One way to address the known limitations of backpropagation through time is to directly adjust neural activities during the learning process. However, it remains unclear how to effectively use feedback to shape RNN dynamics. Here, we introduce error forcing (EF), where the network activity is guided orthogonally toward the zero-error manifold during learning. This method contrasts with alternatives like teaching forcing, which impose stronger constraints on neural activity and thus induce larger feedback influence on circuit dynamics. Furthermore, EF can be understood from a Bayesian perspective as a form of approximate dynamic inference. Empirically, EF consistently outperforms other learning algorithms across several tasks and its benefits persist when additional biological constraints are taken into account. Overall, EF is a powerful temporal credit assignment mechanism and a promising candidate model for learning in biological systems.

Connectomics--the mapping of neural connections in an organism's brain--currently requires extraordinary human effort to proofread the data collected from imaging and machine-learning assisted segmentation. With the growing excitement around using AI agents to automate important scientific tasks, we explore whether current AI systems can perform multiple tasks necessary for data proofreading. We introduce ConnectomeBench, a multimodal benchmark evaluating large language model (LLM) capabilities in three critical proofreading tasks: segment type identification, split error correction, and merge error detection. Using expert annotated data from two large open-source datasets--a cubic millimeter of mouse visual cortex and the complete Drosophila brain--we evaluate proprietary multimodal LLMs including Claude 3.7/4 Sonnet, o4-mini, GPT-4.1,

Some of the optical components used in Atom Computing's quantum computer The race to build the first truly useful quantum computer just got more exciting. A quantum computer made from extremely cold atoms has now passed some of the most important milestones towards usefulness, joining a small group of equally able and promising machines. Though there is wide agreement that sufficiently powerful quantum computers would transform our ability to discover new materials and drugs, and break the encryption that underpins the internet, there are many competing ideas about how best to build them. Industry mainstays such as Google and IBM have spent a decade building quantum computers from tiny superconducting circuits, and this approach is currently the front-runner. But an alternate approach that uses electrically neutral ultracold atoms has recently been gaining traction.

The errors that quantum computers make are holding the technology back. Quantum computers won't be truly useful until they can correct their mistakes Quantum computers are already here, but they make far too many errors. This is arguably the biggest obstacle to the technology really becoming useful, but recent breakthroughs suggest a solution may be on the horizon. Errors creep into traditional computers too, but there are well-established techniques for correcting them. They rely on redundancy, where extra bits are used to detect when 0s incorrectly swap to 1s or vice versa.

Neural Information Processing Systems http://nips.cc/

It looks like a golden chandelier and contains the coldest place in the universe. What I am looking at is not just the most powerful computer in the world, but technology pivotal to financial security, Bitcoin, government secrets, the world economy and more. Quantum computing holds the key to which companies and countries win - and lose - the rest of the 21st Century. In front of me suspended a metre in the air, in a Google facility in Santa Barbara California, is Willow. Frankly, it was not what I expected.

Error correction techniques have been used to refine the output sentences from automatic speech recognition (ASR) models and achieve a lower word error rate (WER) than original ASR outputs. Previous works usually use a sequence-to-sequence model to correct an ASR output sentence autoregressively, which causes large latency and cannot be deployed in online ASR services. A straightforward solution to reduce latency, inspired by non-autoregressive (NAR) neural machine translation, is to use an NAR sequence generation model for ASR error correction, which, however, comes at the cost of significantly increased ASR error rate. In this paper, observing distinctive error patterns and correction operations (i.e., insertion, deletion, and substitution) in ASR, we propose FastCorrect, a novel NAR error correction model based on edit alignment. In training, FastCorrect aligns each source token from an ASR output sentence to the target tokens from the corresponding ground-truth sentence based on the edit distance between the source and target sentences, and extracts the number of target tokens corresponding to each source token during edition/correction, which is then used to train a length predictor and to adjust the source tokens to match the length of the target sentence for parallel generation. In inference, the token number predicted by the length predictor is used to adjust the source tokens for target sequence generation. Experiments on the public AISHELL-1 dataset and an internal industrial-scale ASR dataset show the effectiveness of FastCorrect for ASR error correction: 1) it speeds up the inference by 6-9 times and maintains the accuracy (8-14% WER reduction) compared with the autoregressive correction model; and 2) it outperforms the popular NAR models adopted in neural machine translation and text edition by a large margin.

Reinforcement learning has become the dominant paradigm for eliciting reasoning and self-correction capabilities in large language models, but its computational expense motivates exploration of alternatives. Inspired by techniques from autonomous driving and robotics, we investigate whether supervised learning with synthetic error injection can induce self-correction abilities in language models. Our approach inserts artificial errors into reasoning chains, masks them, and supervises the model to recognize and correct these mistakes. Despite the intuitive appeal of this method, we find that it fails to significantly improve performance even on simple synthetic tasks across multiple models. Moreover, even when the model catches its own error, it often parrots the original mistake. We find that the distribution shift of synthetic errors to on-policy errors significantly degrades the error-correction capabilities of the fine-tuned model, even with good synthetic coverage of on-policy errors. Our results help explain why on-policy reinforcement learning methods have proven uniquely effective for eliciting self-correction.