Bounding the sample size of a Machine Learning Algorithm

One common problem with machine learning algorithms is that we don’t know how much training data we need. A common way around this is the often used strategy: keep training until the training error stops decreasing. However, there are still issues with this. How do we know we’re not stuck in a local minimum? What if the training error has strange behavior, sometimes staying flat over training iterations but sometimes decreasing sharply? The bottom line is that without a precise way of knowing how much training data we need, there will always be some uncertainty as to whether or not we are done training.

Read More

Support Vector Machines — It’s not-so-complicated!

Have you ever wondered while coming across the ‘house price’ (denoted by y) prediction task using Linear Regression that the price might not be directly a linear combination of the different features such as ‘size of house’, ‘number of bedrooms’, ‘number of neighbours’ etc? It is absolutely possible that the price might be a non-linear function of the features, isn’t it? It is also quite intuitive that this can be extended to Logistic Regression as well.

Read More

The revolution of Computational Creativity

When the finance company JPMorgan Chase announced that it was replacing human copywriters with an algorithm, some thought it was a marketing ploy riding the crest of the AI ​​wave. However, those who knew something about so-called artificial intelligence knew that this was just the beginning. Computational creativity was breaking from academia into real-life applications in the creative industries.

Read More

The revolving door for Machine Learning Models

Revolving door

I would like to argue that nowadays specializing in a certain domain allows you to accumulate cross-domain tools, which can ease your transition between domains. Once you acquire the basics in statistics, probability theory, information theory, mathematics, algorithms, and machine learning, you realize that you can reuse nearly every algorithm for various purposes and use cases.

Read More

5 questions to help you choose the right Data Labeling Tool

Data labeling tool

The versatility of a data labeling tool can make or break your data quality. And the data quality can make or break your algorithms. And what happens when our algorithms misinterpret or fail? — Karthik Vasudevan, Founder at Traindata Inc. This post will guide you to ask five questions to help you choose the best data labeling tool.

Read More

[Book] Five Reasoning Methods to rule them all

The five reasoning methods are also called the five tribes. They help to solve the Master Algorithm. Each of the five tribes has a different technique and strategy for solving problems that result in unique algorithms. If we are successful to combine these algorithms, then it will lead us to (theoretically) the master algorithm. These are defined by the Portugues author, Pedro Domingos in his book The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World.

Read More

New $35M AI Research Center at Indiana University created to grow AI Education

Luddy Center for Artificial Intelligence

The Luddy Center for Artificial Intelligence, which was unveiled June 23 and will open in August for the start of the fall semester, includes 58,000 square feet of space designed to enable multidisciplinary research in the constantly expanding AI field.

Read More

Artificial intelligence is the simulation of human intelligence processes on machines

AI Simulation

Artificial intelligence is the simulation of human intelligence processes on machines. AI systems work by using labeled data, analysing the data for patterns and using these patterns to make predictions about future or about interests of a prospective customer. Examples of AI are chatbots, image recognition tools, voice recognition tools etc. AI programming focuses on acquiring data and creating rules about the data.

Read More

Kellogg exec on AI uses cases, implementation, and ‘culture change’

Shipping yard

In this era of evolving technology, organizations must be highly adaptive to succeed. A Statistics report reveals that before the pandemic over 4.7 million people in the U.S. were working remotely at least half the time — a percentage that has since increased. And fully 75% of people using digital channels for the first time indicate that they’ll continue to use them when things return to a post-pandemic “normal.”

Read More

Most business managers don’t care about fancy algorithms you might have learned

Woman reading a book

It might surprise you to hear that most business leaders who hire Data Scientists don’t actually care about fancy statistical/ML algorithms you might have read or learned. Certainly, it seems counter-intuitive based on opinion of aspiring data hopefuls that you must have complex, state-of-the-art algorithms under your technical tool-kit. But veterans in the industry see it differently. I’m only in the third year as a Data Scientist in Silicon Valley of California, but I’ve only become more and more suspicious and hesitant about complex algorithmic solutions to business problems. Here are the basic reasons why that is the case, what these business leaders look for when hiring Data Scientists instead.

Read More

Making a case for Serverless Machine Learning

The scale and complexity of machine learning make it hard to provide and manage data and resources efficiently. This hinders and decreases productivity. The easiest way to approach the problem is serverless machine learning. It is an excellent solution to the problem of data center resource management. Machine learning users face several daunting challenges that have a significant impact on their productivity and efficiency.

Read More

This new tool can track the Environmental Cost of your Machine Learning Model

EnergyVis

Energy consumption is a major factor to plan for when implementing a long-term project or service that uses large-scale machine learning algorithms. Now, a team of researchers from Georgia Tech has created an interactive tool called EnergyVis that allows users to compare energy consumption across locations and against other models. 

Read More

Machine Learning in the Automotive Industry

automobiles & ML

AI simulates the applications that extend the automotive manufacturing ground. Automakers can use #AI-driven systems to create directories and manage workflows, enabling the robots to operate safely alongside humans on manufacturer grounds and panel lines and identify defects in segments going into cars and trucks.

Read More

Algorithms are not Sexist — We are!

algorithms

The AI that famously created an image of AOC in a bikini was behaving inappropriately but rationally. If you believe the AI that created male and female images from pictures of their heads, men typically wear suits, while women prefer low-cut tops and bikinis. It’s been reported in the press that this is evidence that the future of AI is sexist. The argument being that because the internet is awash with pictures of scantily-clad females, AI will assume that this is normal. A fair point? Not necessarily…

Read More

Federated Learning: A decentralized form of Machine Learning

Most major consumer tech companies that are focused on AI and machine learning now use federated learning – a form of machine learning that trains algorithms on devices distributed across a network, without the need for data to leave each device. Given the increasing awareness of privacy issues, federated learning could become the preferred method of machine learning for use cases that use sensitive data (such as location, financial, or health data).

Read More
1 2 3 4