They do not study how the population loss decreases during the first stage.

December 5, 2025 Abstract We argue that language models (LMs) have strong potential as investigative tools for probing the distinction between possible and impossible natural languages and thus uncovering the inductive biases that support human language learning. We outline a phased research program in which LM architectures are iteratively refined to better discriminate between possible and impossible languages, supporting linking hypotheses to human cognition. Which conceivable linguistic systems are possible for humans to learn and use as natural languages? A complete answer to this question would yield profound insights into the human capacity for language. However, our tools for addressing the question are very limited.

This work presents the development and evaluation of an NLP-enabled probabilistic classifier designed to estimate the probability of technical and regulatory success (pTRS) for clinical trials in the field of neuroscience. While pharmaceutical R&D is plagued by high attrition rates and enormous costs, particularly within neuroscience, where success rates are below 10%, timely identification of promising programs can streamline resource allocation and reduce financial risk. Leveraging data from the ClinicalTrials.gov database and success labels from the recently developed Clinical Trial Outcome dataset, the classifier extracts text-based clinical trial features using statistical NLP techniques. These features were integrated into several non-LLM frameworks (logistic regression, gradient boosting, and random forest) to generate calibrated probability scores. Model performance was assessed on a retrospective dataset of 101,145 completed clinical trials spanning 1976-2024, achieving an overall ROC-AUC of 0.64. An LLM-based predictive model was then built using BioBERT, a domain-specific language representation encoder. The BioBERT-based model achieved an overall ROC-AUC of 0.74 and a Brier Score of 0.185, indicating its predictions had, on average, 40% less squared error than would be observed using industry benchmarks. The BioBERT-based model also made trial outcome predictions that were superior to benchmark values 70% of the time overall. By integrating NLP-driven insights into drug development decision-making, this work aims to enhance strategic planning and optimize investment allocation in neuroscience programs.

This paper implements deep reinforcement learning (DRL) for spacecraft reorientation control with a single pointing keep-out zone. The Soft Actor-Critic (SAC) algorithm is adopted to handle continuous state and action space. A new state representation is designed to explicitly include a compact representation of the attitude constraint zone. The reward function is formulated to achieve the control objective while enforcing the attitude constraint. A curriculum learning approach is used for the agent training. Simulation results demonstrate the effectiveness of the proposed DRL-based method for spacecraft pointing-constrained attitude control.

Then each client alternately estimates its cluster labels and refines the model estimates based on FedAvg or FedProx.

Then each client alternately estimates its cluster labels and refines the model estimates based on FedAvg or FedProx.

We thank all reviewers for their detailed feedback. Please see individual responses below. Thank you for your positive comments! "The decoding procedure in Phase 3 is quite elaborate... " In Algorithm 4 line 28, why is noise added to the optimal policy...? " This is closely related to the point above: Since However, since the noise decays with O (ε), the resulting controller is still ε -suboptimal. Most practical systems are only locally linear... How difficult is it to extend this algorithm to the locally-linear " Extending our algorithm to the locally linear setting is a very exciting direction for future research, but we are "The paper lacks any empirical evaluation...With such an experiment this paper "The paper uses "decoder" in place of what in traditional autoencoding... " "...it is certainly not the case that this paper is introducing this as a novel problem... " We will update the abstract to be more precise. "... if F is a family of neural networks, isn't the search space (capacity) |F | increases exponentially?

However, the task is challenging due to the noisy nature of fMRI signals and the intricate pattern of brain visual representations.