Engineer Ray's Blog: 2020

2020年12月27日星期日

Testing the tests: What should we do when the test case self is buggy?

Nowadays, software testing is widely using to ensure quality and prevent bug, like unit test/AB test…. etc, But increasingly test case will causing some unwanted problem like the test case is buggy so the testing will failed. And causing developer waste time chasing down problems that potentially didn’t really exist

Facebook provide a method to detect the test case bug:

1. Using ML technology to predict what test case to run

2. All end-to-end tests will have some degree of flakiness, So make a index about how reliable in these test case

3. Assert that a test is sufficiently reliable and provide a scale to illustrate which tests are less reliable than they should be.

Software testing is a good tool to ensure the product quality, with more complicated code base and more complicated test case, it need carefully consider introduce method that test the test case to prevent waste time and increase testing quality.

2020年11月13日星期五

Performance tuning- Leveraging modern CPU branch predict mechanism.

In modern CPU the branch predictor is complicated and the deeper pipeline causing the cost of miss branch prediction higher.

How can we leveraging the modern CPU branch predictor, In [1] example code, the branch miss rate can be reduce by just sorting input data before exciting original algorithm.

Conclusion

1. Adding pattern to your algorithm(sorted data).

2. Using likely()/unlikely() marco to help branch prediction more accuracy.

Reference:

1. Performance analysis in Linux

2. Code DetectiveHow to investigate LinuxPerformance Issues

3. Linux likely() unlikely() MARCO.

2020年6月22日星期一

Machine Learning Foundations NLP(2): Using the Sequencing APIs

Using the Sequencing APIs

Sequencing is use to format sentence array using token,
For example:

"I have a car"
"Car have a door"

Tokenize: [I:1] [have:2] [a:3] [car:4] [door:5]

then these two sentence can be represent to:
[1 2 3 4]
[4 2 3 5]

Sequencing is useful to represent sentence data and take as input for a neuron network.

Code

	import tensorflow as tf
	from tensorflow import keras

	from tensorflow.keras.preprocessing.text import Tokenizer
	from tensorflow.keras.preprocessing.sequence import pad_sequences

	# Sentences data
	sentences = {
	'I have a car',
	'I have a pen',
	'I have a bike',
	'He have a apple and a banana'
	}

	## Make a tokenizer with max 100 tokens, and label stranger token with OOV index
	tokenizer = Tokenizer(num_words = 100, oov_token="<OOV>")

	# Tokenize the sentence data
	tokenizer.fit_on_texts(sentences)
	word_index = tokenizer.word_index

	# Print the word index tokenized from sentence data
	# The more often a word be use the lower index
	word_index = tokenizer.word_index
	print(word_index)


	## Make to sentence array
	sequences = tokenizer.texts_to_sequences(sentences)

	## Make all sequence array length to 5, it will helpful when we train a NLP neuron network
	padded = pad_sequences(sequences, maxlen=5)
	print("\nWord Index = " , word_index)
	print("\nSequences = " , sequences)
	print("\nPadded Sequences:")
	print(padded)

	# Try with words that the tokenizer wasn't fit to
	# manatee is unseen token so it will be replace to OOV:1
	test_data = [
	'I have a big car',
	'He like banana'
	]

	test_seq = tokenizer.texts_to_sequences(test_data)
	print("\nTest Sequence = ", test_seq)

view raw gistfile1.txt hosted with ❤ by GitHub

Result:

{'<OOV>': 1, 'a': 2, 'have': 3, 'i': 4, 'car': 5, 'pen': 6, 'bike': 7, 'he': 8, 'apple': 9, 'and': 10, 'banna': 11}

Word Index = {'<OOV>': 1, 'a': 2, 'have': 3, 'i': 4, 'car': 5, 'pen': 6, 'bike': 7, 'he': 8, 'apple': 9, 'and': 10, 'banna': 11}

Sequences = [[4, 3, 2, 5], [4, 3, 2, 6], [4, 3, 2, 7], [8, 3, 2, 9, 10, 2, 11]]

Padded Sequences:

[[ 0 4 3 2 5]

[ 0 4 3 2 6]

[ 0 4 3 2 7]

[ 2 9 10 2 11]]

Test Sequence = [[4, 3, 2, 1, 5], [8, 1, 11]]

view raw gistfile1.txt hosted with ❤ by GitHub

Reference:

Machine Learning Foundations: Ep #9 - Using the Sequencing APIs

2020年6月17日星期三

Machine Learning Foundations NLP(1): Tokenization for Natural Language Processing

Tokenization for Natural Language Processing

Tokenize is mean to break down a sentence to server work, for example:

I have a car. -> I / have / a/ car

Tensorflow provide a tokenize tool:Tokenizer, it can easily to use for tokenize input sentence.

