Data Engineering for AI, a Real Game-Changer
It is a well-known fact that an AI model is only as good as the data it is fed. Model-based AI is judged against benchmark datasets and is compared on specific parameters. Most researchers depend on algorithmic improvement rather than on data quality. Most companies hesitate to adopt AI due to this very reason – lack of good data. Andrew Ng, the prominent AI researcher believes data-centric AI to turn this game on its head. And there are pretty many good reasons for this proposition.
An AI cycle involves, for most of its part involves preparing high-quality data, which makes up at least 80% of the AI lifecycle. Even though training the AI model constitutes 20% of training the model, it has been the center of focus for researchers and businesses as well.
The benefits of a data-centric approach go beyond this single advantage. For example, labeling data which is basically done by humans comes with certain inconsistencies. It introduces noise in the data which can prove detrimental to the AI cycle. This happens because of differences in interpreting guidelines. Andrew says, by identifying the points of disagreement for labeling, discrepancies can be reduced to a great extent. When a model is fine-tuned based on faulty data, it is bound to have lower accuracy. In a data-centric approach, as the data is pre-engineered for designing a model, one can expect data accuracy apart from achieving model optimization.
Data engineering has a lot to do with fixing noisy data. In most cases, AI projects fail to see light due to a lack of sufficient data. When a small dataset is used for a model, it makes room for a lot of noise and therefore errors. When you have pre modeled dataset, the noise level can be reduced significantly. Data-centric AI addresses all the fringe cases with clearly defined data focusing on the model architecture.
More Trending Stories
- Shiba Inu is All Set to Reach US$0.000017 this July! But Won’t Break Past That
- Google to Delete Sensitive Data on Users’ Location History after Roe Revocation
- Books almost Losing Copyright gets an NFT Twist to Preserve its Status
- Deepmind’s ‘Democratic AI’ Distributes Public Money! Might Make an AI Govt
- Foundation AI Models Like DALL.E & GPT-3 Need Community Norms Before they Harm
- Top 10 Famous Ethical Hackers in India to Know About in 2022
- Can Dumping Commercial Paper Holding Help Tether Gain US Dollar-Peg Trust?
The post Data Engineering for AI, a Real Game-Changer appeared first on .
It is a well-known fact that an AI model is only as good as the data it is fed. Model-based AI is judged against benchmark datasets and is compared on specific parameters. Most researchers depend on algorithmic improvement rather than on data quality. Most companies hesitate to adopt AI due to this very reason – lack of good data. Andrew Ng, the prominent AI researcher believes data-centric AI to turn this game on its head. And there are pretty many good reasons for this proposition.
An AI cycle involves, for most of its part involves preparing high-quality data, which makes up at least 80% of the AI lifecycle. Even though training the AI model constitutes 20% of training the model, it has been the center of focus for researchers and businesses as well.
The benefits of a data-centric approach go beyond this single advantage. For example, labeling data which is basically done by humans comes with certain inconsistencies. It introduces noise in the data which can prove detrimental to the AI cycle. This happens because of differences in interpreting guidelines. Andrew says, by identifying the points of disagreement for labeling, discrepancies can be reduced to a great extent. When a model is fine-tuned based on faulty data, it is bound to have lower accuracy. In a data-centric approach, as the data is pre-engineered for designing a model, one can expect data accuracy apart from achieving model optimization.
Data engineering has a lot to do with fixing noisy data. In most cases, AI projects fail to see light due to a lack of sufficient data. When a small dataset is used for a model, it makes room for a lot of noise and therefore errors. When you have pre modeled dataset, the noise level can be reduced significantly. Data-centric AI addresses all the fringe cases with clearly defined data focusing on the model architecture.
More Trending Stories
- Shiba Inu is All Set to Reach US$0.000017 this July! But Won’t Break Past That
- Google to Delete Sensitive Data on Users’ Location History after Roe Revocation
- Books almost Losing Copyright gets an NFT Twist to Preserve its Status
- Deepmind’s ‘Democratic AI’ Distributes Public Money! Might Make an AI Govt
- Foundation AI Models Like DALL.E & GPT-3 Need Community Norms Before they Harm
- Top 10 Famous Ethical Hackers in India to Know About in 2022
- Can Dumping Commercial Paper Holding Help Tether Gain US Dollar-Peg Trust?
The post Data Engineering for AI, a Real Game-Changer appeared first on .
Denial of responsibility! Techno Blender is an automatic aggregator of the all world’s media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, all materials to their authors. If you are the owner of the content and do not want us to publish your materials, please contact us by email – [email protected]. The content will be deleted within 24 hours.