React Tutorial

React Tutorial


React Tutorial for Beginners - Hooks API Reference

Hooks API Reference

Hooks are a new addition in React 16.8. They let you use state and other React features without writing a class.

This page describes the APIs for the built-in Hooks in React.

If you’re new to Hooks, you might want to check out the overview first. You may also find useful information in the frequently asked questions section.

  • Basic Hooks
    • useState
    • useEffect
    • useContext
  • Additional Hooks
    • useReducer
    • useCallback
    • useMemo
    • useRef
    • useImperativeHandle
    • useLayoutEffect
    • useDebugValue

Basic Hooks


const [state, setState] = useState(initialState);

Returns a stateful value, and a function to update it.

During the initial render, the returned state (state) is the same as the value passed as the first argument (initialState).

The setState function is used to update the state. It accepts a new state value and enqueues a re-render of the component.


During subsequent re-renders, the first value returned by useState will always be the most recent state after applying updates.


React guarantees that setState function identity is stable and won’t change on re-renders. This is why it’s safe to omit from the useEffect or useCallback dependency list.

Functional updates

If the new state is computed using the previous state, you can pass a function to setState. The function will receive the previous value, and return an updated value. Here’s an example of a counter component that uses both forms of setState:

function Counter({initialCount}) {
  const [count, setCount] = useState(initialCount);
  return (
      Count: {count}
      <button onClick={() => setCount(initialCount)}>Reset</button>
      <button onClick={() => setCount(prevCount => prevCount - 1)}>-</button>
      <button onClick={() => setCount(prevCount => prevCount + 1)}>+</button>

The ”+” and ”-” buttons use the functional form, because the updated value is based on the previous value. But the “Reset” button uses the normal form, because it always sets the count back to the initial value.

If your update function returns the exact same value as the current state, the subsequent rerender will be skipped completely.


Unlike the setState method found in class components, useState does not automatically merge update objects. You can replicate this behavior by combining the function updater form with object spread syntax:

setState(prevState => {
  // Object.assign would also work
  return {...prevState, ...updatedValues};

Another option is useReducer, which is more suited for managing state objects that contain multiple sub-values.

Lazy initial state

The initialState argument is the state used during the initial render. In subsequent renders, it is disregarded. If the initial state is the result of an expensive computation, you may provide a function instead, which will be executed only on the initial render:

const [state, setState] = useState(() => {
  const initialState = someExpensiveComputation(props);
  return initialState;
Bailing out of a state update

If you update a State Hook to the same value as the current state, React will bail out without rendering the children or firing effects. (React uses the comparison algorithm.)

Note that React may still need to render that specific component again before bailing out. That shouldn’t be a concern because React won’t unnecessarily go “deeper” into the tree. If you’re doing expensive calculations while rendering, you can optimize them with useMemo.



Accepts a function that contains imperative, possibly effectful code.

Mutations, subscriptions, timers, logging, and other side effects are not allowed inside the main body of a function component (referred to as React’s render phase). Doing so will lead to confusing bugs and inconsistencies in the UI.

Instead, use useEffect. The function passed to useEffect will run after the render is committed to the screen. Think of effects as an escape hatch from React’s purely functional world into the imperative world.

By default, effects run after every completed render, but you can choose to fire them only when certain values have changed.

Cleaning up an effect

Often, effects create resources that need to be cleaned up before the component leaves the screen, such as a subscription or timer ID. To do this, the function passed to useEffect may return a clean-up function. For example, to create a subscription:

useEffect(() => {
  const subscription = props.source.subscribe();
  return () => {
    // Clean up the subscription

The clean-up function runs before the component is removed from the UI to prevent memory leaks. Additionally, if a component renders multiple times (as they typically do), the previous effect is cleaned up before executing the next effect. In our example, this means a new subscription is created on every update. To avoid firing an effect on every update, refer to the next section.

Timing of effects

Unlike componentDidMount and componentDidUpdate, the function passed to useEffect fires after layout and paint, during a deferred event. This makes it suitable for the many common side effects, like setting up subscriptions and event handlers, because most types of work shouldn’t block the browser from updating the screen.

However, not all effects can be deferred. For example, a DOM mutation that is visible to the user must fire synchronously before the next paint so that the user does not perceive a visual inconsistency. (The distinction is conceptually similar to passive versus active event listeners.) For these types of effects, React provides one additional Hook called useLayoutEffect. It has the same signature as useEffect, and only differs in when it is fired.

Although useEffect is deferred until after the browser has painted, it’s guaranteed to fire before any new renders. React will always flush a previous render’s effects before starting a new update.

Conditionally firing an effect

The default behavior for effects is to fire the effect after every completed render. That way an effect is always recreated if one of its dependencies changes.

However, this may be overkill in some cases, like the subscription example from the previous section. We don’t need to create a new subscription on every update, only if the source prop has changed.

To implement this, pass a second argument to useEffect that is the array of values that the effect depends on. Our updated example now looks like this:

  () => {
    const subscription = props.source.subscribe();
    return () => {

Now the subscription will only be recreated when props.source changes.


If you use this optimization, make sure the array includes all values from the component scope (such as props and state) that change over time and that are used by the effect. Otherwise, your code will reference stale values from previous renders. Learn more about how to deal with functions and what to do when the array values change too often.

If you want to run an effect and clean it up only once (on mount and unmount), you can pass an empty array ([]) as a second argument. This tells React that your effect doesn’t depend on any values from props or state, so it never needs to re-run. This isn’t handled as a special case — it follows directly from how the dependencies array always works.

If you pass an empty array ([]), the props and state inside the effect will always have their initial values. While passing [] as the second argument is closer to the familiar componentDidMount and componentWillUnmount mental model, there are usually better solutions to avoid re-running effects too often. Also, don’t forget that React defers running useEffect until after the browser has painted, so doing extra work is less of a problem.

