Cuộc sống Việt Nam / ベトナム生活|Hồ Chí Minh / ホーチミン
Quảng cáo này xuất hiện trên các Blog không cập nhật bài viết trên 1 tháng
Nếu bạn cập nhật bài viết mới thì quảng cáo này sẽ mất đi
上記の広告は1ヶ月以上記事の更新がないブログに表示されます。
新しい記事を書くことでこちらの広告は消えます。
こんばんは!
ミンです、
今、AIは人気があることをなります。
AIから健康管理、自動車、マーケティングのアプリケーションが多いです。
AI時代の最も重要なことの1つは、データです。
データは信号、個人情報、写真です。
長い時間の前に、写真は視覚的にしか理解できないです。
でも、AI時代に、写真は豊富な知識のリソースです。
これは稲のしゃしんです。でも、それは特殊です。ウイルスから稲は葉がねじれています。
農夫だけがこれが知ります。
しかし、AIはこれを理解できるようになりましたが、方法の理解がすすこし難しいです。
今、毎夜、私は稲の写真から疾患を研究者しています。私はそれが成功することを願っています。

こんばんは!
今日、私は浜松のIT大学生にVideo Call MTGで会いました。
これらの大学生は才能があるとダイナミックです。
二年生大学生にもかかわらず、彼らはAIに野心的です。
今日、学生は私の1つの良い質問をします:十分なデータがないときにAIを使用する方法はなんですか?
これはても面白い質問です、AIの大きな問題です。
経験で、私は学生に2つのキーワードを送ります:Sparse DataとData Augmentation.
Sparse Dataは稀薄データです、未知のデータと同じです。稀薄データは多きのデータはZeroです。それは重要ではないです。
Data Augmentationはデータベースで数学でよって既存のデータを新しいデータに変換しますので、データベースがますます大きくなります。単純演算子はプラス、マイナス、ローテーションです。
どちらもAIで非常に難しい問題です。したがって、学生がこれを理解したいときはとても良いことです。

こんばんは!
ミンです、
今日、私はモントリオールAから良い技術のニュースを読みました。Kalimdorという名前の新しいライブラリのMachine Learningについて紹介しました。
今、Artificial Intelligenceは人気がありつつあります。チャットボット、自動車などのディープ・ラーニングに基づくのアプリケーションはビジネスに適用されました。AIプログラミングでTensorflowとPytorchは有名なライブラリーです。それは複雑な製品と大規模なビジネスのためにDeep Learningを良くサポートします。
その間に、中小企業がAIを統合し始めますがTensorflowは必要以上にあります。従って、実行可能なライブラリーを事欠きます。Pythonプログラミング言語で、ScikitライブラリーはMachine Learningのためにありますが、Javascript言語はまだありません。Javascriptは、Websiteアプリのために便利な言語の1つですので方法はJavaScriptを使用してAIをWebサイトに統合するは高い需要です。
それはKalimdorの哲学です、Javascript言語を使用してMachine Learningアルゴリズムを完全サポートします。
KalimdorはWorld of Warcraftエリアの名前です。そこは多きの木があります。Machine LearningでDecision Treeアルゴリズムと同じ名前です。
Machine Learningアルゴリズムをサポート、良かったです。特に、Javascript言語を使用することができます。これはウェブサイトやルアプリを開発する上で大きな利点です。
さらに、Typescript、Javascriptフレームワークは一度にウェブサイトとアプリを開発しするは、Kalimdorをスポットしますので、フロントエンド、バックエンド、Machine LearningのためにJavascript言語だけを使用します、とても便利です。
今、SpotifyはAIを使用した歌の推薦KalimdorとTypescriptを使用しました。べとなむで音楽ウェブサイト「Zing mp3」も使用すると思います。
この図書館を深く勉強します。Javascriptは処理のために1つスレッドだけがありますので、Kalimdorの問題は処理時間を最適化する方法だと思います。
昨日、シーポイントエンジニアはウェブサイト開発の技術「Redux」を学ぶために30分の会議があります。
将来的には、チームに知識を共有するために、知識を共有するために、もっと30分会議を整理したい。Kalimdorは次のクラスの話題です。

