Vu's Blog

Bài đăng

Các bài đăng gần đây

[AI] I’ve obtained a new certification: Certified Artificial Intelligence Engineer (CAIE™)

I’m happy to share that I’ve obtained a new certification: Certified Artificial Intelligence Engineer (CAIE™) from United States Artificial Intelligence Institute one of leading AI organization over the world! 📚 📖

[AI] Artificial Intelligence Overview

Artificial Intelligence Overview - Supervised Learning - UnSupervised Learning - Reinforced Learning - Overfit Problem - Natural Language Processing (NLP) - Computer Vision

[AI] Reinforcement learning: Find Optimal Policy

Two method to find optimal policy 👉 Model Dynamic 👉 Model Free

[AI] Fundamental concepts of Reinforcement Learning

Agent : The agent is the software program that learns to make intelligent decisions, such as a software program that plays chess intelligently. Environment : The environment is the world of the agent. If we continue with the chess example, a chessboard is the environment where the agent plays chess. State : A state is a position or a moment in the environment that the agent can be in. For example, all the positions on the chessboard are called states. Action : The agent interacts with the environment by performing an action and moves from one state to another, for example, moves made by chessmen are actions. Reward : A reward is a numerical value that the agent receives based on its action. Consider a reward as a point. For instance, an agent receives +1 point (reward) for a good action and -1 point (reward) for a bad action. Action space: The set of all possible actions in the environment is called the action space. The action space is called a discrete action space when our action...

[AI] Generative Adversarial Networks (GANs)

Generative Adversarial Networks (GANs): - Type of deep neural network architecture that uses unsupervised machine learning - Made up by generator and a discriminator network. Both networks train each other, while simultaneously trying to outwit each other. Generator network - Generate new data from a randomly generated vector of numbers, called a latent space. Discriminator network - Tries to differentiate between the real data and the data generated. - It can either perform multi-class classification or binary classification. Important concepts related to GANs - Important measure quality of the models use divergence (KL divergence, JS divergence...). - Nash equilibrium, which is a state that we try to achieve during training. - Objective functions: To measure the similarity. - Scoring algorithms: Calculating the accuracy of a GAN is simple. Some scoring algorithms: some scoring algorithms, some scoring algorithms, Mode Score... Problems with traini...

[Book Note] Trắc nghiệm ước mơ - John C. Maxwell

1. Khi tài năng của một người không phù hợp với những ước mơ của họ và họ không nhận ra điều đó, thì họ sẽ mãi mãi cố gắng làm việc nhưng sẽ chẳng bao giờ biết đến thành công. 2. Bạn có đang tập trung vào những việc mà bạn thích làm hay không? Bạn có đang phát triển thế mạnh của bạn hay không? Cái gì sẽ dễ đến với tôi nhưng khó đến với người khác? 3. Thành công không phải là một sự kiện, nó là một lối sống. 4. Người thành công luôn luôn đưa thời gian, nghị lực và các nguồn lực của họ và trong những thế mạnh, bởi vì họ nhận được sự đền đáp cao nhất từ đó. 5. Chúng ta thường là cái gì mà chúng ta thường làm lặp đi lặp lại. Sự hoàn hảo sau đó không phải là một hành động, mà là một thói quen. 6. Có thể thất bại chỉ vì không biết truyền đạt tầm nhìn cho người khác. 7. Mọi người làm việc vì lý do riêng của họ không phải vì lý do của tôi hay của bạn. 8. Nếu bạn muốn chinh phục ai đó theo giấc mơ của mình, hãy nói bằng thứ "Ngôn ngữ lợi ích" của họ, chứ không phải của chính bạn. ...

[AI] Text Data Analysis methologys

Text Data Analysis methologys 1. Tokenization Process of dividing text into a set of meaningful pieces. There are three methods of the nltk.tokenize: Sentence-tokenized, Word_tokenize and WordPunctTokenizer. 2. Stemming Word can appear in various forms, reduce these different forms into a common base form. Three stemmer: PorterStemmer, LancasterStemmer, and SnowballStemmer. The LANCASTER is the strictest. 3. Lemmatization: Reduce words to their base forms but more structured approach. 4. Chunking Divide the input text into pieces with no constraints, chunks do not need to be meaningful at all. 5. Bag-of-words model Dealing with text documents that consist of millions of words, converting them into numerical representations which usable for machine learning algorithms. Models each document by building a histogram of all of the words. Counts the number of occurrences of each word in the document use scikit-learn, represent documents by ignoring the word order. ...

[AI] Two Methology convert text data into data structure in NLP

2 Methology convert text data into data structure (vector and matrix) 👉 Bag of Word (BoW): Evaluation the frequency of the words in that particular document. Sentence can be represented as a vector with length would be equal to the size of vocabulary. CountVectorizer is python libarary conveniently help in building BoW model. Limitations of the BoW: work well for certain tasks or use cases with a limited vocabulary, not scale to large vocabularies efficiently. 👉 TF-IDF vectors: Approach with weigh terms, vectorizing text and extracting features out of it. TF : account how frequently a term occurs in a document. IDF : justice to terms that occur not so frequently across documents. TF-IDF is computationally fast however does not take into account co-occurrence of terms, semantics, the context associated with terms. Both method use Cosine to evaluation how similar or dissimilar text documents.

[AI] Interesting Deep Learning Algorithms

4 Interesting Deep Learning Algorithms 1. Artificial Neural Networks (ANNs) 👉 With 3 Layers - Input layer : No computation is performed, just used for passing information from the outside to the network. - Hidden layer : Deriving complex relationships between input and output. Identifies the pattern in the dataset, learning the data representation and for extracting the features. We cloud use many hidden layers, each layer is responsible for extracting important features. - Output layer: Number of neurons in the output layer is based on the type of problem we want our network to solve (Example: binary classification, number of neurons in the output layer is one). ⏩ Other Concept: - Activation functions ( ex: sigmoid, tanh, ReLU, softmax ): To introduce a non-linear transformation to learn the complex underlying patterns in the data. - Cost function: Use Optimization Algorithms (Gradient descent) to minimize cost so that Network predictions will b...