AI Definitions: Machine Learning

Machine learning (ML) - This subset of AI makes predictions or decisions based on patterns it spots in data sets. The process evolves and adapts on its own as it is exposed to new data, improving the output without explicit programming from a human. An example would be algorithms recommending ads for users, which become more tailored the longer it observes the users‘ habits (someone’s clicks, likes, time spent, etc.). Data scientists combine ML with other disciplines (like big data analytics and cloud computing) to solve real-world problems. However, the results are limited to probabilities, not absolutes. It doesn’t reveal causation. There are four types of machine learning: supervised, unsupervised, semi-supervised, and reinforcement learning. A clever computer program that simply mimics human-like behavior can be considered AI, but the computer system itself is not machine learning unless its parameters are automatically informed by data without human intervention. Video: Introduction to Machine Learning

More definitions here

37 Articles about Data Science & AI from March

The current limitations of ML and AI systems suitable for use in adversarial environments 

11 Articles about the Ethics of AI

Space Force: needs to “improve how it harnesses AI & ML tools” 

The European Space Agency plans to build a ChatGPT-style digital assistant  

The role of transformers in how chain-of-thought reasoning helps neural networks compute

AI teaching AI 

A record number of objects went into space last year

The adoption of new AI capabilities is making the benefits of geospatial intell more accessible

AI definitions: Training Data   

TimeGPT a generative pre-trained model specifically designed for predicting time-series data

AI definitions: Extractive summarization

Meet the 8 Google employees credited with inventing modern AI

SpaceX is building a network of hundreds of spy satellites  

Google DeepMind has developed an AI model that outperforms techniques for quantum circuits

Here are five ways to use LLMs on your laptop

Shrinking the distance between acquisition of data to actionable insights from multiple geospatial modalities

Build an AI application with Python in 10 easy steps 

AI Terms: Transformers  

A mathematical formula can explain how neural networks detect relevant patterns

12 Recent Articles about AI’s future

NASA's first open-source geospatial artificial intelligence foundation model for Earth observation data

Generative AI Landscape: Trends of 2024 and Beyond

Understanding the similarities and differences between data science and applied statistics

AI Terms: Abstractive summarization

How to Succeed With Predictive AI: An MIT webinar

A breakthrough in storing quantum data without the need for cryogenic cooling

How Large X Models (LXMs) can help generative AI complete new tasks 

The role of Large Language Models in enhancing the process of extractive summarization

The Role of Satellite Technology in Protecting Critical Infrastructure

Google Cloud adds vector support to all its database offerings

Google just entered the race of foundation models for time-series forecasting

How To Use A Vector Database

“Some of the ways we’re applying AI to the world of design systems”

“Generative AI can improve -- not replace -- predictive analytics”

The Dept of Defense is reportedly working with startup Scale AI to test generative AI models for military use

NGA launches National GEOINT Operations Center

The White House wants developers to abandon C and C++ over memory management security concerns

20 Data Science articles from February 2023

Five statistical paradoxes that data scientists should be aware of in order to do accurate analysis

What Pentagon leaders say they have learned from a year of battle in Ukraine:"The power of information is winning”

Software to sow doubts as you meta-analyze  

Machine learning is vulnerable to a wide variety of attacks. How the adversary can disrupt model training and even introduce backdoors

How Pandas alternatives—Polars, DuckDB, Vaex, and Modin—stack up to one of the most popular libraries in Python

Six of the most important types of machine learning algorithm 

“Big Data is real, but most people may not need to worry about it”

The ChatGPT prompts any data scientist must use

No, chatbots aren’t sentient. Here’s how their underlying technology works

5 Common Data Analytics Types Explained in Laymen’s Terms

Using the metaverse to virtually assemble and test AI war machines for the US military

Researchers discover a more flexible approach to machine learning—liquid neural nets

The evolving role of the data engineer

Top Predictive Analytics Trends in 2023

Even the pentagon Is using ChatGPT—the DoD’s used it to write a press release about a new counter-drone task force

How NGA Is integrating commercial analytic services into agency workflows

Python string matching without complex RegEx Syntax

Six python libraries especially useful to data engineers and natural language processing

Can ChatGPT write better code than Data Scientist? 

Researchers say ChatGPT can “weed out errors with sample code and fix it better than existing programs designed to do the same.”

