Related papers: Training new operators - the first six months

A study on future large accelerators [1] has considered a facility, which is designed, built and operated by a worldwide collaboration of equal partner institutions, and which is remote from most of these institutions. The full range of…

How important are training details and datasets to recent optical flow models like RAFT? And do they generalize? To explore these questions, rather than develop a new model, we revisit three prominent models, PWC-Net, IRR-PWC and RAFT, with…

We present a two-level model of organizational training and agent production. Managers decide whether or not to train based on both the costs of training compared to the benefits and on their expectations and observations of the number of…

A central question in robotics is how to design a control system for an agile mobile robot. This paper studies this question systematically, focusing on a challenging setting: autonomous drone racing. We show that a neural network…

Reinforcement learning (RL) is a promising solution for autonomous vehicles to deal with complex and uncertain traffic environments. The RL training process is however expensive, unsafe, and time consuming. Algorithms are often developed…

We present a new approach for neural machine translation (NMT) using the morphological and grammatical decomposition of the words (factors) in the output side of the neural network. This architecture addresses two main problems occurring in…

First-order reinforcement learning with differentiable simulation is promising for quadrotor control, but practical progress remains fragmented across task-specific settings. To support more systematic development and evaluation, we present…

In order to meet the needs of Fermilabs planned post-collider experimental program, the total proton throughput of the 8 GeV Booster accelerator must be nearly doubled within the next two years. A system of 48 ramped corrector magnets has…

The pretraining-finetuning paradigm has facilitated numerous transformative advancements in artificial intelligence research in recent years. However, in the domain of reinforcement learning (RL) for robot locomotion, individual skills are…

This paper investigates learning effects and human operator training practices in variable autonomy robotic systems. These factors are known to affect performance of a human-robot system and are frequently overlooked. We present the results…

We introduce a new method for training deep Boltzmann machines jointly. Prior methods require an initial learning pass that trains the deep Boltzmann machine greedily, one layer at a time, or do not perform well on classifi- cation tasks.

Large language model (LLM) activations are notoriously difficult to understand, with most existing techniques using complex, specialized methods for interpreting them. Recent work has proposed a simpler approach known as LatentQA: training…

Classifier models are prevalent in natural language processing (NLP), often with high accuracy. Yet in real world settings, human-in-the-loop systems can foster trust in model outputs and even higher performance. Selective Prediction (SP)…

The SiliconMicro-strip Tracker (SMT) at the D{\O} experiment in the Fermilab Tevatron collider has been operating since 2001. In 2006, an additional layer, referred to as 'Layer 0', was installed to improve impact parameter resolution and…

Integrating human knowledge into neural networks has the potential to improve their robustness and interpretability. We have developed a novel approach to integrate knowledge from ontologies into the structure of a Transformer network which…

A language model's ability to reflect on its own reasoning provides a key advantage for solving complex problems. While most recent research has focused on how this ability develops during reinforcement learning, we show that it actually…

One-on-one tutoring is widely acknowledged as an effective instructional method, conditioned on qualified tutors. However, the high demand for qualified tutors remains a challenge, often necessitating the training of novice tutors (i.e.,…

Building AI agents can be costly. Consider a question answering agent such as Jill Watson that automatically answers students' questions on the discussion forums of online classes based on their syllabi and other course materials. Training…

Training efficiency is one of the main problems for Neural Machine Translation (NMT). Deep networks need for very large data as well as many training iterations to achieve state-of-the-art performance. This results in very high computation…

Accelerator beam commissioning is a challenging and exciting period. It is generally the first integrated operation of the many systems in an accelerator and, most importantly, of the beam. First, general preparation is discussed. Then…