We recommend using the exhaustive-depsrule as part of oureslint-plugin-react-hooks package. It warns when dependencies are specified incorrectly and suggests a fix.

The array of dependencies is not passed as arguments to the effect function. Conceptually, though, that’s what they represent: every value referenced inside the effect function should also appear in the dependencies array. In the future, a sufficiently advanced compiler could create this array automatically.


const value = useContext(MyContext);

Accepts a context object (the value returned from React.createContext) and returns the current context value for that context. The current context value is determined by the value prop of the nearest <MyContext.Provider> above the calling component in the tree.

When the nearest <MyContext.Provider> above the component updates, this Hook will trigger a rerender with the latest context value passed to that MyContext provider. Even if an ancestor uses React.memoorshouldComponentUpdate, a rerender will still happen starting at the component itself using useContext.

Don’t forget that the argument to useContext must be the context object itself:

  • Correct: useContext(MyContext)
  • Incorrect: useContext(MyContext.Consumer)
  • Incorrect: useContext(MyContext.Provider)

A component calling useContext will always re-render when the context value changes. If re-rendering the component is expensive, you can optimize it by using memoization.


If you’re familiar with the context API before Hooks, useContext(MyContext) is equivalent to static contextType = MyContext in a class, or to <MyContext.Consumer>.

useContext(MyContext) only lets you read the context and subscribe to its changes. You still need a <MyContext.Provider> above in the tree to provide the value for this context.

Putting it together with Context.Provider

const themes = {
  light: {
    foreground: "#000000",
    background: "#eeeeee"
  dark: {
    foreground: "#ffffff",
    background: "#222222"

const ThemeContext = React.createContext(themes.light);

function App() {
  return (
    <ThemeContext.Provider value={themes.dark}>
      <Toolbar />

function Toolbar(props) {
  return (
      <ThemedButton />

function ThemedButton() {
  const theme = useContext(ThemeContext);  return (    <button style={{ background: theme.background, color: theme.foreground }}>      I am styled by theme context!    </button>  );

This example is modified for hooks from a previous example in the Context Advanced Guide, where you can find more information about when and how to use Context.

Additional Hooks

The following Hooks are either variants of the basic ones from the previous section, or only needed for specific edge cases. Don’t stress about learning them up front.


const [state, dispatch] = useReducer(reducer, initialArg, init);

An alternative to useState. Accepts a reducer of type (state, action) => newState, and returns the current state paired with a dispatch method. (If you’re familiar with Redux, you already know how this works.)

useReducer is usually preferable to useState when you have complex state logic that involves multiple sub-values or when the next state depends on the previous one. useReducer also lets you optimize performance for components that trigger deep updates because you can pass dispatch down instead of callbacks.

Here’s the counter example from the useState section, rewritten to use a reducer:

const initialState = {count: 0};

function reducer(state, action) {
  switch (action.type) {
    case 'increment':
      return {count: state.count + 1};
    case 'decrement':
      return {count: state.count - 1};
      throw new Error();

function Counter() {
  const [state, dispatch] = useReducer(reducer, initialState);
  return (
      Count: {state.count}
      <button onClick={() => dispatch({type: 'decrement'})}>-</button>
      <button onClick={() => dispatch({type: 'increment'})}>+</button>


React guarantees that dispatch function identity is stable and won’t change on re-renders. This is why it’s safe to omit from the useEffect or useCallback dependency list.

Specifying the initial state

There are two different ways to initialize useReducer state. You may choose either one depending on the use case. The simplest way is to pass the initial state as a second argument:

  const [state, dispatch] = useReducer(
    {count: initialCount}  );


React doesn’t use the state = initialState argument convention popularized by Redux. The initial value sometimes needs to depend on props and so is specified from the Hook call instead. If you feel strongly about this, you can call useReducer(reducer, undefined, reducer) to emulate the Redux behavior, but it’s not encouraged.

Lazy initialization

You can also create the initial state lazily. To do this, you can pass an init function as the third argument. The initial state will be set to init(initialArg).

It lets you extract the logic for calculating the initial state outside the reducer. This is also handy for resetting the state later in response to an action:

function init(initialCount) {  return {count: initialCount};}
function reducer(state, action) {
  switch (action.type) {
    case 'increment':
      return {count: state.count + 1};
    case 'decrement':
      return {count: state.count - 1};
    case 'reset':      return init(action.payload);    default:
      throw new Error();

function Counter({initialCount}) {
  const [state, dispatch] = useReducer(reducer, initialCount, init);  return (
      Count: {state.count}
        onClick={() => dispatch({type: 'reset', payload: initialCount})}>        Reset
      <button onClick={() => dispatch({type: 'decrement'})}>-</button>
      <button onClick={() => dispatch({type: 'increment'})}>+</button>
Bailing out of a dispatch

If you return the same value from a Reducer Hook as the current state, React will bail out without rendering the children or firing effects. (React uses the comparison algorithm.)

Note that React may still need to render that specific component again before bailing out. That shouldn’t be a concern because React won’t unnecessarily go “deeper” into the tree. If you’re doing expensive calculations while rendering, you can optimize them with useMemo.


const memoizedCallback = useCallback(
  () => {
    doSomething(a, b);
  [a, b],

Returns a memoized callback.

Pass an inline callback and an array of dependencies. useCallback will return a memoized version of the callback that only changes if one of the dependencies has changed. This is useful when passing callbacks to optimized child components that rely on reference equality to prevent unnecessary renders (e.g. shouldComponentUpdate).

useCallback(fn, deps) is equivalent to useMemo(() => fn, deps).


The array of dependencies is not passed as arguments to the callback. Conceptually, though, that’s what they represent: every value referenced inside the callback should also appear in the dependencies array. In the future, a sufficiently advanced compiler could create this array automatically.

