François Chollet

Luke Wood, Francois Chollet

Evaluating the COCO mean average precision (MaP) and COCO recall metrics as part of the static computation graph of modern deep learning frameworks poses a unique set of challenges. These challenges include the need for maintaining a dynamic-sized state to compute mean average precision, reliance on global dataset-level statistics to compute the metrics, and managing differing numbers of bounding boxes between images in a batch. As a consequence, it is common practice for researchers and practitioners to evaluate COCO metrics as a post training evaluation step. With a graph-friendly algorithm to compute COCO Mean Average Precision and recall, these metrics could be evaluated at training time, improving visibility into the evolution of the metrics through training curve plots, and decreasing iteration time when prototyping new model versions. Our contributions include an accurate approximation algorithm for Mean Average Precision, an open source implementation of both COCO mean average precision and COCO recall, extensive numerical benchmarks to verify the accuracy of our implementations, and an open-source training loop that include train-time evaluation of mean average precision and recall.

Lance Ying, Katherine M. Collins, Prafull Sharma et al.

Human intelligence exhibits a remarkable capacity for rapid adaptation and effective problem-solving in novel and unfamiliar contexts. We argue that this profound adaptability is fundamentally linked to the efficient construction and refinement of internal representations of the environment, commonly referred to as world models, and we refer to this adaptation mechanism as world model induction. However, current understanding and evaluation of world models in artificial intelligence (AI) remains narrow, often focusing on static representations learned from training on massive corpora of data, instead of the efficiency and efficacy in learning these representations through interaction and exploration within a novel environment. In this Perspective, we provide a view of world model induction drawing on decades of research in cognitive science on how humans learn and adapt so efficiently; we then call for a new evaluation framework for assessing adaptive world models in AI. Concretely, we propose a new benchmarking paradigm based on suites of carefully designed games with genuine, deep and continually refreshing novelty in the underlying game structures -- we refer to this class of games as novel games. We detail key desiderata for constructing these games and propose appropriate metrics to explicitly challenge and evaluate the agent's ability for rapid world model induction. We hope that this new evaluation framework will inspire future evaluation efforts on world models in AI and provide a crucial step towards developing AI systems capable of human-like rapid adaptation and robust generalization -- a critical component of artificial general intelligence.

François Chollet

To make deliberate progress towards more intelligent and more human-like artificial systems, we need to be following an appropriate feedback signal: we need to be able to define and evaluate intelligence in a way that enables comparisons between two systems, as well as comparisons with humans. Over the past hundred years, there has been an abundance of attempts to define and measure intelligence, across both the fields of psychology and AI. We summarize and critically assess these definitions and evaluation approaches, while making apparent the two historical conceptions of intelligence that have implicitly guided them. We note that in practice, the contemporary AI community still gravitates towards benchmarking intelligence by comparing the skill exhibited by AIs and humans at specific tasks such as board games and video games. We argue that solely measuring skill at any given task falls short of measuring intelligence, because skill is heavily modulated by prior knowledge and experience: unlimited priors or unlimited training data allow experimenters to "buy" arbitrary levels of skills for a system, in a way that masks the system's own generalization power. We then articulate a new formal definition of intelligence based on Algorithmic Information Theory, describing intelligence as skill-acquisition efficiency and highlighting the concepts of scope, generalization difficulty, priors, and experience. Using this definition, we propose a set of guidelines for what a general AI benchmark should look like. Finally, we present a benchmark closely following these guidelines, the Abstraction and Reasoning Corpus (ARC), built upon an explicit set of priors designed to be as close as possible to innate human priors. We argue that ARC can be used to measure a human-like form of general fluid intelligence and that it enables fair general intelligence comparisons between AI systems and humans.

Ashish Vaswani, Samy Bengio, Eugene Brevdo et al.

Tensor2Tensor is a library for deep learning models that is well-suited for neural machine translation and includes the reference implementation of the state-of-the-art Transformer model.

Lukasz Kaiser, Aidan N. Gomez, Francois Chollet

Depthwise separable convolutions reduce the number of parameters and computation used in convolutional operations while increasing representational efficiency. They have been shown to be successful in image classification models, both in obtaining better models than previously possible for a given parameter count (the Xception architecture) and considerably reducing the number of parameters required to perform at a given level (the MobileNets family of architectures). Recently, convolutional sequence-to-sequence networks have been applied to machine translation tasks with good results. In this work, we study how depthwise separable convolutions can be applied to neural machine translation. We introduce a new architecture inspired by Xception and ByteNet, called SliceNet, which enables a significant reduction of the parameter count and amount of computation needed to obtain results like ByteNet, and, with a similar parameter count, achieves new state-of-the-art results. In addition to showing that depthwise separable convolutions perform well for machine translation, we investigate the architectural changes that they enable: we observe that thanks to depthwise separability, we can increase the length of convolution windows, removing the need for filter dilation. We also introduce a new "super-separable" convolution operation that further reduces the number of parameters and computational cost for obtaining state-of-the-art results.

Cezary Kaliszyk, François Chollet, Christian Szegedy

Large computer-understandable proofs consist of millions of intermediate logical steps. The vast majority of such steps originate from manually selected and manually guided heuristics applied to intermediate goals. So far, machine learning has generally not been used to filter or generate these steps. In this paper, we introduce a new dataset based on Higher-Order Logic (HOL) proofs, for the purpose of developing new machine learning-based theorem-proving strategies. We make this dataset publicly available under the BSD license. We propose various machine learning tasks that can be performed on this dataset, and discuss their significance for theorem proving. We also benchmark a set of simple baseline machine learning models suited for the tasks (including logistic regression, convolutional neural networks and recurrent neural networks). The results of our baseline models show the promise of applying machine learning to HOL theorem proving.

François Chollet

We present an interpretation of Inception modules in convolutional neural networks as being an intermediate step in-between regular convolution and the depthwise separable convolution operation (a depthwise convolution followed by a pointwise convolution). In this light, a depthwise separable convolution can be understood as an Inception module with a maximally large number of towers. This observation leads us to propose a novel deep convolutional neural network architecture inspired by Inception, where Inception modules have been replaced with depthwise separable convolutions. We show that this architecture, dubbed Xception, slightly outperforms Inception V3 on the ImageNet dataset (which Inception V3 was designed for), and significantly outperforms Inception V3 on a larger image classification dataset comprising 350 million images and 17,000 classes. Since the Xception architecture has the same number of parameters as Inception V3, the performance gains are not due to increased capacity but rather to a more efficient use of model parameters.

François Chollet

We present a method for training multi-label, massively multi-class image classification models, that is faster and more accurate than supervision via a sigmoid cross-entropy loss (logistic regression). Our method consists in embedding high-dimensional sparse labels onto a lower-dimensional dense sphere of unit-normed vectors, and treating the classification problem as a cosine proximity regression problem on this sphere. We test our method on a dataset of 300 million high-resolution images with 17,000 labels, where it yields considerably faster convergence, as well as a 7% higher mean average precision compared to logistic regression.

Alex A. Alemi, Francois Chollet, Niklas Een et al.

We study the effectiveness of neural sequence models for premise selection in automated theorem proving, one of the main bottlenecks in the formalization of mathematics. We propose a two stage approach for this task that yields good results for the premise selection task on the Mizar corpus while avoiding the hand-engineered features of existing state-of-the-art models. To our knowledge, this is the first time deep learning has been applied to theorem proving on a large scale.