View

1:Note: The Android framework is responsible for measuring, laying out and
 * drawing views. You should not call methods that perform these actions on
 * views yourself unless you are actually implementing a
 * {@link android.view.ViewGroup}.

2:To implement a custom view, you will usually begin by providing overrides for
 * some of the standard methods that the framework calls on all views. You do
 * not need to override all of these methods. In fact, you can start by just
 * overriding {@link #onDraw(android.graphics.Canvas)}.

3:<p>
 * The first pair is known as <em>measured width</em> and
 * <em>measured height</em>. These dimensions define how big a view wants to be
 * within its parent (see <a href="#Layout">Layout</a> for more details.) The
 * measured dimensions can be obtained by calling {@link #getMeasuredWidth()}
 * and {@link #getMeasuredHeight()}.
 * </p>

4:To measure its dimensions, a view takes into account its padding. The padding
 * is expressed in pixels for the left, top, right and bottom parts of the view.
 * Padding can be used to offset the content of the view by a specific amount of
 * pixels. For instance, a left padding of 2 will push the view's content by
 * 2 pixels to the right of the left edge. Padding can be set using the
 * {@link #setPadding(int, int, int, int)} or {@link #setPaddingRelative(int, int, int, int)}
 * method and queried by calling {@link #getPaddingLeft()}, {@link #getPaddingTop()},
 * {@link #getPaddingRight()}, {@link #getPaddingBottom()}, {@link #getPaddingStart()},
 * {@link #getPaddingEnd()}.

5:Layout is a two pass process: a measure pass and a layout pass. The measuring
 * pass is implemented in {@link #measure(int, int)} and is a top-down traversal
 * of the view tree. Each view pushes dimension specifications down the tree
 * during the recursion. At the end of the measure pass, every view has stored
 * its measurements. The second pass happens in
 * {@link #layout(int,int,int,int)} and is also top-down. During
 * this pass each parent is responsible for positioning all of its children
 * using the sizes computed in the measure pass.

6:When a view's measure() method returns, its {@link #getMeasuredWidth()} and
 * {@link #getMeasuredHeight()} values must be set, along with those for all of
 * that view's descendants. A view's measured width and measured height values
 * must respect the constraints imposed by the view's parents. This guarantees
 * that at the end of the measure pass, all parents accept all of their
 * children's measurements. A parent view may call measure() more than once on
 * its children. For example, the parent may measure each child once with
 * unspecified dimensions to find out how big they want to be, then call
 * measure() on them again with actual numbers if the sum of all the children's
 * unconstrained sizes is too big or too small.

7:The measure pass uses two classes to communicate dimensions. The
 * {@link MeasureSpec} class is used by views to tell their parents how they
 * want to be measured and positioned. The base LayoutParams class just
 * describes how big the view wants to be for both width and height. For each
 * dimension, it can specify one of:
 * <ul>
 * <li> an exact number
 * <li>MATCH_PARENT, which means the view wants to be as big as its parent
 * (minus padding)
 * <li> WRAP_CONTENT, which means that the view wants to be just big enough to
 * enclose its content (plus padding).
 * </ul>
 * There are subclasses of LayoutParams for different subclasses of ViewGroup.
 * For example, AbsoluteLayout has its own subclass of LayoutParams which adds
 * an X and Y value.

8:MeasureSpecs are used to push requirements down the tree from parent to
 * child. A MeasureSpec can be in one of three modes:
 * <ul>
 * <li>UNSPECIFIED: This is used by a parent to determine the desired dimension
 * of a child view. For example, a LinearLayout may call measure() on its child
 * with the height set to UNSPECIFIED and a width of EXACTLY 240 to find out how
 * tall the child view wants to be given a width of 240 pixels.
 * <li>EXACTLY: This is used by the parent to impose an exact size on the
 * child. The child must use this size, and guarantee that all of its
 * descendants will fit within this size.
 * <li>AT_MOST: This is used by the parent to impose a maximum size on the
 * child. The child must guarantee that it and all of its descendants will fit
 * within this size.

9:To initiate a layout, call {@link #requestLayout}. This method is typically
 * called by a view on itself when it believes that is can no longer fit within
 * its current bounds.
  view显示区域发生变化的时候,就需要调用requestLayout这个方法回调onLayout方法。

10:
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
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