We recommend using the exhaustive-depsrule as part of oureslint-plugin-react-hooks package. It warns when dependencies are specified incorrectly and suggests a fix.


const memoizedValue = useMemo(() => computeExpensiveValue(a, b), [a, b]);

Returns a memoized value.

Pass a “create” function and an array of dependencies. useMemo will only recompute the memoized value when one of the dependencies has changed. This optimization helps to avoid expensive calculations on every render.

Remember that the function passed to useMemo runs during rendering. Don’t do anything there that you wouldn’t normally do while rendering. For example, side effects belong in useEffect, not useMemo.

If no array is provided, a new value will be computed on every render.

You may rely on **useMemo** as a performance optimization, not as a semantic guarantee. In the future, React may choose to “forget” some previously memoized values and recalculate them on next render, e.g. to free memory for offscreen components. Write your code so that it still works without useMemo — and then add it to optimize performance.


The array of dependencies is not passed as arguments to the function. Conceptually, though, that’s what they represent: every value referenced inside the function should also appear in the dependencies array. In the future, a sufficiently advanced compiler could create this array automatically.

We recommend using the exhaustive-depsrule as part of oureslint-plugin-react-hooks package. It warns when dependencies are specified incorrectly and suggests a fix.


const refContainer = useRef(initialValue);

useRef returns a mutable ref object whose .current property is initialized to the passed argument (initialValue). The returned object will persist for the full lifetime of the component.

A common use case is to access a child imperatively:

function TextInputWithFocusButton() {
  const inputEl = useRef(null);
  const onButtonClick = () => {
    // `current` points to the mounted text input element
  return (
      <input ref={inputEl} type="text" />
      <button onClick={onButtonClick}>Focus the input</button>

Essentially, useRef is like a “box” that can hold a mutable value in its .current property.

You might be familiar with refs primarily as a way to access the DOM. If you pass a ref object to React with <div ref={myRef} />, React will set its .current property to the corresponding DOM node whenever that node changes.

However, useRef() is useful for more than the ref attribute. It’s handy for keeping any mutable value around similar to how you’d use instance fields in classes.

This works because useRef() creates a plain JavaScript object. The only difference between useRef() and creating a {current: ...} object yourself is that useRef will give you the same ref object on every render.

Keep in mind that useRef doesn’t notify you when its content changes. Mutating the .current property doesn’t cause a re-render. If you want to run some code when React attaches or detaches a ref to a DOM node, you may want to use a callback ref instead.


useImperativeHandle(ref, createHandle, [deps])

useImperativeHandle customizes the instance value that is exposed to parent components when using ref. As always, imperative code using refs should be avoided in most cases. useImperativeHandle should be used with forwardRef:

function FancyInput(props, ref) {
  const inputRef = useRef();
  useImperativeHandle(ref, () => ({
    focus: () => {
  return <input ref={inputRef} ... />;
FancyInput = forwardRef(FancyInput);

In this example, a parent component that renders <FancyInput ref={inputRef} /> would be able to call inputRef.current.focus().


The signature is identical to useEffect, but it fires synchronously after all DOM mutations. Use this to read layout from the DOM and synchronously re-render. Updates scheduled inside useLayoutEffect will be flushed synchronously, before the browser has a chance to paint.

Prefer the standard useEffect when possible to avoid blocking visual updates.


If you’re migrating code from a class component, note useLayoutEffect fires in the same phase as componentDidMount and componentDidUpdate. However, we recommend starting with **useEffect** first and only trying useLayoutEffect if that causes a problem.

If you use server rendering, keep in mind that neither useLayoutEffect nor useEffect can run until the JavaScript is downloaded. This is why React warns when a server-rendered component contains useLayoutEffect. To fix this, either move that logic to useEffect (if it isn’t necessary for the first render), or delay showing that component until after the client renders (if the HTML looks broken until useLayoutEffect runs).

To exclude a component that needs layout effects from the server-rendered HTML, render it conditionally with showChild && <Child /> and defer showing it with useEffect(() => { setShowChild(true); }, []). This way, the UI doesn’t appear broken before hydration.



useDebugValue can be used to display a label for custom hooks in React DevTools.

For example, consider the useFriendStatus custom Hook described in “Building Your Own Hooks”:

function useFriendStatus(friendID) {
  const [isOnline, setIsOnline] = useState(null);

  // ...

  // Show a label in DevTools next to this Hook  // e.g. "FriendStatus: Online"  useDebugValue(isOnline ? 'Online' : 'Offline');
  return isOnline;


We don’t recommend adding debug values to every custom Hook. It’s most valuable for custom Hooks that are part of shared libraries.

Defer formatting debug values

In some cases formatting a value for display might be an expensive operation. It’s also unnecessary unless a Hook is actually inspected.

For this reason useDebugValue accepts a formatting function as an optional second parameter. This function is only called if the Hooks are inspected. It receives the debug value as a parameter and should return a formatted display value.

For example a custom Hook that returned a Date value could avoid calling the toDateString function unnecessarily by passing the following formatter:

useDebugValue(date, date => date.toDateString());

#react #api #javascript #web-development #developer

What is GEEK

Buddha Community

React Tutorial for Beginners - Hooks API Reference
Archie  Powell

Archie Powell


Tensorflex: Tensorflow Bindings for The Elixir Programming Language


The paper detailing Tensorflex was presented at NeurIPS/NIPS 2018 as part of the MLOSS workshop. The paper can be found here


How to run

  • You need to have the Tensorflow C API installed. Look here for details.
  • You also need the C library libjpeg. If you are using Linux or OSX, it should already be present on your machine, otherwise be sure to install (brew install libjpeg for OSX, and sudo apt-get install libjpeg-dev for Ubuntu).
  • Simply add Tensorflex to your list of dependencies in mix.exs and you are good to go!:
{:tensorflex, "~> 0.1.2"}

In case you want the latest development version use this:

{:tensorflex, github: "anshuman23/tensorflex"}


Tensorflex contains three main structs which handle different datatypes. These are %Graph, %Matrix and %Tensor. %Graph type structs handle pre-trained graph models, %Matrix handles Tensorflex 2-D matrices, and %Tensor handles Tensorflow Tensor types. The official Tensorflow documentation is present here and do note that this README only briefly discusses Tensorflex functionalities.


