Machine Learning: Difference between revisions

Latest revision as of 09:43, 6 July 2025

Machine Learning is a subset of artificial intelligence (AI) that involves the development of algorithms and statistical models that enable computers to perform tasks without explicit instructions. Instead, these systems learn from data inputs and adapt through experience, thereby improving their performance in tasks over time. Machine Learning has a wide array of applications, ranging from natural language processing to computer vision, and has fundamentally transformed industries such as finance, healthcare, and transportation.

History

The roots of machine learning can be traced back to the inception of artificial intelligence in the mid-20th century. Early artificial intelligence research focused predominantly on symbolic approaches, where knowledge was explicitly programmed into systems. However, the limitations of these methods became evident, prompting researchers to explore alternative approaches.

The Beginnings

In the 1950s and 1960s, pioneering work by scientists such as Alan Turing and Marvin Minsky began to lay the groundwork for machine learning. The Turing Test, proposed by Turing, evaluated a machine’s ability to exhibit intelligent behavior indistinguishable from that of a human. During this period, Minsky and Seymour Papert published insights into neural networks, albeit with limited success due to computational constraints.

Growth and Decline

By the 1980s, interest in machine learning awakened with the re-discovery of backpropagation algorithms to train multi-layer neural networks, allowing for more complex functions to be learned. However, the rise of expert systems—AI that relied heavily on predefined rules—over machine learning techniques led to the "AI winter," a period characterized by reduced funding and interest in AI research from the late 1970s to the early 1990s.

The Resurgence

The early 21st century witnessed a resurgence in machine learning prompted by the advent of big data and advances in computational power. The rise of the internet provided vast quantities of data, enabling the training of more sophisticated models. In addition, increased interest in algorithms like Support Vector Machines (SVMs) and decision trees led to breakthroughs in supervised learning capabilities. This period is often referred to as the "deep learning revolution," where algorithms based on neural networks achieved remarkable successes in tasks such as speech recognition and image classification.

Key Concepts

Machine learning encompasses various techniques and methodologies. Understanding these concepts is essential for comprehending how machine learning systems operate.

Types of Machine Learning

Machine learning can be broadly categorized into three types: supervised learning, unsupervised learning, and reinforcement learning.

In supervised learning, the model is trained on a labeled dataset, which contains input-output pairs. The objective is for the model to learn to map inputs to the correct outputs so that it can make accurate predictions on unseen data. Common algorithms used include linear regression, logistic regression, decision trees, and neural networks.

Unsupervised learning, in contrast, works with unlabeled data. The goal is to identify patterns or structures within the dataset. Applications include clustering, where the system groups similar data points, and dimensionality reduction, where techniques like Principal Component Analysis (PCA) are used to simplify datasets by reducing the number of features.

Reinforcement learning (RL) is a distinct area that focuses on training agents to make sequential decisions by rewarding desirable outcomes and penalizing undesired ones. RL has gained traction for applications such as robotics, game playing, and autonomous driving.

Algorithms and Techniques

Machine learning leverages a range of algorithms, each suited to different types of tasks and data. Some of the most widely used algorithms include decision trees, neural networks, support vector machines, random forests, and k-nearest neighbors (KNN).

Decision trees create a model based on a series of questions that split the data into branches. This model is intuitive and easy to interpret, making it suitable for various applications. Neural networks, inspired by the human brain, consist of interconnected nodes (neurons) that process inputs through multiple layers. Their ability to model complex relationships has led to significant advancements in areas such as image and speech recognition.

Support vector machines are another powerful algorithm that performs classification by finding the optimal hyperplane that separates data points of different classes. Random forests build multiple decision trees and aggregate their results to improve prediction accuracy and reduce overfitting.

Evaluation Metrics

Evaluating machine learning models is critical to determine their effectiveness. Common metrics vary based on whether the task is classification or regression. For classification tasks, accuracy, precision, recall, F1 score, and area under the receiver operating characteristic curve (AUC-ROC) are commonly used. In regression tasks, evaluation metrics such as mean absolute error (MAE), mean squared error (MSE), and R-squared provide insights into model performance.

Implementation in Various Domains

Machine learning finds applications across numerous fields, enhancing processes and enabling new possibilities.

Healthcare

In healthcare, machine learning algorithms assist in predictive analytics, aiding doctors in diagnosing diseases based on patient data. For example, algorithms have shown success in identifying patterns in medical imaging, such as detecting tumors in radiographs or analyzing pathology reports. Moreover, natural language processing applications allow for the interpretation of unstructured data from medical records, streamlining patient care.

