We Have Met the Enemy...and He Is Us

We Have Met the Enemy…and He Is Us

Barbee Davis, MA, PHR, PMP Omaha, Nebraska, U.S.
CARToonIST WAlT KEllEy, who inked the long-running comic strip Pogo, is famous for the quote, “We have met the enemy…and he is us.” Nowhere is this sentiment more accurate than when describing a software project man- ager who is new to the software development process. Here’s how to avoid having “the enemy” be you:
• As a project manager, you expect your team members to estimate the amount of time it will take them to complete a specific task. It is detri- mental to the schedule if they go too far over the budgeted time. One of your tasks on the project is to hold meetings to drive team communica- tion. You need to demonstrate the ability to estimate and deliver the meet- ings as meticulously as you expect your developers to estimate and deliver their code.
When your meetings run long, you are stealing the precious program- ming time developers count on to meet your project schedule deadlines.
• If your project team spoke a foreign language, you would take some les- sons and get a translator. Your developers do not speak your language. Buy a book, take a class, make Google your friend, and find a developer who has the gift to explain complex things in a simple way. You cannot bluff your way through this project without learning some of the concepts, terms, and challenges your team faces.
• Regardless of the perfect methodology you used to build toasters and cars, develop pharmaceuticals, or even construct skyscrapers, it won’t work here. Let the trusted members of your team explain about agile method- ologies. They aren’t new or risky. But they are your best chance at having a working product at the end of your project.
70 97 Things Every Project Manager Should Know

           • Developers are craftsmen and artists. They work differently than accoun- tants, attorneys, or bank tellers. When they are meeting in pairs and talk- ing animatedly, they are actually working. When they are bouncing a ball against a wall or doodling on a whiteboard, they may be crafting a solu- tion to an architecture problem that can’t be solved by staring at a com- puter screen. Give them space.

• Your team will work odd hours. We’ve all seen the cashier at the local food emporium switch with her replacement: she opens the register and exchanges the money drawer, and the new cashier is up and running. A programmer can’t switch places with a cohort and just pick up where his teammate left off. When your team member is feverishly at work, leave him or her alone. Researchers say it may take an hour or more for the person to regain productivity if interrupted.
• It is unnecessary to have every person program in exactly the same pro- gramming language. Some endeavors are better approached with newer languages that require fewer lines of code to write, test, store on your serv- ers, and maintain. Don’t refuse to let your developers use the best tool for the job.
Open your mind to this new world of software development, and you can be a support for your software development team, not the enemy.

深度学习是机器学习的一个子领域,它基于人工神经网络的研究,特别是利用多层次的神经网络来进行学习和模式识别。深度学习模型能够学习数据的高层次特征,这些特征对于图像和语音识别、自然语言处理、医学图像分析等应用至关重要。以下是深度学习的一些关键概念和组成部分: 1. **神经网络(Neural Networks)**:深度学习的基础是人工神经网络,它是由多个层组成的网络结构,包括输入层、隐藏层和输出层。每个层由多个神经元组成,神经元之间通过权重连接。 2. **前馈神经网络(Feedforward Neural Networks)**:这是最常见的神经网络类型,信息从输入层流向隐藏层,最终到达输出层。 3. **卷积神经网络(Convolutional Neural Networks, CNNs)**:这种网络特别适合处理具有网格结构的数据,如图像。它们使用卷积层来提取图像的特征。 4. **循环神经网络(Recurrent Neural Networks, RNNs)**:这种网络能够处理序列数据,如时间序列或自然语言,因为它们具有记忆功能,能够捕捉数据中的时间依赖性。 5. **长短期记忆网络(Long Short-Term Memory, LSTM)**:LSTM 是一种特殊的 RNN,它能够学习长期依赖关系,非常适合复杂的序列预测任务。 6. **生成对抗网络(Generative Adversarial Networks, GANs)**:由两个网络组成,一个生成器和一个判别器,它们相互竞争,生成器生成数据,判别器评估数据的真实性。 7. **深度学习框架**:如 TensorFlow、Keras、PyTorch 等,这些框架提供了构建、训练和部署深度学习模型的工具和库。 8. **激活函数(Activation Functions)**:如 ReLU、Sigmoid、Tanh 等,它们在神经网络中用于添加非线性,使得网络能够学习复杂的函数。 9. **损失函数(Loss Functions)**:用于评估模型的预测与真实值之间的差异,常见的损失函数包括均方误差(MSE)、交叉熵(Cross-Entropy)等。 10. **优化算法(Optimization Algorithms)**:如梯度下降(Gradient Descent)、随机梯度下降(SGD)、Adam 等,用于更新网络权重,以最小化损失函数。 11. **正则化(Regularization)**:技术如 Dropout、L1/L2 正则化等,用于防止模型过拟合。 12. **迁移学习(Transfer Learning)**:利用在一个任务上训练好的模型来提高另一个相关任务的性能。 深度学习在许多领域都取得了显著的成就,但它也面临着一些挑战,如对大量数据的依赖、模型的解释性差、计算资源消耗大等。研究人员正在不断探索新的方法来解决这些问题。
评论 1
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值