Used for loading a Tensorflow .pb graph model in Tensorflex.

Reads in a pre-trained Tensorflow protobuf (.pb) Graph model binary file.

Returns a tuple {:ok, %Graph}.

%Graph is an internal Tensorflex struct which holds the name of the graph file and the binary definition data that is read in via the .pb file.


Used for listing all the operations in a Tensorflow .pb graph.

Reads in a Tensorflex %Graph struct obtained from read_graph/1.

Returns a list of all the operation names (as strings) that populate the graph model.


Creates a 2-D Tensorflex matrix from custom input specifications.

Takes three input arguments: number of rows in matrix (nrows), number of columns in matrix (ncols), and a list of lists of the data that will form the matrix (datalist).

Returns a %Matrix Tensorflex struct type.


Used for accessing an element of a Tensorflex matrix.

Takes in three input arguments: a Tensorflex %Matrix struct matrix, and the row (row) and column (col) values of the required element in the matrix. Both row and col here are NOT zero indexed.

Returns the value as float.


Used for obtaining the size of a Tensorflex matrix.

Takes a Tensorflex %Matrix struct matrix as input.

Returns a tuple {nrows, ncols} where nrows represents the number of rows of the matrix and ncols represents the number of columns of the matrix.


Appends a single row to the back of a Tensorflex matrix.

Takes a Tensorflex %Matrix matrix as input and a single row of data (with the same number of columns as the original matrix) as a list of lists (datalist) to append to the original matrix.

Returns the extended and modified %Matrix struct matrix.


Converts a Tensorflex matrix (back) to a list of lists format.

Takes a Tensorflex %Matrix struct matrix as input.

Returns a list of lists representing the data stored in the matrix.

NOTE: If the matrix contains very high dimensional data, typically obtained from a function like load_csv_as_matrix/2, then it is not recommended to convert the matrix back to a list of lists format due to a possibility of memory errors.

float64_tensor/2, float32_tensor/2, int32_tensor/2:

Creates a TF_DOUBLE, TF_FLOAT, or TF_INT32 tensor from Tensorflex matrices containing the values and dimensions specified.

Takes two arguments: a %Matrix matrix (matrix1) containing the values the tensor should have and another %Matrix matrix (matrix2) containing the dimensions of the required tensor.

Returns a tuple {:ok, %Tensor} where %Tensor represents an internal Tensorflex struct type that is used for holding tensor data and type.

float64_tensor/1, float32_tensor/1, int32_tensor/1, string_tensor/1:

Creates a TF_DOUBLE, TF_FLOAT, TF_INT32, or TF_STRING constant value one-dimensional tensor from the input value specified.

Takes in a float, int or string value (depending on function) as input.

Returns a tuple {:ok, %Tensor} where %Tensor represents an internal Tensorflex struct type that is used for holding tensor data and type.

float64_tensor_alloc/1, float32_tensor_alloc/1, int32_tensor_alloc/1:

Allocates a TF_DOUBLE, TF_FLOAT, or TF_INT32 tensor of specified dimensions.

This function is generally used to allocate output tensors that do not hold any value data yet, but will after the session is run for Inference. Output tensors of the required dimensions are allocated and then passed to the run_session/5 function to hold the output values generated as predictions.

Takes a Tensorflex %Matrix struct matrix as input.

Returns a tuple {:ok, %Tensor} where %Tensor represents an internal Tensorflex struct type that is used for holding the potential tensor data and type.


Used to get the datatype of a created tensor.

Takes in a %Tensor struct tensor as input.

Returns a tuple {:ok, datatype} where datatype is an atom representing the list of Tensorflow TF_DataType tensor datatypes. Click here to view a list of all possible datatypes.


Loads JPEG images into Tensorflex directly as a TF_UINT8 tensor of dimensions image height x image width x number of color channels.

This function is very useful if you wish to do image classification using Convolutional Neural Networks, or other Deep Learning Models. One of the most widely adopted and robust image classification models is the Inception model by Google. It makes classifications on images from over a 1000 classes with highly accurate results. The load_image_as_tensor/1 function is an essential component for the prediction pipeline of the Inception model (and for other similar image classification models) to work in Tensorflex.

Reads in the path to a JPEG image file (.jpg or .jpeg).

Returns a tuple {:ok, %Tensor} where %Tensor represents an internal Tensorflex struct type that is used for holding the tensor data and type. Here the created Tensor is a uint8 tensor (TF_UINT8).

NOTE: For now, only 3 channel RGB JPEG color images can be passed as arguments. Support for grayscale images and other image formats such as PNG will be added in the future.


Loads high-dimensional data from a CSV file as a Tensorflex 2-D matrix in a super-fast manner.

The load_csv_as_matrix/2 function is very fast-- when compared with the Python based pandas library for data science and analysis' function read_csv on the test.csv file from MNIST Kaggle data (source), the following execution times were obtained:

  • read_csv: 2.549233 seconds
  • load_csv_as_matrix/2: 1.711494 seconds

This function takes in 2 arguments: a path to a valid CSV file (filepath) and other optional arguments opts. These include whether or not a header needs to be discarded in the CSV, and what the delimiter type is. These are specified by passing in an atom :true or :false to the header: key, and setting a string value for the delimiter: key. By default, the header is considered to be present (:true) and the delimiter is set to ,.