Code:

	import tensorflow as tf
	from tensorflow import keras
	from tensorflow.keras.preprocessing.text import Tokenizer

	# Sentences data
	sentences = {
	'I have a car',
	'I have a pen',
	'I have a bike',
	'He have a apple'
	}

	# Create an tokenizer with maxium 10 words
	tokenizer = Tokenizer(num_words = 10)

	# Tokenize the sentence data
	tokenizer.fit_on_texts(sentences)

	# Print the word index tokenized from sentence data
	# The more often a word be use the lower index will be assign
	word_index = tokenizer.word_index
	print(word_index)

view raw gistfile1.txt hosted with ❤ by GitHub

Result:

{'have': 1, 'a': 2, 'i': 3, 'he': 4, 'apple': 5, 'bike': 6, 'pen': 7, 'car': 8}

Reference:

Machine Learning Foundations: Ep #8 - Tokenization for Natural Language Processing

2020年5月28日星期四

Linux kernel module: Add entry to debugfs and a read/write file.

Goal

This module will create Ray_DBG directory in debugfs and create a file REG for read/write. it's similar to previous module(create /proc/Ray), but easier to use and need less line of code.

Code

	#include <linux/module.h>
	#include <linux/init.h>
	#include <linux/kernel.h>
	#include <linux/debugfs.h>


	MODULE_LICENSE("GPL");
	MODULE_AUTHOR("Ray Tsai");

	static u32 reg_data = 0xaa;
	static struct dentry *root;

	static int __init test_module_init(void)
	{
	pr_info("Init Ray test module!\n");

	pr_info("Create Ray_DBG !\n");
	root = debugfs_create_dir("Ray_DBG", NULL);

	if(!root)
	return -EFAULT;

	pr_info("Create REG debug !\n");
	if(!debugfs_create_x32("REG", 0777, root, (u32 *)&reg_data))
	goto err;

	return 0;

	err:
	debugfs_remove_recursive(root);
	pr_info("Failed to initialize debugfs\n");
	return -EFAULT;
	}

	static void __exit test_module_exit(void)
	{
	debugfs_remove_recursive(root);
	pr_info("Exit Ray test module!\n");
	}

	module_init(test_module_init);
	module_exit(test_module_exit);

view raw gistfile1.txt hosted with ❤ by GitHub

Makefile

	obj-m := test_module.o



	KERNELDIR ?= /lib/modules/$(shell uname -r)/build



	all default: modules

	install: modules_install



	modules modules_install help clean:

	$(MAKE) -C $(KERNELDIR) M=$(shell pwd) $@

view raw gistfile1.txt hosted with ❤ by GitHub

Usage

1. Insert module

$>insmod test_module.ko
$>dmesg | tail
[7501255.251763] Create Ray_DBG !
[7501255.251776] Create REG debug !

2. Mount debugfs

$>mount -t debugfs none /sys/kernel/debug

3. Read data

$>cat /sys/kernel/debug/Ray_DBG/REG
$>0x000000aa

4. Write data

$>echo 0xff>  /sys/kernel/debug/Ray_DBG/REG 
$>cat /sys/kernel/debug/Ray_DBG/REG 
$>0x000000ff

Reference

Linux kernel module: Add entry in /proc and passing args while insmod

Goal

Implement a kernel module that can passing parameter when insert module, and add an entry to /proc that can write/read data. In embedded linux often using similar facility to help debug.

