The run-time for optimization tools used in chip design has grown with the complexity of designs to the point where it can take several days to go through one design cycle which has become a bottleneck. Designers want fast tools that can quickly give feedback on a design. Using the input and output data of the tools from past designs, one can attempt to build a machine learning model that predicts the outcome of a design in significantly shorter time than running the tool. The accuracy of such models is affected by the representation of the design data, which is usually a netlist that describes the elements of the digital circuit and how they are connected. Graph representations for the netlist together with graph neural networks have been investigated for such models. However, the characteristics of netlists pose several challenges for existing graph learning frameworks, due to the large number of nodes and the importance of long-range interactions between nodes. To address these challenges, we represent the netlist as a directed hypergraph and propose a Directional Equivariant Hypergraph Neural Network (DE-HNN) for the effective learning of (directed) hypergraphs. Theoretically, we show that our DE-HNN can universally approximate any node or hyperedge based function that satisfies certain permutation equivariant and invariant properties natural for directed hypergraphs. We compare the proposed DE-HNN with several State-of-the-art (SOTA) machine learning models for (hyper)graphs and netlists, and show that the DE-HNN significantly outperforms them in predicting the outcome of optimized place-and-route tools directly from the input netlists.

Large language models are increasingly solving tasks that are commonly believed to require human-level reasoning ability. However, these models still perform very poorly on benchmarks of general intelligence such as the Abstraction and Reasoning Corpus (ARC). In this paper, we approach the ARC as a programming-by-examples problem, and introduce a novel and scalable method for language model self-improvement called Code Iteration (CodeIt). Our method iterates between 1) program sampling and hindsight relabeling, and 2) learning from prioritized experience replay. By relabeling the goal of an episode (i.e., the program output given input) to the output actually produced by the sampled program, our method effectively deals with the extreme sparsity of rewards in program synthesis. Applying CodeIt to the ARC dataset, we demonstrate that prioritized hindsight replay, along with pre-training and data-augmentation, leads to successful inter-task generalization. CodeIt is the first neuro-symbolic approach that scales to the full ARC evaluation dataset. Our method solves 15% of ARC evaluation tasks, achieving state-of-the-art performance and outperforming existing neural and symbolic baselines. Our code is available at https://github.com/Qualcomm-AI-research/codeit.

Proteins are typically represented by discrete atomic coordinates providing an accessible framework to describe different conformations. However, in some fields proteins are more accurately represented as near-continuous surfaces, as these are imprinted with geometric (shape) and chemical (electrostatics) features of the underlying protein structure. Protein surfaces are dependent on their chemical composition and, ultimately determine protein function, acting as the interface that engages in interactions with other molecules. In the past, such representations were utilized to compare protein structures on global and local scales and have shed light on functional properties of proteins. Here we describe RosettaSurf, a surface-centric computational design protocol, that focuses on the molecular surface shape and electrostatic properties as means for protein engineering, offering a unique approach for the design of proteins and their functions. The RosettaSurf protocol combines the explicit optimization of molecular surface features with a global scoring function during the sequence design process, diverging from the typical design approaches that rely solely on an energy scoring function. With this computational approach, we attempt to address a fundamental problem in protein design related to the design of functional sites in proteins, even when structurally similar templates are absent in the characterized structural repertoire. Surface-centric design exploits the premise that molecular surfaces are, to a certain extent, independent of the underlying sequence and backbone configuration, meaning that different sequences in different proteins may present similar surfaces. We benchmarked RosettaSurf on various sequence recovery datasets and showcased its design capabilities by generating epitope mimics that were biochemically validated. Overall, our results indicate that the explicit optimization of surface features may lead to new routes for the design of functional proteins.

In this work, we are interested in generalizing convolutional neural networks (CNNs) from low-dimensional regular grids, where image, video and speech are represented, to high-dimensional irregular domains, such as social networks, brain connectomes or words’ embedding, represented by graphs. We present a formulation of CNNs in the context of spectral graph theory, which provides the necessary mathematical background and efficient numerical schemes to design fast localized convolutional filters on graphs. Importantly, the proposed technique offers the same linear computational complexity and constant learning complexity as classical CNNs, while being universal to any graph structure. Experiments on MNIST and 20NEWS demonstrate the ability of this novel deep learning system to learn local, stationary, and compositional features on graphs.