top of page

AI & Machine Learning

PROBLEM AREAS

sample-ai.jpg

COMPUTER VISION

Detection

Can it see an object?

Classification

Can it identify what the object is?

Segmentation

Can it break images down into smaller parts and identify them?

Generation

Can it generate a stylized image based on input?

sample-ai-2.jpg

DECISION AIDES

Optimization

Can it improve key factors in solutions (requirements coverage, run time, analysis time, risk, cost, etc.)?

Recommendation

Can it isolate influencing information and generate courses of action?

Knowledge Capture

Can it learn from the human overlords and eventually replace them?

kidding.jpg

GENERAL

Anomaly Detection

Can it find/predict problems for us?

Automation

Can it perform the task for us?

TECHNIQUES

sample-machine-learning.jpg

Convolutional Neural Networks (CNN) – A type of artificial neural network commonly used for image recognition and processing. CNNs are composed of multiple layers, with each layer containing a number of convolutional units. The convolutional units are responsible for detecting features in the input image, and the layers are stacked on top of each other to learn more complex features. CNNs have been shown to be very effective for image recognition and processing tasks, including image classification, object detection, and semantic segmentation.

 

Generative Adversarial Networks (GAN) – A type of machine learning model to generate realistic images, videos, and audio. They work by training two neural networks against each other: a generator and a discriminator. The generator's job is to create new data that looks like the training data, while the discriminator's job is to distinguish between real and fake data. The generator and discriminator are given rewards for their actions, and they learn to improve their performance over time.

 

Semantic Segmentation – A technique to assign a label or category to each pixel in an image. The labels can represent objects, such as people, cars, and buildings, or they can represent abstract concepts, such as sky, grass, and road.

 

Autoencoders – An artificial neural network that learns to reconstruct its input from a compressed representation. It is an unsupervised learning technique, as it does not require labeled data. Autoencoders are used for a variety of tasks, including image denoising, image compression, and feature learning. They can also be used to generate new data, such as images or text.

 

Diffusion – Generative models to produce images by starting with a random image and then gradually adding Gaussian noise to it. The noise is added in a controlled way so that the image gradually becomes more and more distorted. The final image is then denoised to recover the original image. Diffusion models learn to predict how the noise will affect the image, and it uses this knowledge to denoise the final image.

 

Transformers – A neural network that learns context and thus meaning by tracking relationships in sequential data. They are based on the attention mechanism, which allows them to learn long-range dependencies between data.

TRAINING

We use multiple training methods and combinations of methods to ensure accuracy and performance of our models.

Supervised – A type of machine learning in which the model is trained on a dataset of labeled data. The labels provide the model with information about the desired output for each input. 

 

Unsupervised – A type of machine learning in which the model is trained on a dataset of unlabeled data. The model is left to find patterns in the data on its own.

 

Reinforcement – A  type of machine learning in which the model learns to take actions in an environment in order to maximize a reward. The model is given a set of actions that it can take, and it learns to take the actions that are most likely to lead to a reward.

bottom of page