Returns a %Matrix Tensorflex struct type.


Runs a Tensorflow session to generate predictions for a given graph, input data, and required input/output operations.

This function is the final step of the Inference (prediction) pipeline and generates output for a given set of input data, a pre-trained graph model, and the specified input and output operations of the graph.

Takes in five arguments: a pre-trained Tensorflow graph .pb model read in from the read_graph/1 function (graph), an input tensor with the dimensions and data required for the input operation of the graph to run (tensor1), an output tensor allocated with the right dimensions (tensor2), the name of the input operation of the graph that needs where the input data is fed (input_opname), and the output operation name in the graph where the outputs are obtained (output_opname). The input tensor is generally created from the matrices manually or using the load_csv_as_matrix/2 function, and then passed through to one of the tensor creation functions. For image classification the load_image_as_tensor/1 can also be used to create the input tensor from an image. The output tensor is created using the tensor allocation functions (generally containing alloc at the end of the function name).

Returns a List of Lists (similar to the matrix_to_lists/1 function) containing the generated predictions as per the output tensor dimensions.


Adds scalar value to matrix.

Takes two arguments: %Matrix matrix and scalar value (int or float)

Returns a %Matrix modified matrix.


Subtracts scalar value from matrix.

Takes two arguments: %Matrix matrix and scalar value (int or float)

Returns a %Matrix modified matrix.


Multiplies scalar value with matrix.

Takes two arguments: %Matrix matrix and scalar value (int or float)

Returns a %Matrix modified matrix.


Divides matrix values by scalar.

Takes two arguments: %Matrix matrix and scalar value (int or float)

Returns a %Matrix modified matrix.


Adds two matrices of same dimensions together.

Takes in two %Matrix matrices as arguments.

Returns the resultant %Matrix matrix.


Subtracts matrix2 from matrix1.

Takes in two %Matrix matrices as arguments.

Returns the resultant %Matrix matrix.


Converts the data stored in a 2-D tensor back to a 2-D matrix.

Takes in a single argument as a %Tensor tensor (any TF_Datatype).

Returns a %Matrix 2-D matrix.

NOTE: Tensorflex doesn't currently support 3-D matrices, and therefore tensors that are 3-D (such as created using the load_image_as_tensor/1 function) cannot be converted back to a matrix, yet. Support for 3-D matrices will be added soon.


Examples are generally added in full description on my blog here. A blog post covering how to do classification on the Iris Dataset is present here.


Here we will briefly touch upon how to use the Google V3 Inception pre-trained graph model to do image classficiation from over a 1000 classes. First, the Inception V3 model can be downloaded here:

After unzipping, see that it contains the graphdef .pb file (classify_image_graphdef.pb) which contains our graph definition, a test jpeg image that should identify/classify as a panda (cropped_panda.pb) and a few other files I will detail later.

Now for running this in Tensorflex first the graph is loaded:

iex(1)> {:ok, graph} = Tensorflex.read_graph("classify_image_graph_def.pb")
2018-07-29 00:48:19.849870: W tensorflow/core/framework/] Op BatchNormWithGlobalNormalization is deprecated. It will cease to work in GraphDef version 9. Use tf.nn.batch_normalization().
   def: #Reference<0.2597534446.2498625538.211058>,
   name: "classify_image_graph_def.pb"

Then the cropped_panda image is loaded using the new load_image_as_tensor function:

iex(2)> {:ok, input_tensor} = Tensorflex.load_image_as_tensor("cropped_panda.jpg")
   datatype: :tf_uint8,
   tensor: #Reference<0.2597534446.2498625538.211093>

Then create the output tensor which will hold out output vector values. For the inception model, the output is received as a 1008x1 tensor, as there are 1008 classes in the model:

iex(3)> out_dims = Tensorflex.create_matrix(1,2,[[1008,1]])
  data: #Reference<0.2597534446.2498625538.211103>,
  ncols: 2,
  nrows: 1

iex(4)> {:ok, output_tensor} = Tensorflex.float32_tensor_alloc(out_dims)
   datatype: :tf_float,
   tensor: #Reference<0.2597534446.2498625538.211116>

Then the output results are read into a list called results. Also, the input operation in the Inception model is DecodeJpeg and the output operation is softmax:

iex(5)> results = Tensorflex.run_session(graph, input_tensor, output_tensor, "DecodeJpeg", "softmax")
2018-07-29 00:51:13.631154: I tensorflow/core/platform/] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA
  [1.059142014128156e-4, 2.8240500250831246e-4, 8.30648496048525e-5,
   1.2982363114133477e-4, 7.32232874725014e-5, 8.014426566660404e-5,
   6.63459359202534e-5, 0.003170756157487631, 7.931600703159347e-5,
   3.707312498590909e-5, 3.0997329304227605e-5, 1.4232713147066534e-4,
   1.0381334868725389e-4, 1.1057958181481808e-4, 1.4321311027742922e-4,
   1.203602587338537e-4, 1.3130248407833278e-4, 5.850398520124145e-5,
   2.641105093061924e-4, 3.1629020668333396e-5, 3.906813799403608e-5,
   2.8646905775531195e-5, 2.2863158665131778e-4, 1.2222197256051004e-4,
   5.956588938715868e-5, 5.421260357252322e-5, 5.996063555357978e-5,
   4.867801326327026e-4, 1.1005574924638495e-4, 2.3433618480339646e-4,
   1.3062104699201882e-4, 1.317620772169903e-4, 9.388553007738665e-5,
   7.076268957462162e-5, 4.281177825760096e-5, 1.6863139171618968e-4,
   9.093972039408982e-5, 2.611844101920724e-4, 2.7584232157096267e-4,
   5.157176201464608e-5, 2.144951868103817e-4, 1.3628098531626165e-4,
   8.007588621694595e-5, 1.7929042223840952e-4, 2.2831936075817794e-4,
   6.216531619429588e-5, 3.736453436431475e-5, 6.782123091397807e-5,
   1.1538144462974742e-4, ...]

