Timezone: »

Model-agnostic Measure of Generalization Difficulty
Akhilan Boopathy · Kevin Liu · Jaedong Hwang · Shu Ge · Asaad Mohammedsaleh · Ila R. Fiete

Wed Jul 26 02:00 PM -- 03:30 PM (PDT) @ Exhibit Hall 1 #433
The measure of a machine learning algorithm is the difficulty of the tasks it can perform, and sufficiently difficult tasks are critical drivers of strong machine learning models. However, quantifying the generalization difficulty of machine learning benchmarks has remained challenging. We propose what is to our knowledge the first model-agnostic measure of the inherent generalization difficulty of tasks. Our inductive bias complexity measure quantifies the total information required to generalize well on a task minus the information provided by the data. It does so by measuring the fractional volume occupied by hypotheses that generalize on a task given that they fit the training data. It scales exponentially with the intrinsic dimensionality of the space over which the model must generalize but only polynomially in resolution per dimension, showing that tasks which require generalizing over many dimensions are drastically more difficult than tasks involving more detail in fewer dimensions. Our measure can be applied to compute and compare supervised learning, reinforcement learning and meta-learning generalization difficulties against each other. We show that applied empirically, it formally quantifies intuitively expected trends, e.g. that in terms of required inductive bias, MNIST $<$ CIFAR10 $<$ Imagenet and fully observable Markov decision processes (MDPs) $<$ partially observable MDPs. Further, we show that classification of complex images $<$ few-shot meta-learning with simple images. Our measure provides a quantitative metric to guide the construction of more complex tasks requiring greater inductive bias, and thereby encourages the development of more sophisticated architectures and learning algorithms with more powerful generalization capabilities.

Author Information

Akhilan Boopathy (MIT)
Kevin Liu (MIT)
Jaedong Hwang (MIT)
Shu Ge (Massachusetts Institute of Technology)
Asaad Mohammedsaleh (Massachusetts Institute of Technology)
Ila R. Fiete (MIT)

More from the Same Authors