライブラリに既にある画像文書の種類から情報を抽出するには、次の手順を実行します。 文書テンプレートの選択 まず、画面右上のライブラリをクリックすると、FPT AI Readerのインターフェースに、アプリケーションに統合された文書テンプレートが表示され、抽出精度は最大98%です。 現在、FPT AI Readerは、個人書類や異なる業界分野の文書テンプレートから情報をデジタル化することをサポートしており、例えば、身分証明書、運転免許証、パスポート、出生証明書、請求書、保険証書などがあります。 使用を開始するには、データを抽出する文書の種類をクリックし、プロジェクトに名前を付けて、作成をクリックする必要があります。この記事では、運転免許証の文書を例にします。 この時点で、画面のインターフェースには、運転免許証の情報を抽出する際に使用する重要な情報フィールドと関連データ(名前、ステータス、更新時間、作成時間、モデルの使用、アクション)が表示されます。 データのアップロード 続行するには、画面右上のアップロードを選択し、運転免許証の画像をシステムにアップロードします。注意: 1ファイルのアップロード容量は5MBを超えないようにし、jpg、png、pdfなどの画像形式を受け入れます。 画面の左下隅にあるファイル選択 (1) をクリックして、複数の画像ファイルを同時にアップロードできます。アップロードする運転免許証の画像ファイルを選択した後、アップロード (2) をクリックしてこの手順を完了します。 運転免許証の画像がシステムに正常にアップロードされると、FPT AI Readerは数秒で画像を自動的に処理し、画像のステータスを処理済みに変更します。この時点で、各行をクリックして運転免許証の画像と抽出結果を表示できます。 (注意: 画像はあくまで参考用であり、実際の使用価値はありません) 他の正常にアップロードされた運転免許証の画像でも同様に操作します。OCR技術を使用して、FPT AI Readerは、画像がぼやけている、文字が不鮮明である、異なる色の領域がある、文字が重なっている、画像が整列していないなどの条件でも高精度で認識できます(例: 下の画像の「ngày/date 05 tháng/month … Continued

Get ahead with AI-powered technology updates!

Subscribe now to our newsletter for exclusive insights, expert analysis, and cutting-edge developments delivered straight to your inbox!