Finance

The finance sector utilizes machine learning for credit scoring, fraud detection, algorithmic trading, and risk management. By analyzing transaction patterns, institutions can identify anomalies that may indicate fraudulent activity. In addition, machine learning models are employed to forecast stock trends and optimize trading strategies based on patterns in historical data.

Transportation

In transportation, particularly in the development of autonomous vehicles, machine learning plays a crucial role in enabling vehicles to navigate real-world environments. Algorithms process data from sensors like cameras and LiDAR, helping vehicles understand their surroundings and make informed decisions. This technology is also used in optimizing traffic flow through smart traffic management systems.

Retail

Retail organizations employ machine learning to analyze consumer behavior and personalize marketing strategies. Predictive analytics assist in inventory management, ensuring popular products remain in stock while minimizing excess inventory. Machine learning algorithms enable recommendation systems that enhance the customer shopping experience by suggesting relevant products.

Real-world Examples

The impact of machine learning can be observed in various real-world applications, highlighting its transformative potential.

Natural Language Processing

Natural language processing (NLP) techniques have led to advancements in virtual assistants such as Siri and Alexa, allowing for conversational interfaces that understand and respond to human queries. Sentiment analysis in social media monitoring tools employs machine learning to gauge public opinion and brand perception.

Image and Video Analysis

Machine learning powers many image and video analysis applications, such as facial recognition technology used in security systems and social media platforms. Companies like Facebook and Google utilize machine learning algorithms to tag photos automatically and enhance user experience.

Fraud Detection

Financial institutions implement machine learning algorithms to analyze transaction data in real-time for signs of fraudulent behavior. These models improve over time, adapting to new fraud patterns and constantly enhancing security measures without requiring manual updates.

Autonomous Vehicles

Companies such as Tesla and Waymo rely on machine learning to drive innovations in autonomous vehicle technology. The ability of vehicles to process immense amounts of real-time data allows for dynamic decision-making, significantly enhancing safety and efficiency on roads.

Criticism and Limitations

Despite its advancements, machine learning is not without criticisms and limitations that merit consideration.

Data Bias

One of the primary concerns in machine learning is the potential for biased outcomes due to biased training data. If the data used to train machine learning models reflects societal biases, the algorithms may perpetuate or even exacerbate these biases when making decisions, particularly in fields such as hiring, law enforcement, and lending.

Interpretability

Many complex machine learning models, particularly deep learning networks, often function as "black boxes." This lack of transparency can pose challenges in understanding how decisions are made, leading to difficulties in accountability and trust. Stakeholders may be hesitant to adopt these technologies if they cannot ascertain the rationale behind specific outputs.

Overfitting

Overfitting occurs when a model learns to perform exceedingly well on training data but fails to generalize to new, unseen data. This issue can stem from models that are too complex relative to the amount of training data available. Techniques such as cross-validation and regularization are commonly employed to mitigate this risk.

Ethical Considerations

As machine learning systems become more integrated into everyday life, ethical considerations arise regarding privacy, consent, and surveillance. The potential for misuse of data and algorithmic decision-making necessitates ongoing discussions surrounding regulatory frameworks and ethical standards to guide machine learning deployment.

Future Directions

Looking forward, the future of machine learning is promising, with numerous advancements anticipated across various domains.

Explainable AI

Research into explainable AI (XAI) seeks to address issues of interpretability, striving to make machine learning models more understandable to humans. Developing techniques that clarify how models arrive at decisions will enhance trust and facilitate broader adoption in sensitive areas such as healthcare and law.

Integration of Multimodal Data

Future machine learning applications may increasingly involve integrating multimodal data—combining visual, textual, and auditory information to drive more holistic understanding and decision-making. Such advancements could lead to enhanced customer experiences in consumer-facing industries and more refined analytical capabilities in research.

Open-source Collaboration

The open-source movement within machine learning fosters collaboration and the democratization of technology. As advances in models and frameworks such as TensorFlow and PyTorch become widely accessible, it enables organizations across different sectors to harness machine learning capabilities, thereby accelerating innovation.