25 Data Science Articles from Dec 2022

A Pandas DataFrame cheatsheet for exploratory analysis & data manipulation 

Five ways that data roles will change in 2023 related to Chief Data Officers

AI & machine learning are “top of mind for the Army, especially as it pertains to protecting its assets in space”

10 weird things about SpaceX's more than 3,000 Starlink satellites (and that number keeps growing)

Initial specific steps toward launching a machine learning project 

Adobe has just released a remarkable and free AI-powered enhanced speech tool

The four biggest trends they expect to shape the AI landscape in 2023

Synthetic data applications, limitations & vulnerabilities

A guide to the roles and responsibilities on a data migration team

A tech journalist goes back to high school to find out what OpenAI’s Chatbot can pass AP Lit

The current limitations of AI’s military impact & where tech could one day spark “revolutionary changes” 

How Bayesian network structure learning can incorporate missing data 

The NGA has plans to develop an overarching cloud-based enterprise management system capable of automating its data collection and dissemination and ultimately replacing the overall Foundation GEOINT storage and management process 

A new paper on “Localization and classification of space objects using EfficientDet detector for space situational awareness”

Potential uses of ChatGPT for data scientists

McKinsey on the state of AI since the research firm began tracking it five years ago

A new collaborative effort is designed to “support interoperable open map data as a shared asset that can strengthen mapping services worldwide”

Different kinds of geospatial specialists are needed in different situations

China outpaces efforts by U.S. intelligence agencies to harness power of publicly available data 

The Space Dev Agency’s first major satellite launch has been delayed again

A look under the hood: How does ChatGPT work internally? 

An AI method from MIT and IBM research “improves the training and inference performance of deep learning models on large graphs”

Some basics about the new AI called ChatGPT 

Why Neural Network explainability is important, how to do it, & the tools for it

“The FCC approved part of SpaceX’s application for the second generation of the Starlink constellation, which will allow SpaceX to deploy up to 7,500 satellites”

20 Data Science Articles from Nov 2022

3 approaches to track and reduce space debris

Big tech has a “blind spot about the severe limitations of large language models.” Meta’s Galactica is another example

A key advancement toward commercial quantum computing applications

Creating and deleting a SQL database on Azure Cloud

A dataset to train AI systems that can generate code

The NRO has released its formal request for proposals from commercial operators who can provide the spy satellite agency with hyperspectral imagery

Tracking the typical lifecycle of AI innovations in the aerospace and defense industry

Scientists call for better space weather alerts to prevent commercial space disasters

Data preparation is key to expanding military benefits of AI

The difficult search for dangerous space Junk that could potentially trigger devastating chain reactions 

The National Geospatial-Intelligence Agency plans to double spending on contracts to monitor global economic activity from space

“Our adversaries are coming after us in cyberspace” says a U.S. Space Force Col.

5 ways to use predictive insights to get the most from your data

Where is the satellite intelligence industry heading next?

15 of the little-known uses of artificial intelligence

Geopolitical threats in the space environment—debating a potential new cold war

AI creators have problems explaining how it works and determining why it has the outputs it has

Top C++ based data science and machine learning libraries

A moderately large atomic detonation could knock out satellites in low Earth orbit

7 Steps to Mastering Machine Learning with Python in 2022

27 Data Science Articles from June 2022

The priorities of the first-ever assistant secretary of the Air Force for space acquisition and integration (& top acquisition executive for the Space Force)

Google Cloud expands Earth Engine to help businesses and governments

Comparing C++ to Python (with examples)

Can synthetic data help AI get quicker results —and be less discriminatory? Here comes the fake data

OpenAI says its latest AI has learned to play Minecraft

US intelligence artificial intelligence use is booming but it's not the secret weapon you might imagine

“A major challenge facing the DoD at the moment is disparate data, spread across many different databases and stakeholders. Future winners will be those that can take all the data into a single location and make sense of it.”

“AI solutions for defense are much more mundane and focused on improving decision-making for humans” than many would imagine”

Space 2.0: “The shape of space is expanding beyond traditional defense & aerospace to an expansive range of practical & profitable applications.” A look at the 2022 trends

China launches first crewless drone carrier—experts suggest that it could also be used as a military vessel  

Space-based assets aren’t immune to cyberattacks: Russia's attack on Viasat satellites exposed how vulnerable space-based assets are and the potential for spillover damage

Which is better for data science visualization—R or Python? (hint: it all depends on the nature of the problem to be solved) 

Overcoming overfitting a model in machine learning

How space debris threatens modern life  

Ranking Pandas for Python, Dask & Datatable based on their performance

Snowflake ups support for python Build and offers Native Application Framework to run applications inside the Snowflake Data Cloud platform

Pentagon’s new AI and data chief waited days just for an ID card: ‘Let me say honestly that the bureaucracy is real’

The basic process of handling satellite image data for geospatial deep learning

6 Types of “feature importance” — a useful (and yet slippery) machine learning concept

Google Cloud’s new machine learning tools for its Vertex AI are now making their debut after being featured at the recent Applied ML Summit

The remarkable story of deploying the satellite communication system Starlink in Ukraine

Creating a simple, interactive dashboard with Panel & Python

Wanted: artificial intelligence & machine autonomy algorithms for military command and control

A visual breakdown of threats to space-based services such as Starlink & GPS

Google won’t allow people to create deepfakes using its collaborative machine learning platform any longer

“Python may be the second choice to R, but its popularity and ease of use positions it to dominate data science” 

Top YouTube channels for learning data science

Some basic data cleaning issues and possible solutions

Daily Data Science stories here.