Good morning,
the first day of the lunar new year is such a good day to start to learn something new.
I started to reading Deep learning book written by Ian Goodfellow and Yoshua Bengio who are the famous research scientist in Artificial Intelligent. Goodfellow is one of the young idols in A.I who proposed GAN model which is capable to classify real or fake object under noise effect.
This book introduced from basic to advance the requirements, theories, and technicals of Deep learning. In my opinion, it's a good approach for a new-comer. But it would be better if you had a good knowledge of linear algebra, calculus, probability and base knowledge of machine learning first.
It's a long story for this book coming to my home. I ordered from Amazon France, my Ph.D. friend picked up and send to my classmate of French university to hold while I was being worked in Thailand, and now it stays here with me, thank all my friends.
Now, if you can not have a book, you can get all chapters in HTML forrmat via this link, now it's open and free for all people :
http://www.deeplearningbook.org/

New year, a new target, a new challenge, I hope everything will be fine in this years... the best luck for me!
こんばんは!
Good evening!
ミンです。
I'm back
昨日私がITチー厶と会議ありました。プロジェクトの品質を改善するためのAIツールについて説明しました。私はこの分野に十分な知識がありませんでも私は問題を知れます。それはNatural Language Processing(NLP)です。
Yesterday I had a short meeting with IT team, we discussed AI tool to improve project quality. Honestly, I have not enough knowledge in this specific branch, but I could imagine generally what's a problem we're dealing, an engine for natural language processing.
この分野の私の限界ので、API、時間は短縮しましたためのFrameworkを選び出した。Wit.aiはよいAPIです、そのAPIはのですからMessengerを支えま。
Due to my limit in this field, I decided to look for API or framework to cut down prototype time. Selecting a tool is based on not only its core matching with our problem but also a pricing acceptable to the business model, it's a very important thing that I learned from my senior. After searching and studying I chose Wit.ai, natural language interface (API) which is a part of Facebook. Hence, it definitely supported Facebook Messenger.
むずかしですね!私は働いてから勉強しました。それは多きのことをしますがよいです。前に進み続けなさい!
It's not easy to get a "quick and run" result to apply, I'm still studying and practicing after working hours. Many things to do but it's good for me and work in future, keep moving forward !!!
https://wit.ai/
Good evening everyone,
Today I found one interesting startup in France, it's Deepomatic.
Deepomatic is an ambitious start-up in Paris region, their target is aiming at building an AI-powered model to solve image-related problems.

Deepomatic had achieved the 2nd price at Websummit 2017 in Lisbon, such a great company with an innovated product.
I really excited to understand the product of this start-up, so I took a small challenge via the demo on their website.
I give an input image including a lot of pedestrians, this image belongs to a typical dataset using for pedestrian detection problem. After few seconds, a result will be returned with attached details

In that image, there are 12 people, but the results was almost correct with 13 people. Moreover, it told us there are also 2 bicycles. Amazing !!! It's multi-object detection. The quality detection is very good, there is no false negative and only 1 false positive.
You can have a funny test on this website https://developers.deepomatic.com/demos
Two years ago, I already worked in pedestrians detection project but the result was not good like this. I fell like being out of date now.
Technology changed so fast. Thanks to "complete Deep learning", AI is becoming more and more popular, many frameworks released with useful API, so is it a time to turn our business into AI?
こんばんは!
ミンです。
昨日FAIR ( Facebook AI Research) はSLACデータセットを放ちました。
行動認識とビデオ分類に関連するデータセットであります。

そのデータセットは520,000以上のトリムされていないビデオと1.75Mのクリップ注釈で構成します。
ここはテストの例があります。
綱引き、バレーボール、棒高跳びはテストに行動です。
「Positive」は正しい行動です、「Negative」は間違った行動、またわ偽の行動です。

私はすぐにそれをテストします。
S |
M |
T |
W |
T |
F |
S |
|
|
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
9 |
10 |
11 |
12 |
13 |
14 |
15 |
16 |
17 |
18 |
19 |
20 |
21 |
22 |
23 |
24 |
25 |
26 |
27 |
28 |
29 |
30 |
|
|
|
読者登録
Chúng tôi sẽ gửi những bài viết mới nhất đến địa chỉ email đăng ký. Xóa Tại đây
Số lượng người đọc hiện tại là 1 người
プロフィール
ミン
From ancient and peaceful Troyes to modern and entertaining Saigon
|