References

@@ Line 1: / Line 1: @@
-== Introduction ==
+'''Machine Learning''' is a subset of artificial intelligence (AI) that involves the development of algorithms and statistical models that enable computers to perform tasks without explicit instructions. Instead, these systems learn from data inputs and adapt through experience, thereby improving their performance in tasks over time. Machine Learning has a wide array of applications, ranging from natural language processing to computer vision, and has fundamentally transformed industries such as finance, healthcare, and transportation.
-Machine learning (ML) is a subset of artificial intelligence (AI) that enables systems to learn from data and improve their performance on specific tasks without being explicitly programmed. It employs algorithms that identify patterns in data and make predictions based on those patterns. Machine learning intersects multiple disciplines, including statistics, computer science, and cognitive psychology, drawing from each to develop sophisticated methodologies for automating decision-making processes.
-Machine learning is generally categorized into three main types: supervised learning, unsupervised learning, and reinforcement learning. In supervised learning, models are trained on labeled datasets, which means that the input data is paired with its corresponding output. In unsupervised learning, algorithms identify patterns in data without prior labeling, discovering hidden structures or groupings. Reinforcement learning, on the other hand, involves training agents to make sequences of decisions by rewarding them for performing correctly and penalizing them for errors.
+== History ==
+The roots of machine learning can be traced back to the inception of artificial intelligence in the mid-20th century. Early artificial intelligence research focused predominantly on symbolic approaches, where knowledge was explicitly programmed into systems. However, the limitations of these methods became evident, prompting researchers to explore alternative approaches.
-As machine learning continues to evolve, it plays an increasingly significant role in various industries, enabling advancements in fields such as healthcare, finance, transportation, and more. Its applications range from predicting stock prices to diagnosing diseases, showcasing its versatile capabilities and growing importance in both academic research and industrial applications.
+=== The Beginnings ===
+In the 1950s and 1960s, pioneering work by scientists such as Alan Turing and Marvin Minsky began to lay the groundwork for machine learning. The Turing Test, proposed by Turing, evaluated a machine’s ability to exhibit intelligent behavior indistinguishable from that of a human. During this period, Minsky and Seymour Papert published insights into neural networks, albeit with limited success due to computational constraints.
-== History or Background ==
+=== Growth and Decline ===
-The roots of machine learning can be traced back to the 1950s, when computer scientists began exploring the concept of artificial intelligence. Pioneers like Alan Turing and John McCarthy were instrumental in laying the groundwork for AI, but it was not until the 1960s and 1970s that the term "machine learning" gained traction. Early work in this field focused on simple algorithms and rule-based systems, with limited success in practical applications.
+By the 1980s, interest in machine learning awakened with the re-discovery of backpropagation algorithms to train multi-layer neural networks, allowing for more complex functions to be learned. However, the rise of expert systems—AI that relied heavily on predefined rules—over machine learning techniques led to the "AI winter," a period characterized by reduced funding and interest in AI research from the late 1970s to the early 1990s.
-During the 1980s, the introduction of new computational techniques and increased processing power led to the revival of neural networks, a foundational concept in machine learning. Researchers such as Geoffrey Hinton and David Rumelhart developed backpropagation algorithms, which enabled multi-layer networks to learn complex patterns from data. This era saw the emergence of "expert systems," which aimed to replicate human decision-making in specific domains, though these systems often struggled with adaptability and scalability.
+=== The Resurgence ===
+The early 21st century witnessed a resurgence in machine learning prompted by the advent of big data and advances in computational power. The rise of the internet provided vast quantities of data, enabling the training of more sophisticated models. In addition, increased interest in algorithms like Support Vector Machines (SVMs) and decision trees led to breakthroughs in supervised learning capabilities. This period is often referred to as the "deep learning revolution," where algorithms based on neural networks achieved remarkable successes in tasks such as speech recognition and image classification.
-In the late 1990s and 2000s, machine learning began gaining mainstream attention, fueled by exponential growth in data generation and advancements in computing power, particularly with the advent of graphics processing units (GPUs). Significant progress in areas such as natural language processing (NLP) and computer vision further propelled machine learning into public consciousness. The rise of large datasets, epitomized by the availability of the Internet, enabled the training of more accurate and sophisticated models.
