Startup Ensemble receives $3.3 million in funding to solve data quality issues related to Dark Matter

Ensemble AI Inc. It aims to tackle data quality challenges and help companies build more powerful AI models after completing a $3.3 million seed funding round.

today’s tour Amplo, M13 and Motivate also take part in the event managed by Salesforce Ventures. They support Ensemble because the startup has created a pioneering approach to data representation to improve the performance of AI models without pumping huge amounts of extra data into them or creating more complex model architectures.

What the startup does is use machine learning techniques to improve AI models by helping them uncover hidden relationships between data sets. The company explains that artificial intelligence needs access to more and better quality data to solve real-world problems. Many companies struggle with limited, sparse or one-dimensional data sets, preventing AI models from producing meaningful or useful results.

To overcome this, data scientists spend hours curating their data, and some progress has been made with more complex AI model architectures, but such efforts require extensive resources and technical expertise that not every company has.

To solve these problems, Ensemble created a new installation model called . Dark MatterUsing an “objective function” to create richer representations of data for predictive tasks. The company says Dark Matter can understand complex, non-linear relationships within datasets through a lightweight data transformation. By transforming the complexity of these relationships into a simple “data representation,” it enables engineers to create higher-quality AI models that can tackle much more difficult problems.

Ensemble co-founder and Chief Executive Alex Reneau explained that Dark Matter sits between feature engineering and model training and inference processes within data pipelines.

“We can enable customers to maximize their own data on which they work, even if it is limited, sparse or extremely complex, allowing them to train effective models with less extensive information,” he said. “This core technology allows data scientists to focus on experiments and also unlocks new capabilities for our customers by making ML applicable to problems that could not previously be modeled.”

The startup believes Dark Matter is a superior solution to synthetic data, which is often used by AI developers to compensate for low-quality or sparse datasets. He explains that although Dark Matter creates new variables, its mechanics are fundamentally different.

Since synthetic data reconstructs existing distributions from Gaussian noise, this means that no new information is actually created. The company explained that the synthetic data only reflects the statistical properties of existing data, so it has no significant impact on forecast accuracy.

Dark Matter, on the other hand, is learning how to create new embeddings with fundamentally different statistical properties and distributions, providing measurably improved prediction accuracy.

Caroline Fiegel of Salesforce Ventures said: VentureBeat Ensemble offers a promising solution that could potentially accelerate the adoption of AI. He explained that many organizations struggle to use AI models in production due to issues with poor data quality and potential use of personally identifiable information.

“When you peel that back and really start to understand why, it’s because the data is different. “This is a bit of a low-quality thing,” he said. “It’s full of PII.”

Ensemble says Dark Matter is already being used by a number of early adopters with promising results in areas such as biotech, healthcare, personalization and ad tech. For example, one biotech customer used its technology to create a model that could better predict virus-host interactions within the gut microbiome.

Looking forward, Ensemble said it will use the funds raised from today’s round to expand its team and accelerate its product development and go-to-market plan.

Image: SiliconANGLE/Microsoft Designer