In settings where both spurious and causal predictors are available, standard neural networks trained under the objective of empirical risk minimization (ERM) with no additional inductive biases tend to have a dependence on a spurious feature. As a result, it is necessary to integrate additional inductive biases in order to guide the network toward generalizable hypotheses. Often these spurious features are shared across related tasks, such as estimating disease prognoses from image scans coming from different hospitals, making the challenge of generalization more difficult. In these settings, it is important that methods are able to integrate the proper inductive biases to generalize across both nuisance-varying families as well as task families. Motivated by this setting, we present RIME (Robustly Informed Meta lEarning), a new method for meta learning under the presence of both positive and negative inductive biases (what to learn and what not to learn). We first develop a theoretical causal framework showing why existing approaches at knowledge integration can lead to worse performance on distributionally robust objectives. We then show that RIME is able to simultaneously integrate both biases, reaching state of the art performance under distributionally robust objectives in informed meta-learning settings under nuisance-varying families.

Preference-based Reinforcement Learning (PbRL) circumvents the need for reward engineering by harnessing human preferences as the reward signal. However, current PbRL methods excessively depend on high-quality feedback from domain experts, which results in a lack of robustness. In this paper, we present RIME, a robust PbRL algorithm for effective reward learning from noisy preferences. Our method utilizes a sample selection-based discriminator to dynamically filter out noise and ensure robust training. To counteract the cumulative error stemming from incorrect selection, we suggest a warm start for the reward model, which additionally bridges the performance gap during the transition from pre-training to online training in PbRL. Our experiments on robotic manipulation and locomotion tasks demonstrate that RIME significantly enhances the robustness of the state-of-the-art PbRL method. Code is available at https://github.com/CJReinforce/RIME_ICML2024.

Language models (LMs) can generate hallucinations and incoherent outputs, which highlights their weak context dependency. Cache-LMs, which augment LMs with a memory of recent history, can increase context dependency and have shown remarkable performance in diverse language generation tasks. However, we find that even with training, the performance gain stemming from the cache component of current cache-LMs is suboptimal due to the misalignment between the current hidden states and those stored in the memory. In this work, we present HistAlign, a new training approach to ensure good cache alignment such that the model receives useful signals from the history. We first prove our concept on a simple and synthetic task where the memory is essential for correct predictions, and we show that the cache component of HistAlign is better aligned and improves overall performance. Next, we evaluate HistAlign on diverse downstream language generation tasks, including prompt continuation, abstractive summarization, and data-to-text. We demonstrate that HistAlign improves text coherence and faithfulness in open-ended and conditional generation settings respectively. HistAlign is also generalizable across different model families, showcasing its strength in improving context dependency of LMs in diverse scenarios. Our code is publicly available at https://github.com/meetdavidwan/histalign

Many tasks in natural language processing require the extraction of relationship information for a given condition, such as event argument extraction, relation extraction, and task-oriented semantic parsing. Recent works usually propose sophisticated models for each task independently and pay less attention to the commonality of these tasks and to have a unified framework for all the tasks. In this work, we propose to take a unified view of all these tasks and introduce TAGPRIME to address relational structure extraction problems. TAGPRIME is a sequence tagging model that appends priming words about the information of the given condition (such as an event trigger) to the input text. With the self-attention mechanism in pre-trained language models, the priming words make the output contextualized representations contain more information about the given condition, and hence become more suitable for extracting specific relationships for the condition. Extensive experiments and analyses on three different tasks that cover ten datasets across five different languages demonstrate the generality and effectiveness of TAGPRIME.

We describe the constraint modelling tool Savile Row, its input language and its main features. Savile Row translates a solver-independent constraint modelling language to the input languages for various solvers including constraint, SAT, and SMT solvers. After a brief introduction, the manual describes the Essence Prime language, which is the input language of Savile Row. Then we describe the functions of the tool, its main features and options and how to install and use it.

It's time to prep for Apple's next big event and carefully consider what could happen if the government legalizes "hacking back." Tune in Monday.What to expect from Apple at WWDC 2017 This week Apple will put on a big show for developers but, as usual, we're also listening. That's because we could get news about everything from an Echo-fighting Siri speaker to refreshed MacBooks and iPads. On the software side, we're expecting to find out how Siri will keep up with its AI assistant competition, and what's next for both iOS and MacOS. E3 is still more than a week away, but EA has already dropped off one big game announcement: a name, release date and trailer for the next Need for Speed.

From in-depth features and interviews to the daily torrent of trailers and news, we write a lot about video games here. But there's only so much one team can cover, and often some of our favorite games never grace the digital pages of Engadget. To remedy that shortcoming, we're introducing Gaming IRL, a monthly segment where several editors talk about what they've been playing in their downtime. Sometimes these'll be the latest AAA game, but you'll also see free-to-play mobile distraction and classics revisited (or criminally ignored until now). Gaming IRL is part of a broader series in which you'll find stories from all of the areas we cover: gadgets we use every day, the apps and services we adore, what we're watching and the music and podcasts we can't live without.

It takes no less than 45 minutes of playing Tequila Works' upcoming game for their creative director to tell a story from his childhood. It is no aimless reminiscence -- Rime, as Raúl Rubio says, "is about childhood memories. So we put a little bit of ourselves in the game." The Serrano-based studio's upcoming release is a "single-player puzzle adventure game" and has already drawn comparisons to classics like LucasArts adventure games, The Legend of Zelda and projects from Team Ico. Rubio finds these comparisons to industry staples equal parts flattering and terrifying.