+== Key Concepts ==
+Machine learning encompasses various techniques and methodologies. Understanding these concepts is essential for comprehending how machine learning systems operate.
-From 2010 onwards, machine learning has experienced a meteoric rise, catalyzed by breakthroughs in deep learning, a subfield focused on neural networks with many layers (deep neural networks). Algorithms such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) have transformed fields like image and speech recognition, enabling machines to achieve superhuman performance in specific tasks.
+=== Types of Machine Learning ===
+Machine learning can be broadly categorized into three types: supervised learning, unsupervised learning, and reinforcement learning.
-== Technical Details or Architecture ==
+In supervised learning, the model is trained on a labeled dataset, which contains input-output pairs. The objective is for the model to learn to map inputs to the correct outputs so that it can make accurate predictions on unseen data. Common algorithms used include linear regression, logistic regression, decision trees, and neural networks.
-Machine learning systems typically consist of three key components: the data, the learning algorithm, and the model. The data serves as the fundamental input, containing the features and target variables used to train the model. The selection and preprocessing of this data are crucial, as they directly impact the performance of the machine learning algorithms.
-=== Types of Machine Learning ===
+Unsupervised learning, in contrast, works with unlabeled data. The goal is to identify patterns or structures within the dataset. Applications include clustering, where the system groups similar data points, and dimensionality reduction, where techniques like Principal Component Analysis (PCA) are used to simplify datasets by reducing the number of features.
-. **Supervised Learning**
-   - **Definition**: Algorithms learn from labeled datasets, where input-output pairs are provided.
-   - **Common Algorithms**:
-     - Linear Regression
-     - Decision Trees
-     - Support Vector Machines (SVM)
-     - Neural Networks
-   - **Use Cases**:
-     - Image classification
-     - Sentiment analysis
-     - Fraud detection
-. **Unsupervised Learning**
+Reinforcement learning (RL) is a distinct area that focuses on training agents to make sequential decisions by rewarding desirable outcomes and penalizing undesired ones. RL has gained traction for applications such as robotics, game playing, and autonomous driving.
-   - **Definition**: Algorithms identify patterns in data without labeled outputs.
-   - **Common Algorithms**:
-     - K-means Clustering
-     - Hierarchical Clustering
-     - Principal Component Analysis (PCA)
-   - **Use Cases**:
-     - Market segmentation
-     - Anomaly detection
-     - Topic modeling
-. **Reinforcement Learning**
+=== Algorithms and Techniques ===
-   - **Definition**: Agents learn to make decisions by performing actions and receiving feedback.
+Machine learning leverages a range of algorithms, each suited to different types of tasks and data. Some of the most widely used algorithms include decision trees, neural networks, support vector machines, random forests, and k-nearest neighbors (KNN).
-   - **Common Techniques**:
-     - Q-learning
-     - Deep Q-Networks (DQN)
-     - Policy Gradients
-   - **Use Cases**:
-     - Game playing (e.g., AlphaGo)
-     - Robotics
-     - Recommendation systems
-=== Model Training and Validation ===
+Decision trees create a model based on a series of questions that split the data into branches. This model is intuitive and easy to interpret, making it suitable for various applications. Neural networks, inspired by the human brain, consist of interconnected nodes (neurons) that process inputs through multiple layers. Their ability to model complex relationships has led to significant advancements in areas such as image and speech recognition.
-The training process involves feeding the algorithm with training data to "teach" it how to make predictions or decisions. This process can be summarized in the following steps:
-. **Data Splitting**: The dataset is typically divided into training, validation, and test sets to evaluate model performance and avoid overfitting.
+Support vector machines are another powerful algorithm that performs classification by finding the optimal hyperplane that separates data points of different classes. Random forests build multiple decision trees and aggregate their results to improve prediction accuracy and reduce overfitting.
-. **Feature Selection**: Identifying the most relevant features allows the model to focus on important attributes, improving efficiency.
-. **Hyperparameter Tuning**: Fine-tuning model parameters through grid search, random search, or Bayesian optimization enhances predictive performance.
-. **Cross-Validation**: Employing techniques such as k-fold cross-validation ensures robust evaluation of model generalization.
-Models are evaluated using metrics appropriate for the specific task, including accuracy, precision, recall, F1 score, and area under the ROC curve (AUC-ROC). The choice of metric often depends on the problem domain, particularly with imbalanced datasets.
+=== Evaluation Metrics ===
