azure机器学习

by Déborah Mesquita

由DéborahMesquita

我如何打造一款赢得2016 Azure机器学习奖的游戏 (How I built a game that won the 2016 Azure Machine Learning Award)

Every year, Microsoft hosts the Imagine Cup. Young developers often call it the “Olympics of Technology” and consider it one of the top competitions related to software design. As a result, it attracts lots of young participants from around the world, who collaborate to solve some of the world’s toughest challenges.

每年，微软都会举办“ 想象杯” 。 年轻的开发人员通常将其称为“ 技术奥林匹克 ”，并将其视为与软件设计相关的顶级竞赛之一。 结果，它吸引了来自世界各地的许多年轻参与者，他们共同解决了世界上最严峻的挑战。

In 2016, they hosted the Hello Cloud Machine Learning Award, where winners were selected from all entries based on quality, creativity and the effectiveness of their use of Azure Machine Learning Studio.

他们在2016年举办了Hello Cloud机器学习奖 ，根据质量，创造力和使用Azure机器学习Studio的效率从所有条目中选出获奖者。

I jumped into the competition to learn more about machine learning and ended up as one of the winners of the challenge. What got my interested in the first place was the focus of the competition: to build creative and inventive systems using machine learning.

我参加了比赛，以学习更多有关机器学习的知识，并最终成为挑战赛的获胜者之一。 首先让我感兴趣的是比赛的焦点： 使用机器学习构建具有创造力和创造力的系统 。

When a user stops using a good or service (in this case, a game), we call that “churn.” Based on past player history, or data for similar players over time, we can create a machine learning model to predict when a player is most likely to quit.

当用户停止使用商品或服务(在这种情况下为游戏)时，我们称其为“搅动”。 基于过去的玩家历史记录或一段时间内类似玩家的数据，我们可以创建机器学习模型来预测最有可能退出的玩家。

In the first part of the competition we had to build, train, score, and evaluate a model to do just that in the Azure ML Studio. We then had take a basic game provided by them, connect it to the Azure ML service, and publish it to the web.

在竞赛的第一部分中，我们必须构建，训练，评分和评估模型，才能在Azure ML Studio中做到这一点。 然后，我们接受了他们提供的基本游戏，将其连接到Azure ML服务，然后将其发布到Web。

One of the key factors that makes a player abandon a game is its difficulty level. If it's too easy the game gets boring, and if it's too hard it demotivates the user to keep playing.

使玩家放弃游戏的关键因素之一是游戏的难度等级。 如果太容易了，那么游戏就会变得无聊，如果太难了，就会使用户无法继续玩下去。

I decided to use the rock-paper-scissors dynamics in the game. To earn new superpowers in the game, the players (math students) would have to solve some math equations (these were used like attacks in a fighting game).

我决定在游戏中使用剪刀石头布的动态效果。 为了在游戏中获得新的超能力，玩家(数学系学生)必须解决一些数学方程式(这些方程式被用作格斗游戏中的攻击)。

Based on the data for each player, we could adjust the difficulty of the math equations to keep them motivated to play the game. We could also identify what kind of equations were causing kids the more trouble (subtraction? multiplication?). This is an amazing opportunity to help teachers and everyone involved in the learning process.

根据每个玩家的数据，我们可以调整数学方程式的难度，以保持他们玩游戏的动力。 我们还可以确定哪种方程式给孩子带来了更多麻烦(减法，乘法？)。 这是一个极好的机会，可以帮助教师和参与学习过程的每个人。

原则1：专注于与众不同 (Principle 1: Focus on being different)

One thing I know for sure in competitions is this: focus on being different instead of just focusing on being better. We don't know the number of competitors for sure, but based on what we heard there in Seattle, the contest had almost 1,000 entries. That's a lot of games for the judges to evaluate. You have to do everything you can to stand out from a crowd that size.

我在比赛中肯定知道的一件事是： 专注于与众不同，而不仅仅是专注于变得更好 。 我们不知道确切的竞争对手数量，但是根据我们在西雅图那里听到的信息，竞赛有近1,000个参赛作品。 评委们需要评估很多游戏。 您必须尽一切可能在如此大的人群中脱颖而出。

I bet that when you read the word “different” your first thought was “Great, now I have to come up with something big and strange out of nowhere.” Don’t worry — that isn’t the case. Because here’s another thing I know for sure: to be different, you can just focus on being yourself. I know it sounds cheesy, but let's elaborate on that.

我敢打赌，当您读到“与众不同”一词时，您的第一个念头是：“好极了，现在我不得不冒昧地提出一些大而奇怪的事情。” 不用担心-事实并非如此。 因为这是我肯定知道的另一件事： 与众不同，您可以专注于做自己 。 我知道这听起来很俗气，但让我们对其进行详细说明。

“Be yourself; everyone else is already taken.” ― Oscar Wilde

“做你自己; 其他人都已经被抓走了。” ― 奥斯卡·王尔德

You are you, right? No one else in the world have had the experience you had, did everything you did or felt exactly everything you felt. That's it, you just have to use this to be different (and original). Now let's go back to what I did in the contest.

