DEV Community

Cover image for Machine Learning : Things you don't know yet !!!
Rishabh Jain
Rishabh Jain

Posted on

1 1 1 1 1

Machine Learning : Things you don't know yet !!!

  1. Machine Learning, although not a recent innovation, has been circulating since the 1950s. Its prominence surged notably in the 1990s with the introduction of programming languages like Python and R. Within the realm of Artificial Intelligence, Machine Learning constitutes a vital subdomain, with Deep Learning nestled as a subset therein.

  2. While Data Analytics and Data Cleaning are perennial components of machine learning endeavours, the selection of an appropriate model holds equal significance, ensuring anticipated outcomes without unwelcome surprises.
    Different Machine Learning Models and their usages. Image Credit:DataScience

  3. The paramount importance of Data Quality cannot be overstated. Regardless of the sophistication of the employed model, erroneous outcomes and pronounced biases ensue if the data quality is compromised. The adage "garbage in, garbage out" remains pertinent.

  4. At times, simplicity triumphs over complexity in predictive accuracy, contingent upon the nature of the data and the predictive task at hand.

  5. Human biases can insidiously infiltrate models, as datasets are crafted by individuals and may harbour inherent biases, as discernible in less-refined Image Generation Systems.
    Human Biases in various sectors used by Machine Learning, Image Credit: Lightly.ai

  6. Model performance is subject to degradation over time. Hence, regular model training and diligent data updates are imperative to sustain optimal performance.

  7. Ensemble Models are a potent yet underutilized tool among novices. Mastery of diverse models empowers practitioners to leverage techniques like bagging, boosting, and stacking, substantially augmenting performance.
    Ensemble Methods

  8. Even meticulously trained and rigorously tested ML models can err due to inadequate data provision. ML models crave copious data inputs for enhanced accuracy.

  9. Ethical considerations demand meticulous attention. Given the meteoric advancement of AI tools and the proliferation of AI-Software as a Service (SaaS) applications, ethical guidelines must be rigorously upheld. Every technological innovation harbors both benefits and drawbacks, and adherence to regulatory frameworks is pivotal in mitigating adverse consequences.

Image of Timescale

🚀 pgai Vectorizer: SQLAlchemy and LiteLLM Make Vector Search Simple

We built pgai Vectorizer to simplify embedding management for AI applications—without needing a separate database or complex infrastructure. Since launch, developers have created over 3,000 vectorizers on Timescale Cloud, with many more self-hosted.

Read full post →

Top comments (0)

Image of Docusign

🛠️ Bring your solution into Docusign. Reach over 1.6M customers.

Docusign is now extensible. Overcome challenges with disconnected products and inaccessible data by bringing your solutions into Docusign and publishing to 1.6M customers in the App Center.

Learn more