Code:

	#include <linux/module.h>
	#include <linux/moduleparam.h>
	#include <linux/init.h>
	#include <linux/kernel.h>
	#include <linux/proc_fs.h>
	#include <linux/uaccess.h>

	#define BUFSIZE 100

	MODULE_LICENSE("GPL");
	MODULE_AUTHOR("Ray Tsai");

	static int mode=0;
	module_param(mode,int,0660);

	static char *entry="Default";
	module_param(entry,charp,0660);

	// Proc related structure
	static struct proc_dir_entry *ent;
	static int private_test_data = 123;

	// Write handler
	static ssize_t test_write(struct file file, const char __user ubuf,size_t count, loff_t *ppos)
	{
	char buf[BUFSIZE];
	int temp = 0;
	int num =0;

	pr_info( KERN_DEBUG "write handler\n");

	if(copy_from_user(buf, ubuf, count))
	{
	pr_info( KERN_DEBUG "Error copy from user \n");
	return -EFAULT;
	}
	num = sscanf(buf,"%d",&temp);
	private_test_data = temp;

	return count;
	}

	// Read handler
	static ssize_t test_read(struct file file, char __user ubuf,size_t count, loff_t *ppos)
	{
	char buf[BUFSIZE];
	int len=0;

	pr_info( KERN_DEBUG "read handler\n");

	if(*ppos > 0 \|\| count < BUFSIZE)
	return 0;

	len += sprintf(buf,"private_test_data = %d\n",private_test_data);

	if(copy_to_user(ubuf,buf,len))
	return -EFAULT;

	*ppos = len;
	return len;
	}

	static struct file_operations myops =
	{
	.owner = THIS_MODULE,
	.read = test_read,
	.write = test_write,
	};

	static int __init test_module_init(void)
	{
	pr_info("Init Ray test module!\n");
	pr_info("Test module: %d!\n", mode);
	pr_info("Create %s \n", entry);
	ent = proc_create(entry,0660,NULL,&myops);
	return 0;
	}

	static void __exit test_module_exit(void)
	{
	proc_remove(ent);
	pr_info("Exit Ray test module!\n");
	}

	module_init(test_module_init);
	module_exit(test_module_exit);

view raw gistfile1.txt hosted with ❤ by GitHub

Makefile:


	obj-m := test_module.o

	KERNELDIR ?= /lib/modules/$(shell uname -r)/build

	all default: modules
	install: modules_install

	modules modules_install help clean:
	$(MAKE) -C $(KERNELDIR) M=$(shell pwd) $@

view raw gistfile1.txt hosted with ❤ by GitHub

Usage:

1. Insert module and make /proc/Ray entry
#>sudo insmod test_module.ko entry="Ray" mode=1238

2. Write data
#>echo 123 > /proc/Ray

3. Read data
#>cat /proc/Ray

Reference:

2020年5月27日星期三

DeepSpeech: Speech to text AI model.

Overview

"DeepSpeech is an open source Speech-To-Text engine, using a model trained by machine learning techniques based on Baidu’s Deep Speech research paper."
DeepSpeech provide lots of lauange api support, Python Javascript, c, and it's easily use to involve in application

Install DeepSpeech

Follow user guide instruction.

	# Create and activate a virtualenv
	virtualenv -p python3 $HOME/tmp/deepspeech-venv/
	source $HOME/tmp/deepspeech-venv/bin/activate

	# Install DeepSpeech
	pip3 install deepspeech

	# Download pre-trained English model files
	curl -LO https://github.com/mozilla/DeepSpeech/releases/download/v0.7.0/deepspeech-0.7.0-models.pbmm
	curl -LO https://github.com/mozilla/DeepSpeech/releases/download/v0.7.0/deepspeech-0.7.0-models.scorer

	# Download example audio files
	curl -LO https://github.com/mozilla/DeepSpeech/releases/download/v0.7.0/audio-0.7.0.tar.gz
	tar xvf audio-0.7.0.tar.gz

view raw gistfile1.txt hosted with ❤ by GitHub

Demo

Using command line tool to inference sound data.

$> deepspeech --model deepspeech-0.7.0-models.pbmm --audio audio/2830-3980-0043.wav

Output:
Loading model from file deepspeech-0.7.0-models.pbmm
TensorFlow: v1.15.0-24-gceb46aa
DeepSpeech: v0.7.1-0-g2e9c281
Loaded model in 0.0093s.
Loading scorer from files deepspeech-0.7.0-models.scorer
Loaded scorer in 0.00023s.
Running inference.
experience proves this
Inference took 1.480s for 1.975s audio file.

The red color string is inference text data of input sound data.