你是你吧？ 世界上没有其他人拥有您所拥有的经验，做过的所有事情或完全感受到了您的所有感觉。 就是这样，您只需使用它即可与众不同(原始)。 现在，让我们回到比赛中的表现。

I bought my first Wacom tablet in time for the competition, and honestly I was just looking for excuses to use it. I like to get adventurous in other areas, and I know that this is something that differentiates me. So I decided to work and change the assets of the game.

我及时买了第一台Wacom平板电脑来参加比赛，说实话，我只是在找借口使用它。 我喜欢在其他领域冒险，这让我与众不同。 因此，我决定工作并更改游戏的资产。

原则2：从您所知道的开始 (Principle 2: Start with what you know)

In the competition we first had to follow a tutorial. Only after that could we begin to create our own version of the game. This is a great way to design the workflow of our projects (and side projects): always find a way to make the start phase easy.

在比赛中，我们首先必须遵循教程。 只有在那之后，我们才能开始创建我们自己的游戏版本。 这是设计我们的项目(和副项目)工作流程的好方法： 总是找到一种使开始阶段变得容易的方法 。

I first heard this advice in the book Think Like a Programmer. It’s true for programming, but it’s also true for a bunch of other aspects of our lives.

我首先在《 像程序员一样思考》一书中听到了这个建议。 对于编程来说确实如此，但是对于我们生活中的其他方面也是如此。

Once you have divided the problem up into pieces, for example, go ahead and complete any pieces you already know how to code. Having a working partial solution may speak ideas about the rest of the problem. Also, as you may have noticed, a common theme in problem solving is making useful progress to build confidence that you will ultimately complete the task. By starting with what you know, you build confidence and momentum toward the goal. ― V. Anton Spraul, Think Like a Programmer

例如，将问题分成几部分后，请继续并完成您已经知道如何编码的所有部分。 有一个可行的部分解决方案可能会说出有关其余问题的想法。 另外，您可能已经注意到， 解决问题的一个共同主题是取得有益的进展，以树立最终完成任务的信心 。 从您所知道的开始， 您就可以朝着目标建立信心和动力 。 ― V 安东·斯普劳(Anton Spraul)， 像程序员一样思考

Let's be honest: programming is hard. During the competition I had some moments of frustration, but things like seeing my first predictive model running and seeing the pieces of the game start to work together got me going. Make sure you can start to see progress in a project, right from the beginning.

说实话：编程很难。 在比赛中，我有些沮丧，但看到我的第一个预测模型运行并看到游戏各部分开始协同工作的事情使我前进。 确保您从一开始就可以开始查看项目的进度。

原则3：适应 (Principle 3: Adapt)

This was the key factor for winning the competition, because without it, my entry wouldn't even have been submitted. Time is a limited source. Everybody knows this, but it’s something that we always have to keep reminding ourselves of — especially us programmers.

这是赢得比赛的关键因素，因为没有它，我的参赛作品甚至都不会提交。 时间是有限的来源 。 每个人都知道这一点，但这是我们始终需要提醒自己的事情-尤其是我们程序员。

If you take a look at my Game Design Document above, you can see that my first idea for the game had a lot of features. For example, we had levels for the players, items they would be able to collect, healing effects, and so on. As the deadline approached, I realized that I didn't have time to execute all of those ideas. So I had to think: what is the one thing that I should have in the game to make sure that it would accomplish my goal? The answer was math equations and the rock-paper-scissors dynamics, and that was what I implemented.

如果您看一下上面的《游戏设计文档》，您会发现我对游戏的第一个想法具有很多功能。 例如，我们为玩家设置了等级，他们可以收集的物品，恢复效果等等。 随着最后期限的临近，我意识到我没有时间执行所有这些想法。 所以我不得不思考：为了确保游戏能够实现我的目标，我在游戏中应该做的一件事是什么？ 答案是数学方程式和剪刀布动力学，这就是我所实现的。

It's not easy to give instructions about how to adapt, because every situation is a different. But you should know that you will have to make choices along the way. Your main focus should be to finish the thing on time, so you can indeed compete in the contest.

提供有关如何适应的指导并不容易，因为每种情况都不尽相同。 但是您应该知道，您将不得不在此过程中做出选择 。 您的主要重点应该是按时完成事情，这样您才能真正参加比赛。

Well, as you may know by now, my project was one of the two winners of the challenge (yay!). I won a trip to the Imagine Cup World Finals and had mentoring sessions with members of the Microsoft Data Platform product.

好吧，正如您现在可能知道的那样，我的项目是挑战的两个赢家之一(是！)。 我赢得了Imagine Cup世界总决赛的旅程，并与Microsoft数据平台产品的成员进行了辅导。

With this project I finally found my main career goal: to design Machine Learning systems that let humans do things they care about.

通过这个项目，我终于找到了我的主要职业目标： 设计机器学习系统，让人们去做他们关心的事情 。

翻译自: https://www.freecodecamp.org/news/how-i-won-the-azure-machine-learning-award-418ff35c6e4d/

azure机器学习