Competition-level code generation with AlphaCode.

Programming is a powerful and ubiquitous problem-solving tool. Systems that can assist programmers or even generate programs themselves could make programming more productive and accessible. Recent transformer-based neural network models show impressive code generation abilities yet still perform poorly on more complex tasks requiring problem-solving skills, such as competitive programming problems. Here, we introduce AlphaCode, a system for code generation that achieved an average ranking in the top 54.3% in simulated evaluations on recent programming competitions on the Codeforces platform. AlphaCode solves problems by generating millions of diverse programs using specially trained transformer-based networks and then filtering and clustering those programs to a maximum of just 10 submissions. This result marks the first time an artificial intelligence system has performed competitively in programming competitions.

Restoring and attributing ancient texts using deep neural networks.

Ancient history relies on disciplines such as epigraphy-the study of inscribed texts known as inscriptions-for evidence of the thought, language, society and history of past civilizations1. However, over the centuries, many inscriptions have been damaged to the point of illegibility, transported far from their original location and their date of writing is steeped in uncertainty. Here we present Ithaca, a deep neural network for the textual restoration, geographical attribution and chronological attribution of ancient Greek inscriptions. Ithaca is designed to assist and expand the historian's workflow. The architecture of Ithaca focuses on collaboration, decision support and interpretability. While Ithaca alone achieves 62% accuracy when restoring damaged texts, the use of Ithaca by historians improved their accuracy from 25% to 72%, confirming the synergistic effect of this research tool. Ithaca can attribute inscriptions to their original location with an accuracy of 71% and can date them to less than 30 years of their ground-truth ranges, redating key texts of Classical Athens and contributing to topical debates in ancient history. This research shows how models such as Ithaca can unlock the cooperative potential between artificial intelligence and historians, transformationally impacting the way that we study and write about one of the most important periods in human history.

Distinct ECG Phenotypes Identified in Hypertrophic Cardiomyopathy Using Machine Learning Associate With Arrhythmic Risk Markers.

Aims: Ventricular arrhythmia triggers sudden cardiac death (SCD) in hypertrophic cardiomyopathy (HCM), yet electrophysiological biomarkers are not used for risk stratification. Our aim was to identify distinct HCM phenotypes based on ECG computational analysis, and characterize differences in clinical risk factors and anatomical differences using cardiac magnetic resonance (CMR) imaging. Methods: High-fidelity 12-lead Holter ECGs from 85 HCM patients and 38 healthy volunteers were analyzed using mathematical modeling and computational clustering to identify phenotypic subgroups. Clinical features and the extent and distribution of hypertrophy assessed by CMR were evaluated in the subgroups. Results: QRS morphology alone was crucial to identify three HCM phenotypes with very distinct QRS patterns. Group 1 (n = 44) showed normal QRS morphology, Group 2 (n = 19) showed short R and deep S waves in V4, and Group 3 (n = 22) exhibited short R and long S waves in V4-6, and left QRS axis deviation. However, no differences in arrhythmic risk or distribution of hypertrophy were observed between these groups. Including T wave biomarkers in the clustering, four HCM phenotypes were identified: Group 1A (n = 20), with primary repolarization abnormalities showing normal QRS yet inverted T waves, Group 1B (n = 24), with normal QRS morphology and upright T waves, and Group 2 and Group 3 remaining as before, with upright T waves. Group 1A patients, with normal QRS and inverted T wave, showed increased HCM Risk-SCD scores (1A: 4.0%, 1B: 1.8%, 2: 2.1%, 3: 2.5%, p = 0.0001), and a predominance of coexisting septal and apical hypertrophy (p < 0.0001). HCM patients in Groups 2 and 3 exhibited predominantly septal hypertrophy (85 and 90%, respectively). Conclusion: HCM patients were classified in four subgroups with distinct ECG features. Patients with primary T wave inversion not secondary to QRS abnormalities had increased HCM Risk-SCD scores and coexisting septal and apical hypertrophy, suggesting that primary T wave inversion may increase SCD risk in HCM, rather than T wave inversion secondary to depolarization abnormalities. Computational ECG phenotyping provides insight into the underlying processes captured by the ECG and has the potential to be a novel and independent factor for risk stratification.

Building machines that learn and think for themselves.

We agree with Lake and colleagues on their list of "key ingredients" for building human-like intelligence, including the idea that model-based reasoning is essential. However, we favor an approach that centers on one additional ingredient: autonomy. In particular, we aim toward agents that can both build and exploit their own internal models, with minimal human hand engineering. We believe an approach centered on autonomous learning has the greatest chance of success as we scale toward real-world complexity, tackling domains for which ready-made formal models are not available. Here, we survey several important examples of the progress that has been made toward building autonomous agents with human-like abilities, and highlight some outstanding challenges.

Learning where to attend with deep architectures for image tracking.

We discuss an attentional model for simultaneous object tracking and recognition that is driven by gaze data. Motivated by theories of perception, the model consists of two interacting pathways, identity and control, intended to mirror the what and where pathways in neuroscience models. The identity pathway models object appearance and performs classification using deep (factored)-restricted Boltzmann machines. At each point in time, the observations consist of foveated images, with decaying resolution toward the periphery of the gaze. The control pathway models the location, orientation, scale, and speed of the attended object. The posterior distribution of these states is estimated with particle filtering. Deeper in the control pathway, we encounter an attentional mechanism that learns to select gazes so as to minimize tracking uncertainty. Unlike in our previous work, we introduce gaze selection strategies that operate in the presence of partial information and on a continuous action space. We show that a straightforward extension of the existing approach to the partial information setting results in poor performance, and we propose an alternative method based on modeling the reward surface as a gaussian process. This approach gives good performance in the presence of partial information and allows us to expand the action space from a small, discrete set of fixation points to a continuous domain.