Finally, we need to find which class has the maximum probability and identify it's label. Since results is a List of Lists, it's better to read in the nested list. Then we need to find the index of the element in the new list which as the maximum value. Therefore:

iex(6)> max_prob = List.flatten(results) |> Enum.max

iex(7)> Enum.find_index(results |> List.flatten, fn(x) -> x == max_prob end)

We can thus see that the class with the maximum probability predicted (0.8849328756332397) for the image is 169. We will now find what the 169 label corresponds to. For this we can look back into the unzipped Inception folder, where there is a file called imagenet_2012_challenge_label_map_proto.pbtxt. On opening this file, we can find the string class identifier for the 169 class index. This is n02510455 and is present on Line 1556 in the file. Finally, we need to match this string identifier to a set of identification labels by referring to the file imagenet_synset_to_human_label_map.txt file. Here we can see that corresponding to the string class n02510455 the human labels are giant panda, panda, panda bear, coon bear, Ailuropoda melanoleuca (Line 3691 in the file).

Thus, we have correctly identified the animal in the image as a panda using Tensorflex!


A brief idea of what this example entails:

  • The Recurrent Neural Network utilizes Long-Short-Term-Memory (LSTM) cells for holding the state for the data flowing in through the network
  • In this example, we utilize the LSTM network for sentiment analysis on movie reviews data in Tensorflex. The trained models are originally created as part of an online tutorial (source) and are present in a Github repository here.

To do sentiment analysis in Tensorflex however, we first need to do some preprocessing and prepare the graph model (.pb) as done multiple times before in other examples. For that, in the examples/rnn-lstm-example directory there are two scripts: and Prior to explaining the working of these scripts you first need to download the original saved models as well as the datasets:

  • For the model, download from here and then store all the 4 model files in the examples/rnn-lstm-example/model folder
  • For the dataset, download from here. After decompressing, we do not need all the files, just the 2 numpy binaries wordsList.npy and wordVectors.npy. These will be used to encode our text data into UTF-8 encoding for feeding our RNN as input.

Now, for the Python two scripts: and

  • This is used to create our pb model from the Python saved checkpoints. Here we will use the downloaded Python checkpoints' model to create the .pb graph. Just running python after putting the model files in the correct directory will do the trick. In the same ./model/ folder, you will now see a file called frozen_model_lstm.pb. This is the file which we will load into Tensorflex. In case for some reason you want to skip this step and just get the loaded graph here is a Dropbox link
  • Even if we can load our model into Tensorflex, we also need some data to do inference on. For that, we will write our own example sentences and convert them (read encode) to a numeral (int32) format that can be used by the network as input. For that, you can inspect the code in the script to get an understanding of what is happening. Basically, the neural network takes in an input of a 24x250 int32 (matrix) tensor created from text which has been encoded as UTF-8. Again, running python will give you two csv files (one indicating positive sentiment and the other a negative sentiment) which we will later load into Tensorflex. The two sentences converted are:
    • Negative sentiment sentence: That movie was terrible.
    • Positive sentiment sentence: That movie was the best one I have ever seen.

Both of these get converted to two files inputMatrixPositive.csv and inputMatrixNegative.csv (by which we load into Tensorflex next.

Inference in Tensorflex: Now we do sentiment analysis in Tensorflex. A few things to note:

  • The input graph operation is named Placeholder_1
  • The output graph operation is named add and is the eventual result of a matrix multiplication. Of this obtained result we only need the first row
  • Here the input is going to be a integer valued matrix tensor of dimensions 24x250 representing our sentence/review
  • The output will have 2 columns, as there are 2 classes-- for positive and negative sentiment respectively. Since we will only be needing only the first row we will get our result in a 1x2 vector. If the value of the first column is higher than the second column, then the network indicates a positive sentiment otherwise a negative sentiment. All this can be observed in the original repository in a Jupyter notebook here: ```elixir iex(1)> {:ok, graph} = Tensorflex.read_graph "examples/rnn-lstm-example/model/frozen_model_lstm.pb" {:ok, %Tensorflex.Graph{ def: #Reference<0.713975820.1050542081.11558>, name: "examples/rnn-lstm-example/model/frozen_model_lstm.pb" }}

iex(2)> Tensorflex.get_graph_ops graph ["Placeholder_1", "embedding_lookup/params_0", "embedding_lookup", "transpose/perm", "transpose", "rnn/Shape", "rnn/strided_slice/stack", "rnn/strided_slice/stack_1", "rnn/strided_slice/stack_2", "rnn/strided_slice", "rnn/stack/1", "rnn/stack", "rnn/zeros/Const", "rnn/zeros", "rnn/stack_1/1", "rnn/stack_1", "rnn/zeros_1/Const", "rnn/zeros_1", "rnn/Shape_1", "rnn/strided_slice_2/stack", "rnn/strided_slice_2/stack_1", "rnn/strided_slice_2/stack_2", "rnn/strided_slice_2", "rnn/time", "rnn/TensorArray", "rnn/TensorArray_1", "rnn/TensorArrayUnstack/Shape", "rnn/TensorArrayUnstack/strided_slice/stack", "rnn/TensorArrayUnstack/strided_slice/stack_1", "rnn/TensorArrayUnstack/strided_slice/stack_2", "rnn/TensorArrayUnstack/strided_slice", "rnn/TensorArrayUnstack/range/start", "rnn/TensorArrayUnstack/range/delta", "rnn/TensorArrayUnstack/range", "rnn/TensorArrayUnstack/TensorArrayScatter/TensorArrayScatterV3", "rnn/while/Enter", "rnn/while/Enter_1", "rnn/while/Enter_2", "rnn/while/Enter_3", "rnn/while/Merge", "rnn/while/Merge_1", "rnn/while/Merge_2", "rnn/while/Merge_3", "rnn/while/Less/Enter", "rnn/while/Less", "rnn/while/LoopCond", "rnn/while/Switch", "rnn/while/Switch_1", "rnn/while/Switch_2", "rnn/while/Switch_3", ...]