Using Python API to inference sound data.

import wave

import numpy as np

data = "audio/8455-210777-0068.wav" # your power is sufficient i said sound data

# using wave lib to read wav file

wf = wave.open(data, 'rb')

frames = wf.getnframes()

pcm_data = wf.readframes(frames)

wf.close()

# transfer audio data to int16 type

audio = np.frombuffer(pcm_data, dtype=np.int16)

from deepspeech import Model

# load pre-trained model

ds = Model("./deepspeech-0.7.0-models.pbmm")

# do inference

output = ds.stt(audio)

# print inference text data

print(output)

view raw gistfile1.txt hosted with ❤ by GitHub

Output:
your power is sufficient i said sound data

Reference:

2020年5月22日星期五

Machine Learning Foundations: Exercise 4 Happy and sad image classify model with 99.9% accuracy .

Machine Learning Foundations: Exercise 4 Happy and sad image classify model with 99.9% accuracy:code lab link

Build a mode to classify happy and sad model with convolution neuron network. with more 99.9% accuracy.

Note:

Model over-fitting: Near 100% accuracy on training data, but lower accuracy on testing data.
ImageDataGenerator: Label data automatic using directory name.

Code:

	import tensorflow as tf
	import os
	import zipfile


	DESIRED_ACCURACY = 0.999

	# Get the happy-or-sad image data
	!wget --no-check-certificate \
	"https://storage.googleapis.com/laurencemoroney-blog.appspot.com/happy-or-sad.zip" \
	-O "/tmp/happy-or-sad.zip"

	zip_ref = zipfile.ZipFile("/tmp/happy-or-sad.zip", 'r')
	zip_ref.extractall("/tmp/h-or-s")
	zip_ref.close()

	# Callback function to check model accuracy
	class RayCallback(tf.keras.callbacks.Callback):
	def on_epoch_end(self, epoch, logs={}):
	if(logs.get('accuracy')>DESIRED_ACCURACY):
	print("\nReached 99.9% accuracy so cancelling training!")
	self.model.stop_training = True

	callbacks = RayCallback()

	# 5 layer neuron network
	model = tf.keras.models.Sequential([
	tf.keras.layers.Conv2D(16, (3,3), activation='relu',
	input_shape=(300, 300, 3)), # Fit ImageDataGenerator output size
	tf.keras.layers.MaxPooling2D(2, 2),
	tf.keras.layers.Conv2D(32, (3,3), activation='relu'),
	tf.keras.layers.MaxPooling2D(2,2),
	tf.keras.layers.Flatten(),
	tf.keras.layers.Dense(512, activation='relu'),
	tf.keras.layers.Dense(1, activation='sigmoid')
	])

	from tensorflow.keras.optimizers import RMSprop

	model.compile(loss='binary_crossentropy',
	optimizer=RMSprop(lr=0.001),
	metrics=['accuracy'])

	# Label image data base on it's directory name
	from tensorflow.keras.preprocessing.image import ImageDataGenerator

	train_datagen = ImageDataGenerator(1./255)

	train_generator = train_datagen.flow_from_directory(
	'/tmp/h-or-s', # Data directory
	target_size=(300, 300), # Output size
	batch_size=128, # Size of the batches of data, each batch is 8 data since the training data are 80 images
	class_mode='binary' # Two class: Happy, Bad
	)

	# Train the model
	history = model.fit(train_generator,
	steps_per_epoch = 8, # Load 8 data each epoch to parallize the training stage
	epochs=15,
	callbacks=callbacks
	)

view raw gistfile1.txt hosted with ❤ by GitHub

2020年5月20日星期三

Conference note: Making C Less Dangerous in the Linux kernel - Kees Cook

In this conference discuss below topic about unsafely C usage, and how Linux kernel has remove or add facility to detect such condition.

1. Variable Length array

Using compiler option to Detect VLA: gcc -W vla
Using guard page tor prevent stack overflow. VLA is needed lots of instruction compared to the fixed-size array.

2. Switch case break or non-break

Mark all non-breaks with a “fall through” to whether programmer intent to fall through or it's a bug.
Compiler support this feature: -Wimplicit-fallthrough

3. Arithmetic overflow detection

Using compiler option to detection overflow in compile time
Support different warning label: ignore or take as warring

4. Compare different API for string copy

The safer string copy function: strscpy().

