IAI | HSLU | Magdalena Picariello
Introduction to AI | HSLU
Introduction to AI | HSLU
Set of flashcards Details
| Flashcards | 92 |
|---|---|
| Language | English |
| Category | Computer Science |
| Level | University |
| Created / Updated | 17.10.2023 / 02.11.2023 |
| Weblink |
https://card2brain.ch/box/20231017_iai_%7C_hslu_%7C_magdalena_picariello
|
| Embed |
<iframe src="https://card2brain.ch/box/20231017_iai_%7C_hslu_%7C_magdalena_picariello/embed" width="780" height="150" scrolling="no" frameborder="0"></iframe>
|
Checklist für data augmentation:
● Does it sound realistic?
● Is the x->y mapping clear (e.g. can humans recognize speech?)
● Is the algorithm currently doing poorly on it?
How can we perform artificial data synthesis?
With unstructured data, we can perform artificial data synthesis
What is feature engineering in machine learning?
With structured data, we can do feature engineering.
Example: Restaurant recommender
Vegetarians are frequently recommended restaurants with only meat options Possible features to add:
● Is person vegetarian (based on past orders)?
● Does restaurant have vegetarian options (based on menu)
What is the role of data iteration in the feature engineering process?
Data iteration involves continuously revising and improving features based on the results of error analysis, user feedback, and benchmarking. It helps enhance the quality and relevance of features.
Why can error analysis be more challenging without a good baseline (HLP)?
- Error analysis becomes challenging without a good baseline (Human-Level Performance or HLP) because you lack a reliable reference point for comparison. A baseline provides insight into how well a human can perform the task, which is essential for identifying areas where the model falls short.
Q1: How can user feedback contribute to the data iteration process in feature engineering?
Q2: Why is benchmarking against competitors a useful source of inspiration for feature engineering?
A1: User feedback is valuable for identifying which features users find relevant or missing. It provides insights that guide the selection and engineering of features to enhance user satisfaction and model performance.
A2: Benchmarking against competitors helps identify areas where your model can gain a competitive edge. By analyzing competitor performance, you can inspire the development of unique features that set your model apart in the field.
Build your system quickly, then iterate
what are the directions you can take to improve speech recognition system:
● Noisy background:
○ Cafe noise
○ Car noise
● Accented speech
● Far from microphone
● Young children
● Stuttering
Data or modeling?
AI system = Code (model) + data
Data or modeling?
Model centric view VS. Data cenrtic view
What is the typical mindset in Data or Modeling?
- Traditional machine learning research driven by improving benchmark dataset performance.
- Researchers often work on a fixed dataset they download.
- This approach has led to significant progress in machine learning.
- In production systems, the dataset doesn't need to remain fixed.
- It's common to edit the training and test sets to enhance data quality for better system performance.
Questions:
- In what cases can focusing on optimizing the data and hyperparameters be more effective than code optimization?
- What does a machine learning system comprise in terms of components?
- How does taking a non-model-centric approach differ in terms of optimization?
- What is a crucial step during the modeling phase of machine learning?
- How can error analysis be utilized in the context of data improvement?
- Why is collecting more data not always the most efficient solution?
- How can error analysis contribute to a high-accuracy model?
How long should you spend obtaining data?