First we will try for positive sentiment:
iex(3)> input_vals = Tensorflex.load_csv_as_matrix("examples/rnn-lstm-example/inputMatrixPositive.csv", header: :false)
  data: #Reference<0.713975820.1050542081.13138>,
  ncols: 250,
  nrows: 24

iex(4)> input_dims = Tensorflex.create_matrix(1,2,[[24,250]])
  data: #Reference<0.713975820.1050542081.13575>,
  ncols: 2,
  nrows: 1

iex(5)> {:ok, input_tensor} = Tensorflex.int32_tensor(input_vals, input_dims)
   datatype: :tf_int32,
   tensor: #Reference<0.713975820.1050542081.14434>

iex(6)> output_dims = Tensorflex.create_matrix(1,2,[[24,2]])
  data: #Reference<0.713975820.1050542081.14870>,
  ncols: 2,
  nrows: 1

iex(7)> {:ok, output_tensor} = Tensorflex.float32_tensor_alloc(output_dims)
   datatype: :tf_float,
   tensor: #Reference<0.713975820.1050542081.15363>

We only need the first row, the rest do not indicate anything:

iex(8)> [result_pos | _ ] = Tensorflex.run_session(graph, input_tensor,output_tensor, "Placeholder_1", "add")
  [4.483788013458252, -1.273943305015564],
  [-0.17151066660881042, -2.165886402130127],
  [0.9569928646087646, -1.131581425666809],
  [0.5669126510620117, -1.3842089176177979],
  [-1.4346938133239746, -4.0750861167907715],
  [0.4680981934070587, -1.3494354486465454],
  [1.068990707397461, -2.0195648670196533],
  [3.427264451980591, 0.48857203125953674],
  [0.6307879686355591, -2.069119691848755],
  [0.35061028599739075, -1.700657844543457],
  [3.7612719535827637, 2.421398878097534],
  [2.7635951042175293, -0.7214710116386414],
  [1.146680235862732, -0.8688814640045166],
  [0.8996094465255737, -1.0183486938476563],
  [0.23605018854141235, -1.893072247505188],
  [2.8790698051452637, -0.37355837225914],
  [-1.7325369119644165, -3.6470277309417725],
  [-1.687785029411316, -4.903762340545654],
  [3.6726789474487305, 0.14170047640800476],
  [0.982108473777771, -1.554244875907898],
  [2.248904228210449, 1.0617655515670776],
  [0.3663095533847809, -3.5266385078430176],
  [-1.009346604347229, -2.901120901107788],
  [3.0659966468811035, -1.7605335712432861]

iex(9)> result_pos
[4.483788013458252, -1.273943305015564]

Thus we can clearly see that the RNN predicts a positive sentiment. For a negative sentiment, next:

iex(10)> input_vals = Tensorflex.load_csv_as_matrix("examples/rnn-lstm-example/inputMatrixNegative.csv", header: :false)
  data: #Reference<0.713975820.1050542081.16780>,
  ncols: 250,
  nrows: 24

iex(11)> {:ok, input_tensor} = Tensorflex.int32_tensor(input_vals,input_dims)
   datatype: :tf_int32,
   tensor: #Reference<0.713975820.1050542081.16788>

iex(12)> [result_neg|_] = Tensorflex.run_session(graph, input_tensor,output_tensor, "Placeholder_1", "add")
  [0.7635725736618042, 10.895986557006836],
  [2.205151319503784, -0.6267685294151306],
  [3.5995595455169678, -0.1240251287817955],
  [-1.6063352823257446, -3.586883068084717],
  [1.9608432054519653, -3.084211826324463],
  [3.772461414337158, -0.19421455264091492],
  [3.9185996055603027, 0.4442034661769867],
  [3.010765552520752, -1.4757057428359985],
  [3.23650860786438, -0.008513949811458588],
  [2.263028144836426, -0.7358709573745728],
  [0.206748828291893, -2.1945853233337402],
  [2.913491725921631, 0.8632720708847046],
  [0.15935257077217102, -2.9757845401763916],
  [-0.7757357358932495, -2.360766649246216],
  [3.7359719276428223, -0.7668198347091675],
  [2.2896337509155273, -0.45704856514930725],
  [-1.5497230291366577, -4.42919921875],
  [-2.8478822708129883, -5.541027545928955],
  [1.894787073135376, -0.8441318273544312],
  [0.15720489621162415, -2.699129819869995],
  [-0.18114641308784485, -2.988100051879883],
  [3.342879056930542, 2.1714375019073486],
  [2.906526565551758, 0.18969044089317322],
  [0.8568912744522095, -1.7559258937835693]
iex(13)> result_neg
[0.7635725736618042, 10.895986557006836]

Thus we can clearly see that in this case the RNN indicates negative sentiment! Our model works!

Pull Requests Made

Author: anshuman23
Source code:
License: Apache-2.0 license

#elixir #tensorflow #machine-learning 

Autumn  Blick

Autumn Blick


How native is React Native? | React Native vs Native App Development

If you are undertaking a mobile app development for your start-up or enterprise, you are likely wondering whether to use React Native. As a popular development framework, React Native helps you to develop near-native mobile apps. However, you are probably also wondering how close you can get to a native app by using React Native. How native is React Native?

In the article, we discuss the similarities between native mobile development and development using React Native. We also touch upon where they differ and how to bridge the gaps. Read on.

