What is Machine Learning?
What do Netflix movie recommendations, Amazon product recommendations, and Siri have in common? They all use machine learning to identify patterns and make decisions. So, what exactly is machine learning?
Machine learning (ML) is an application of artificial intelligence (AI) in which a machine/computer learns from massive amounts of data to make predictions. The computer performs automated data sequencing, combing through data in search of patterns. Once data patterns and predictive behaviors have been identified, the algorithm applies certain rules, which it either creates or modifies, to accomplish its objectives.
The power of ML algorithms lies in their ability to “learn” from past data and make predictions in much the same way humans learn from past experiences to make decisions. The ability to learn from past data is where ML differs from statistical models. Statistical models, such as linear regression, can assess the relationships between variables and can often be used to make predictions. However, statistical models don’t use training data to learn, and predictive accuracy is not their strength.
ML can be used for a variety of objectives. It can make transportation routes faster and more efficient; improve sales conversions by making recommendations or tailoring product-related content; improve patient diagnostics and predict hospitalization rates based on patient data; protect computer networks from cyberthreats; and determine levels of investment or insurance risk.
Conditions for Machine Learning
A few conditions are necessary for ML:
- The relevant data must be consolidated and accessible. Data includes text, images, computer clicks and social media postings.
- An algorithm must be developed to analyze the large amounts of data for similarities and patterns. Enough computational processing power must also be available for the program to comb through vast amounts of data.
- A basic set of rules must be provided to the machine to serve as guidelines.
Trust, Fear and Machine Learning
Both trust and fear are associated with ML today. Trust that the machine will make fitting recommendations or take the appropriate actions is necessary for us to adopt self-driving vehicles or investment suggestions. Fear stems from the idea of technological singularity, a hypothetical point in the future when machine intelligence will exceed that of humans, with unforeseeable results for human civilization.
The main advantage of ML algorithms over humans today is their computational capacity and speed. Computers can analyze massive amounts of data and recognize patterns much more quickly than humans ever will. Humans, however, are still the ones defining and setting the rules the algorithms follow while analyzing and learning from past data to make decisions or recommendations. The objectives of the algorithms, too, are set by humans, and it’s up to us to ensure the algorithms are used in ways that promote rather than endanger human values.
Machine Learning in Cybersecurity
Specific to cybersecurity, it is difficult to keep pace with the volume and increasing sophistication of threats and cyberattacks today. According to Cybersecurity Ventures, cybercrime is predicted to inflict US$6 trillion in damages globally in 2021 and reach $10.5 trillion annually by 2025. Automated tools have made it easier for attackers to execute successful attacks.
Palo Alto Networks takes a unique approach to cybersecurity that incorporates automation and ML to get ahead of attackers. Machine learning can help identify variations of known threats and patterns, predict the next steps of an attack, and automatically create and implement protections across the organization in near-real time.
Want to learn how Palo Alto Networks is leveraging ML to protect enterprises from tomorrow’s threats? Read our e-book 4 Key Elements of an ML-Powered NGFW. You can also watch our on-demand launch event to learn how Palo Alto Networks is delivering intelligent network security with the world’s first ML-Powered NGFW, PAN-OS® 10.0 and more than 70 innovative capabilities.