5. Safe stack - Shadow stack

Separate the local variable stack and return address stack
Support by hardware:

ARM pointer authentication (Sign the return address for distinguish between a local variable and return address

Reference :

1. Making C Less Dangerous in the Linux kernel - Kees Cook

2020年5月18日星期一

Conference note: Safety vs Security: A Tale of Two Updates - Jeremy Rosen, Smile.fr

Safety and Security are different aspects of a system, briefly discuss the differences between them and challenge to build a system with both concerns.

Safety: Proof and make it simple for reliable

Completely define spec.
Proven: Check every state of the system is meet the design.
Change as less as possible: If a bug can solve by period reboot machine, just period reboot machine rather than update the software.

Security: Prevent system been using in unwanted way

Fasting iteration: Change the old encryption algorithm to a new one.
Preventive: Need update the spec to face the new challenge.

These two aspects often neglect in embedded system nowadays, especially in customer market. but with more and more device connected to the internet and responsible for critical tasks like healthcare. Designer need to take care of these two fields, Include them in the beginning of designing a system and try to make an optimized combination of them to meet the system standards.

Reference :

1. Safety vs Security: A Tale of Two Updates - Jeremy Rosen, Smile.fr

2020年5月17日星期日

Machine Learning Foundations: Exercise 3 Improve accuracy of MNIST using Convolutions.

Exercise 3: Improve accuracy of MNIST using Convolutions:code lab link

Improve MNIST to 99.8% accuracy or more using only a single convolutional layer and a single MaxPooling 2D.
The filter amount will affect the accuracy and training time.

Code:

	import tensorflow as tf

	# Callback function to check model accuracy
	class RayCallback(tf.keras.callbacks.Callback):
	def on_epoch_end(self, epoch, logs={}):
	if(logs.get('accuracy')>0.998):
	print("\nReached 99.8% accuracy so cancelling training!")
	self.model.stop_training = True

	# Load the MNIST handwrite digit data set
	mnist = tf.keras.datasets.mnist
	(training_images, training_labels), (test_images, test_labels) = mnist.load_data()

	# Reshap and normalize training data and callback function
	callbacks = RayCallback()
	training_images=training_images.reshape(60000, 28, 28, 1)
	training_images = training_images/255.0

	# Create an 5 layer model:
	# Convolution 16 filter with 3 X 3 size to each Image -> Polling each Image to 1/4
	# Flatten -> 128 input -> 10 output
	model = tf.keras.models.Sequential([
	tf.keras.layers.Conv2D(16, (3,3), activation='relu', input_shape=(28, 28, 1)),
	tf.keras.layers.MaxPooling2D(2, 2),
	tf.keras.layers.Flatten(),
	tf.keras.layers.Dense(128, activation='relu'),
	tf.keras.layers.Dense(10, activation='softmax')
	])

	# Setting optimizer and loss function
	model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])

	# Training model untill accuracy > 99.8%
	model.fit(training_images, training_labels, epochs=10, callbacks=[callbacks])

view raw gistfile1.txt hosted with ❤ by GitHub

2020年5月16日星期六

GO: Fixed warning: go env -w GOPATH=... does not override conflicting OS environment variable

Because the GOPATH is OS level setting variable so it can't be set using go env -v cmd.

We need to use OS specify way to modify this PATH variable.

1. Using CMD to set GOPATH variable (Environment: WIN10)

> setx GOPATH ##YOURPATH##

2. Check the GOPATH been setup

> go env

Reference

1. go 1.13 go env -w 設置GOPROXY報錯

2. GO/env_write.txt

3.https://github.com/golang/go/wiki/SettingGOPATH#windows-10-gui

2020年5月13日星期三

Conference note: Linux I2C in the 21st Century - Wolfram Sang, Consultant / Renesas

This conference is intent to brief overview what't new in Linux I2C subsystem.
I found some interesting part of the I2C subsystem of the latest linux kernel:

The API i2c_dummy_device() for I2C device have more than one slave address.

Declare a dummy I2C deivce share same device but different slave address

Recommend new API to create I2C device: i2c_new_ancillary_device()

Dynamic address assign : In same I2C bus and dynamic detect what address should use.

I2C has been widely used in industry for decades and has some enhance feature like have multi slave address or dynamic assign address of slave in the same bus,

bring lots of challenge for developer and Linux kernel provide some new API for more generic driver development.

Further speaking:

I3C is the next generation serial bus to replace I2C but I didn't saw much application right now.