A brief introduction to React Native

Let’s briefly set the context first. We will briefly touch upon what React Native is and how it differs from earlier hybrid frameworks.

React Native is a popular JavaScript framework that Facebook has created. You can use this open-source framework to code natively rendering Android and iOS mobile apps. You can use it to develop web apps too.

Facebook has developed React Native based on React, its JavaScript library. The first release of React Native came in March 2015. At the time of writing this article, the latest stable release of React Native is 0.62.0, and it was released in March 2020.

Although relatively new, React Native has acquired a high degree of popularity. The “Stack Overflow Developer Survey 2019” report identifies it as the 8th most loved framework. Facebook, Walmart, and Bloomberg are some of the top companies that use React Native.

The popularity of React Native comes from its advantages. Some of its advantages are as follows:

  • Performance: It delivers optimal performance.
  • Cross-platform development: You can develop both Android and iOS apps with it. The reuse of code expedites development and reduces costs.
  • UI design: React Native enables you to design simple and responsive UI for your mobile app.
  • 3rd party plugins: This framework supports 3rd party plugins.
  • Developer community: A vibrant community of developers support React Native.

Why React Native is fundamentally different from earlier hybrid frameworks

Are you wondering whether React Native is just another of those hybrid frameworks like Ionic or Cordova? It’s not! React Native is fundamentally different from these earlier hybrid frameworks.

React Native is very close to native. Consider the following aspects as described on the React Native website:

  • Access to many native platforms features: The primitives of React Native render to native platform UI. This means that your React Native app will use many native platform APIs as native apps would do.
  • Near-native user experience: React Native provides several native components, and these are platform agnostic.
  • The ease of accessing native APIs: React Native uses a declarative UI paradigm. This enables React Native to interact easily with native platform APIs since React Native wraps existing native code.

Due to these factors, React Native offers many more advantages compared to those earlier hybrid frameworks. We now review them.

#android app #frontend #ios app #mobile app development #benefits of react native #is react native good for mobile app development #native vs #pros and cons of react native #react mobile development #react native development #react native experience #react native framework #react native ios vs android #react native pros and cons #react native vs android #react native vs native #react native vs native performance #react vs native #why react native #why use react native

What are hooks in React JS? - INFO AT ONE

In this article, you will learn what are hooks in React JS? and when to use react hooks? React JS is developed by Facebook in the year 2013. There are many students and the new developers who have confusion between react and hooks in react. Well, it is not different, react is a programming language and hooks is a function which is used in react programming language.
Read More:-

#react #hooks in react #react hooks example #react js projects for beginners #what are hooks in react js? #when to use react hooks

Sival Alethea

Sival Alethea


APIs for Beginners - How to use an API (Full Course / Tutorial)

What is an API? Learn all about APIs (Application Programming Interfaces) in this full tutorial for beginners. You will learn what APIs do, why APIs exist, and the many benefits of APIs. APIs are used all the time in programming and web development so it is important to understand how to use them.

You will also get hands-on experience with a few popular web APIs. As long as you know the absolute basics of coding and the web, you’ll have no problem following along.
⭐️ Unit 1 - What is an API
⌨️ Video 1 - Welcome (0:00:00)
⌨️ Video 2 - Defining Interface (0:03:57)
⌨️ Video 3 - Defining API (0:07:51)
⌨️ Video 4 - Remote APIs (0:12:55)
⌨️ Video 5 - How the web works (0:17:04)
⌨️ Video 6 - RESTful API Constraint Scavenger Hunt (0:22:00)

⭐️ Unit 2 - Exploring APIs
⌨️ Video 1 - Exploring an API online (0:27:36)
⌨️ Video 2 - Using an API from the command line (0:44:30)
⌨️ Video 3 - Using Postman to explore APIs (0:53:56)
⌨️ Video 4 - Please please Mr. Postman (1:03:33)
⌨️ Video 5 - Using Helper Libraries (JavaScript) (1:14:41)
⌨️ Video 6 - Using Helper Libraries (Python) (1:24:40)

⭐️ Unit 3 - Using APIs
⌨️ Video 1 - Introducing the project (1:34:18)
⌨️ Video 2 - Flask app (1:36:07)
⌨️ Video 3 - Dealing with API Limits (1:50:00)
⌨️ Video 4 - JavaScript Single Page Application (1:54:27)
⌨️ Video 5 - Moar JavaScript and Recap (2:07:53)
⌨️ Video 6 - Review (2:18:03)
📺 The video in this post was made by
The origin of the article:
🔥 If you’re a beginner. I believe the article below will be useful to you ☞ What You Should Know Before Investing in Cryptocurrency - For Beginner
⭐ ⭐ ⭐The project is of interest to the community. Join to Get free ‘GEEK coin’ (GEEKCASH coin)!
☞ **-----CLICK HERE-----**⭐ ⭐ ⭐
Thanks for visiting and watching! Please don’t forget to leave a like, comment and share!

#apis #apis for beginners #how to use an api #apis for beginners - how to use an api #application programming interfaces #learn all about apis

Ray  Patel

Ray Patel


How to Validate and Geocode a Street Address in Java

When working with location services, it is important that the information you collect is accurate for your users or clients. Find out more!

When working with location services, it is important that the information you collect is accurate for your users or clients. This will prevent any mistakes in shipping, billing, and many other aspects of operations that rely on correct location information. For businesses that have applications using location services, this is especially important as any incorrect data can mean the displacement of goods or interrupted services.

The following APIs will allow you to fully validate street addresses by first parsing address data input and then verifying and normalizing the information. The last two APIs will also allow you to geocode and reverse geocode an address to receive more accurate location data for your applications.

#tutorial #api #address #java api #validation and verification #api access keys #api tutorial #validate #java api tutorials #api tutorials