24 Data Science Articles from May 2022

Spark or Hadoop? Both Apache products can be used by data scientists but which is the better analytics tool? Here’s a comparison—along with which one will fit better based on your project focus

Fixing data lake errors can be time-consuming, and costly—here are some thoughts on standardized an autonomous validation approach to avoid the lake becoming a swamp

Interpol: in a couple of years expect state-developed cyber weapons to be available on the dark net  

Do you think Python is slow? Here’s a fast way to loop in Python

Looking for patterns in satellite image time series with python? Here’s a quick guide for handling time-varying imagery with open python libraries 

Can the new-and-improved Large Hadron Collider save particle physics?   

Want to run Python code in a browser? Soon you might be able to 

The AI Engineering Process: A guide to solving an AI problem

The challenges of organizing geospatial intelligence efficiently 

Making predictions outperforms smart teams of data scientists working on large data sets. Some examples of machine learning mistakes thanks to the narrow thinking of the humans that created them.

Some researchers claim we’re on the cusp of GoPro physics—where a camera can point at an event and an algorithm can identify the underlying physics equation

An in-depth look at Neural architecture search—the AutoML subfield aiming to replace manual designs

In an effort to enhance artificial intelligence & machine learning technologies military researchers are letting it be known they want more accurate processing of covariance information related to environmental variations and noise

Intelligence agencies are starting to coalesce around a set of common standards and data for using open source intelligence

A detailed explanation of handling satellite imagery in the format of .tiff files using Python.

A way to better understand road networks by measuring their spatial homogeneity using machine-learning tools like graph neural networks

The place where machine learning shines

A new deep learning technique shows promise to make robotics systems more stable in handling deformable objects

Small satellites: The implications for national security 

Ukraine may be a tipping point for developing intelligent weapons

Two main types of adversarial attacks in neural networks

It’s not just about gathering data—it’s telling compelling stories 

NGA to Leverage AI, ML for GEOINT Analysis at Scale 

From data scientist to … comedian?

Follow for a daily Data Science article

Data Science articles from Jan 2022

CDC announces plan to send every US household pamphlet on probabilistic thinking & Bayesian inferences

Using machine learning models & clustering network embeddings to get meaningful insights into social network ecosystems

Many PhD candidates are willing to publish findings based on fraudulent data based on some less familiar Bayesian methods

Scientists make first detection of exotic “X” particles in quark-gluon plasma

2022 Trends in Semantic Technologies: Humanizing Artificial Intelligence

NRO Selects 5 Companies for Commercial Radar Development Contracts

Tiny machine learning is bringing neural networks to microcontrollers

Why very few machine learning models actually get deployed

Remote Sensing: Deep Learning for Land Cover Classification of Satellite Imagery Using Python   

Hands-on data visualization for interactive storytelling in Python

The goal of a recommendation algorithm “isn’t to surprise or shock but to affirm. The process looks a lot like prediction, but it’s merely repetition. The result is more of the same: a present that looks like the past and a future that isn’t one.” https://bit.ly/3HW7Ior

The algorithmic feedback loop: "If you want to freeze culture, the 1st step is to reduce it to data & if you want to maintain the frozen status quo algorithms trained on people’s past behaviors & tastes would be the best tools..."

What patent trends can tell us about next gen of AI Tech

Stop talking about “statistical significance and practical significance”

The people behind the AI Curtain

“So much of what passes for automation isn’t really automation,” says writer and documentarian Astra Taylor. She describes a moment when she was waiting to pick up her lunch at a cafe, and another customer walked in, awestruck, wondering aloud how the app knew that his order was ready 20 minutes early. The woman behind the counter just looked at him and said, “I just sent you a message.”

“He was so convinced that it was a robot,” Taylor says. “He couldn’t see the human labor right in front of his eyes.”

She calls this process fauxtomation: “Fauxtomation renders invisible human labor to make computers seem smarter than they are.”

“AI” usually relies on a lot of low-paid human labor.

Katharine Manning Schwab writing in Fast Company