+Evaluating machine learning models is critical to determine their effectiveness. Common metrics vary based on whether the task is classification or regression. For classification tasks, accuracy, precision, recall, F1 score, and area under the receiver operating characteristic curve (AUC-ROC) are commonly used. In regression tasks, evaluation metrics such as mean absolute error (MAE), mean squared error (MSE), and R-squared provide insights into model performance.
-== Applications or Use Cases ==
+== Implementation in Various Domains ==
-Machine learning's versatility has led to its adoption in numerous sectors, enhancing efficiency and enabling novel solutions to longstanding problems.
+Machine learning finds applications across numerous fields, enhancing processes and enabling new possibilities.
 === Healthcare ===
-- **Diagnostic Systems**: ML algorithms aid in the analysis of medical imagery and pathology reports, assisting radiologists in detecting conditions like cancer at early stages.
+In healthcare, machine learning algorithms assist in predictive analytics, aiding doctors in diagnosing diseases based on patient data. For example, algorithms have shown success in identifying patterns in medical imaging, such as detecting tumors in radiographs or analyzing pathology reports. Moreover, natural language processing applications allow for the interpretation of unstructured data from medical records, streamlining patient care.
-- **Predictive Analytics**: Predicts patient outcomes, allowing healthcare providers to customize treatment plans.
-- **Drug Discovery**: Accelerates the identification and development of new medications through pattern recognition of genetic data.
 === Finance ===
-- **Algorithmic Trading**: ML models analyze vast amounts of market data to make real-time trading decisions.
+The finance sector utilizes machine learning for credit scoring, fraud detection, algorithmic trading, and risk management. By analyzing transaction patterns, institutions can identify anomalies that may indicate fraudulent activity. In addition, machine learning models are employed to forecast stock trends and optimize trading strategies based on patterns in historical data.
-- **Credit Scoring**: Enhances the evaluation of an applicant’s creditworthiness through data-driven insights.
-- **Fraud Detection**: Identifies anomalies in transaction patterns, minimizing losses from fraudulent activities.
 === Transportation ===
-- **Autonomous Vehicles**: Machine learning algorithms underpin self-driving technologies by interpreting sensor data and learning from road conditions.
+In transportation, particularly in the development of autonomous vehicles, machine learning plays a crucial role in enabling vehicles to navigate real-world environments. Algorithms process data from sensors like cameras and LiDAR, helping vehicles understand their surroundings and make informed decisions. This technology is also used in optimizing traffic flow through smart traffic management systems.
-- **Traffic Management**: Optimizes traffic signal patterns in smart cities through real-time data analysis, reducing congestion and delays.
 === Retail ===
-- **Personalization**: Recommender systems utilize machine learning to curate individual shopping experiences, boosting customer satisfaction and sales.
+Retail organizations employ machine learning to analyze consumer behavior and personalize marketing strategies. Predictive analytics assist in inventory management, ensuring popular products remain in stock while minimizing excess inventory. Machine learning algorithms enable recommendation systems that enhance the customer shopping experience by suggesting relevant products.
-- **Inventory Management**: Predictive analytics models optimize stock levels based on consumption patterns.
+== Real-world Examples ==
+The impact of machine learning can be observed in various real-world applications, highlighting its transformative potential.
+=== Natural Language Processing ===
+Natural language processing (NLP) techniques have led to advancements in virtual assistants such as [[Siri]] and [[Alexa]], allowing for conversational interfaces that understand and respond to human queries. Sentiment analysis in social media monitoring tools employs machine learning to gauge public opinion and brand perception.
+=== Image and Video Analysis ===
+Machine learning powers many image and video analysis applications, such as facial recognition technology used in security systems and social media platforms. Companies like [[Facebook]] and [[Google]] utilize machine learning algorithms to tag photos automatically and enhance user experience.
+=== Fraud Detection ===
+Financial institutions implement machine learning algorithms to analyze transaction data in real-time for signs of fraudulent behavior. These models improve over time, adapting to new fraud patterns and constantly enhancing security measures without requiring manual updates.
+=== Autonomous Vehicles ===
+Companies such as [[Tesla]] and [[Waymo]] rely on machine learning to drive innovations in autonomous vehicle technology. The ability of vehicles to process immense amounts of real-time data allows for dynamic decision-making, significantly enhancing safety and efficiency on roads.
+== Criticism and Limitations ==
+Despite its advancements, machine learning is not without criticisms and limitations that merit consideration.
+=== Data Bias ===