Reference :

Linux I2C in the 21st Century - Wolfram Sang, Consultant / Renesas
I2C and SMBus Subsystem
I3C

2020年5月12日星期二

Machine Learning Foundations: Exercise 2 Handwriting digit model with 99% accuracy

Exercise 2: Handwriting digit model with 99% accuracy:code lab link

Write an MNIST classifier that trains to 99% accuracy or above, and does it without a fixed number of epochs -- i.e. you should stop training once you reach that level of accuracy.

Code:

	import tensorflow as tf

	# Callback function to check model accuracy
	class RayCallback(tf.keras.callbacks.Callback):
	def on_epoch_end(self, epoch, logs={}):
	if(logs.get('accuracy')>0.99):
	print("\nReached 99% accuracy so cancelling training!")
	self.model.stop_training = True

	# Load the MNIST handwrite digit data set
	mnist = tf.keras.datasets.mnist
	(x_train, y_train),(x_test, y_test) = mnist.load_data()

	# Normalize training data and callback function
	callbacks = RayCallback()
	x_train = x_train/255.0
	x_test = x_test/255.0

	# Create an 3 layer model: Flatten -> 128 input -> 10 output
	model = tf.keras.models.Sequential([
	tf.keras.layers.Flatten(),
	tf.keras.layers.Dense(128, activation=tf.nn.relu),
	tf.keras.layers.Dense(10, activation=tf.nn.softmax)
	])

	# Setting optimizer and loss function
	model.compile(optimizer='adam',
	loss='sparse_categorical_crossentropy',
	metrics=['accuracy'])

	# Training model untill accuracy > 99%
	model.fit(x_train, y_train, epochs=15, callbacks=[callbacks])

	# Evaluate with test data
	model.evaluate(x_test, y_test)

view raw gistfile1.txt hosted with ❤ by GitHub

Result:

	Epoch 1/15
	1875/1875 [==============================] - 3s 2ms/step - loss: 0.2570 - accuracy: 0.9265
	Epoch 2/15
	1875/1875 [==============================] - 4s 2ms/step - loss: 0.1133 - accuracy: 0.9667
	Epoch 3/15
	1875/1875 [==============================] - 3s 2ms/step - loss: 0.0778 - accuracy: 0.9765
	Epoch 4/15
	1875/1875 [==============================] - 3s 2ms/step - loss: 0.0580 - accuracy: 0.9822
	Epoch 5/15
	1875/1875 [==============================] - 3s 2ms/step - loss: 0.0444 - accuracy: 0.9859
	Epoch 6/15
	1875/1875 [==============================] - 3s 2ms/step - loss: 0.0345 - accuracy: 0.9893
	Epoch 7/15
	1863/1875 [============================>.] - ETA: 0s - loss: 0.0268 - accuracy: 0.9916
	Reached 99% accuracy so cancelling training!
	1875/1875 [==============================] - 3s 2ms/step - loss: 0.0271 - accuracy: 0.9916
	313/313 [==============================] - 0s 1ms/step - loss: 0.0843 - accuracy: 0.9774
	[0.08431357890367508, 0.977400004863739]

view raw gistfile1.txt hosted with ❤ by GitHub

Machine Learning Foundations : Exercise 1 House Price Question

Exercise 1 : House Prices Question : code lab link

Build a neural network that predicts the price of a house according to a simple formula, house pricing was as easy as a house costs 50k + 50k per bedroom, so that a 1 bedroom house costs 100k, a 2 bedroom house costs 150k etc.

Training data set

Bedroom amount [1, 2, 3, 4,]
House price [100, 150, 200, 250]

Code:

	import tensorflow as tf
	iport numpy as np
	from tensorflow import keras
	# Create an 1*1 layer neuron
	model = tf.keras.Sequential([keras.layers.Dense(units=1, input_shape=[1])])

	# Setting optimizer and loss function
	model.compile(optimizer='sgd', loss='mean_squared_error')

	# Training data
	xs = np.array([1, 2, 3, 4], dtype=int)
	ys = np.array([100, 150, 200, 250], dtype=int)

	# Training mode for 500 iteration
	model.fit(xs, ys, epochs=500)

	# Predict the output
	print(model.predict([7.0]))m

view raw gistfile1.txt hosted with ❤ by GitHub

2020年5月6日星期三

Octave: Read data from csv file.

Octave/Matlab can read CSV file and do some processing or plot the data quickly and easily.

