Approximate Bayesian Inference via Bitstring Representations
This work addresses scalability and interpretability challenges in machine learning for practitioners dealing with large models, though it is incremental as it builds on existing quantization techniques.
The paper tackles the problem of scaling probabilistic inference by performing it in quantized, discrete parameter spaces, enabling learning of continuous distributions with discrete parameters and achieving inference efficiency without accuracy loss.
The machine learning community has recently put effort into quantized or low-precision arithmetics to scale large models. This paper proposes performing probabilistic inference in the quantized, discrete parameter space created by these representations, effectively enabling us to learn a continuous distribution using discrete parameters. We consider both 2D densities and quantized neural networks, where we introduce a tractable learning approach using probabilistic circuits. This method offers a scalable solution to manage complex distributions and provides clear insights into model behavior. We validate our approach with various models, demonstrating inference efficiency without sacrificing accuracy. This work advances scalable, interpretable machine learning by utilizing discrete approximations for probabilistic computations.