+One of the primary concerns in machine learning is the potential for biased outcomes due to biased training data. If the data used to train machine learning models reflects societal biases, the algorithms may perpetuate or even exacerbate these biases when making decisions, particularly in fields such as hiring, law enforcement, and lending.
+=== Interpretability ===
+Many complex machine learning models, particularly deep learning networks, often function as "black boxes." This lack of transparency can pose challenges in understanding how decisions are made, leading to difficulties in accountability and trust. Stakeholders may be hesitant to adopt these technologies if they cannot ascertain the rationale behind specific outputs.
-== Relevance in Computing or Industry ==
+=== Overfitting ===
-The transformative impact of machine learning is evident across various industries, reshaping businesses and enabling innovative applications. The integration of machine learning into business models has led to substantial improvements in operational efficiency, cost savings, and enhanced customer experiences.
+Overfitting occurs when a model learns to perform exceedingly well on training data but fails to generalize to new, unseen data. This issue can stem from models that are too complex relative to the amount of training data available. Techniques such as cross-validation and regularization are commonly employed to mitigate this risk.
-According to a report by McKinsey, over 70% of organizations reported that they are using machine learning to achieve faster decision-making capabilities. The implementation of ML solutions results in significant cost reductions, with an estimated savings of billions of dollars across sectors via automation and predictive analytics.
+=== Ethical Considerations ===
+As machine learning systems become more integrated into everyday life, ethical considerations arise regarding privacy, consent, and surveillance. The potential for misuse of data and algorithmic decision-making necessitates ongoing discussions surrounding regulatory frameworks and ethical standards to guide machine learning deployment.
-Industries such as finance, healthcare, and transportation have heavily invested in machine learning capabilities, with firms leveraging the technology to stay competitive. The increasing availability of open-source libraries, coupled with advancements in cloud computing, democratize access to powerful machine learning tools, allowing organizations of all sizes to experiment and deploy AI technologies without large-scale infrastructure investments.
+== Future Directions ==
+Looking forward, the future of machine learning is promising, with numerous advancements anticipated across various domains.
-In computation, machine learning plays a pivotal role in the evolution of natural language processing, computer vision, and robotics, expanding the frontiers of what machines can achieve. The development of frameworks such as TensorFlow, PyTorch, and Scikit-learn has significantly simplified the process for researchers and practitioners to build complex ML models.
+=== Explainable AI ===
+Research into explainable AI (XAI) seeks to address issues of interpretability, striving to make machine learning models more understandable to humans. Developing techniques that clarify how models arrive at decisions will enhance trust and facilitate broader adoption in sensitive areas such as healthcare and law.
+=== Integration of Multimodal Data ===
+Future machine learning applications may increasingly involve integrating multimodal data—combining visual, textual, and auditory information to drive more holistic understanding and decision-making. Such advancements could lead to enhanced customer experiences in consumer-facing industries and more refined analytical capabilities in research.
+=== Open-source Collaboration ===
+The open-source movement within machine learning fosters collaboration and the democratization of technology. As advances in models and frameworks such as [[TensorFlow]] and [[PyTorch]] become widely accessible, it enables organizations across different sectors to harness machine learning capabilities, thereby accelerating innovation.
 == See also ==
 * [[Artificial Intelligence]]
 * [[Deep Learning]]
-* [[Neural Networks]]
+* [[Data Mining]]
-* [[Support Vector Machine]]
 * [[Natural Language Processing]]
+* [[Reinforcement Learning]]
 * [[Computer Vision]]
-* [[Data Science]]
 == References ==
-* Alpaydin, E. (2020). "Introduction to Machine Learning." MIT Press. ISBN 9780262035613.
+* [https://www.openai.com OpenAI Official Website]
-* Mitchell, T. M. (1997). "Machine Learning." McGraw Hill. ISBN 9780070428072.
+* [https://tensorflow.org TensorFlow Official Website]
-* Russell, S., & Norvig, P. (2016). "Artificial Intelligence: A Modern Approach." Pearson. ISBN 9780134610996.
+* [https://pytorch.org PyTorch Official Website]
-* Goodfellow, I., Bengio, Y., & Courville, A. (2016). "Deep Learning." MIT Press. ISBN 9780262035613.
+* [https://www.ibm.com/cloud/learn/machine-learning IBM Cloud: Machine Learning]
-* McKinsey Global Institute. (2018). "Artificial Intelligence: The Next Digital Frontier?" Retrieved from [www.mckinsey.com]
+* [https://www.microsoft.com/en-us/research/research-area/machine-learning/ Microsoft Research: Machine Learning]
+* [https://www.nvidia.com/en-us/deep-learning-ai/ Nvidia Deep Learning AI]
 [[Category:Artificial Intelligence]]
+[[Category:Machine Learning]]
 [[Category:Computer Science]]
-[[Category:Machine Learning]]