Below are simple codes that read a CSV file and make some parse.

1. Read Sample_Data.csv file and stored as char array:

    char = strsplit(fileread ("Sample_Data.csv"));

2. Simple parse: Iterate each element and skip first 5 column data

    for i = 1:size(char,2)
        result(i,:) = char((10*i)-5:(10*i));
    end

Ref:

1. Octave - strsplit

2020年3月11日星期三

How to get main() function return value in Linux terminal.

Sometimes we want to get the exit value of main() function, using the "echo $? " can get the most recently program exit value.

Example program:

	#include <stdio.h>

	int main(void)
	{



	printf("Hello world\r\n");

	return 0xff;

	}

view raw gistfile1.txt hosted with ❤ by GitHub

Result:

Reference:
1. UNIX Shell Programming

訂閱：文章 (Atom)

	{'<OOV>': 1, 'a': 2, 'have': 3, 'i': 4, 'car': 5, 'pen': 6, 'bike': 7, 'he': 8, 'apple': 9, 'and': 10, 'banna': 11}

	Word Index = {'<OOV>': 1, 'a': 2, 'have': 3, 'i': 4, 'car': 5, 'pen': 6, 'bike': 7, 'he': 8, 'apple': 9, 'and': 10, 'banna': 11}

	Sequences = [[4, 3, 2, 5], [4, 3, 2, 6], [4, 3, 2, 7], [8, 3, 2, 9, 10, 2, 11]]

	Padded Sequences:
	[[ 0 4 3 2 5]
	[ 0 4 3 2 6]
	[ 0 4 3 2 7]
	[ 2 9 10 2 11]]

	Test Sequence = [[4, 3, 2, 1, 5], [8, 1, 11]]

	import wave
	import numpy as np
	data = "audio/8455-210777-0068.wav" # your power is sufficient i said sound data

	# using wave lib to read wav file
	wf = wave.open(data, 'rb')

	frames = wf.getnframes()
	pcm_data = wf.readframes(frames)

	wf.close()

	# transfer audio data to int16 type
	audio = np.frombuffer(pcm_data, dtype=np.int16)

	from deepspeech import Model

	# load pre-trained model
	ds = Model("./deepspeech-0.7.0-models.pbmm")

	# do inference
	output = ds.stt(audio)

	# print inference text data
	print(output)

2020年12月27日 星期日

2020年11月13日 星期五

2020年6月22日 星期一

Using the Sequencing APIs

Code

Result:

2020年6月17日 星期三

Tokenization for Natural Language Processing

Code:

Result:

Reference:

2020年5月28日 星期四

Goal

Code

Makefile

Usage

Reference

Goal

Code:

Makefile:

Usage:

Reference:

2020年5月27日 星期三

Overview

Install DeepSpeech

Demo

Using command line tool to inference sound data.

2020年5月22日 星期五

Machine Learning Foundations: Exercise 4 Happy and sad image classify model with 99.9% accuracy:code lab link

Code:

2020年5月20日 星期三

2020年5月18日 星期一

Safety and Security are different aspects of a system, briefly discuss the differences between them and challenge to build a system with both concerns.

Safety: Proof and make it simple for reliable

Security: Prevent system been using in unwanted way

2020年5月17日 星期日

Exercise 3: Improve accuracy of MNIST using Convolutions:code lab link

Code:

2020年5月16日 星期六

2020年5月13日 星期三

This conference is intent to brief overview what't new in Linux I2C subsystem.I found some interesting part of the I2C subsystem of the latest linux kernel:

The API i2c_dummy_device() for I2C device have more than one slave address.

Recommend new API to create I2C device: i2c_new_ancillary_device()

Dynamic address assign : In same I2C bus and dynamic detect what address should use.

2020年5月12日 星期二

2020年5月6日 星期三

2020年3月11日 星期三

2020年12月27日星期日

2020年11月13日星期五

2020年6月22日星期一

2020年6月17日星期三

2020年5月28日星期四

2020年5月27日星期三

2020年5月22日星期五

2020年5月20日星期三

2020年5月18日星期一

2020年5月17日星期日

2020年5月16日星期六

2020年5月13日星期三

This conference is intent to brief overview what't new in Linux I2C subsystem.
I found some interesting part of the I2C subsystem of the latest linux kernel:

2020年5月12日星期二

2020年5月6日星期三

2020年3月11日星期三