Machine Learning / AI

What is the concept of computational graphs?

How to model conversions?

What are the features of training detection on heavily overlapping objects?

What does optimization in machine marketing consist of?

How to avoid training the model with heuristics of creating negative examples instead of generalization?

What is a replay buffer in continual learning?

Middle — Senior

HybridСБЕР

Does the dropout probability persist when converting a model to ONNX?

How do you scale activations when dropping out during inference?

What is VAD (voice activity detection) and what models (Silero VAD) are there?

What is the main property that a binding must have for browser identification?

How to create relevant and reliable negative pairs for model training?