Conversational agents (CAs) have become increasingly prominent in education, yet most research focuses on disembodied chatbots for higher education. Because visual representation is central for early childhood, we developed a web application with an LLM-based CA to examine how embodiment and visual appearance relate to children’s dialogic interactions. In an online study with 33 children ages 4–8, participants were randomly assigned to interact with a text-based CA, an embodied CA with a random avatar, or an embodied CA whose appearance children could customize. Children showed high verbal engagement regardless of CA type. Across interactions, children largely remained on topic, though playfulness, disengagement, and social connectedness varied. Interactions with custom avatar CAs more often included playful and disengaged behaviors, while customization patterns suggested that children created avatars resembling themselves or explored fantastical identities. This study provides initial insights into design mechanisms for child-centered pedagogical CAs.

Process reward models (PRMs) play a central role in guiding inference-time scaling algorithms for large language models (LLMs). However, we observe that even state-of-the-art PRMs can be poorly calibrated. Specifically, they tend to overestimate the success probability that a partial reasoning step will lead to a correct final answer, particularly when smaller LLMs are used to complete the reasoning trajectory. To address this, we present a calibration approach—performed via quantile regression—that adjusts PRM outputs to better align with true success probabilities. Leveraging these calibrated success estimates and their associated confidence bounds, we introduce an instance-adaptive scaling (IAS) framework that dynamically adjusts the compute budget based on the estimated likelihood that a partial reasoning trajectory will yield a correct final answer. Unlike conventional methods that allocate a fixed number of reasoning trajectories per query, this approach adapts to each instance and reasoning step when using our calibrated PRMs. Experiments on mathematical reasoning benchmarks show that (i) our PRM calibration method achieves small calibration error, outperforming the baseline methods, (ii) calibration is crucial for enabling effective IAS, and (iii) the proposed IAS strategy reduces inference costs while maintaining final answer accuracy, utilizing less compute on more confident problems as desired.

Self-supervised learning models extract general-purpose representations from data. Quantifying the reliability of these representations is crucial, as many downstream models rely on them as input for their own tasks. To this end, we introduce a formal definition of _representation reliability_: the representation for a given test point is considered to be reliable if the downstream models built on top of that representation can consistently generate accurate predictions for that test point. However, accessing downstream data to quantify the representation reliability is often infeasible or restricted due to privacy concerns. We propose an ensemble-based method for estimating the representation reliability without knowing the downstream tasks a priori. Our method is based on the concept of _neighborhood consistency_ across distinct pre-trained representation spaces. The key insight is to find shared neighboring points as anchors to align these representation spaces before comparing them. We demonstrate through comprehensive numerical experiments that our method effectively captures the representation reliability with a high degree of correlation, achieving robust and favorable performance compared with baseline methods.

Demand forecasting is a crucial component of supply chain management for revenue optimization and inventory planning. Traditional time series forecasting methods, however, have resulted in small models with limited expressive power because they have difficulty in scaling their model size up while maintaining high accuracy. In this paper, we propose Forecasting orchestra (Forchestra), a simple but powerful ensemble framework capable of accurately predicting future demand for a diverse range of items. Forchestra consists of two parts: 1) base predictors and 2) a neural conductor. For a given time series, each base predictor outputs its respective forecast based on historical observations. On top of the base predictors, the neural conductor adaptively assigns the importance weight for each predictor by looking at the representation vector provided by a representation module. Finally, Forchestra aggregates the predictions by the weights and constructs a final prediction. In contrast to previous ensemble approaches, the neural conductor and all base predictors of Forchestra are trained in an end-to-end manner; this allows each base predictor to modify its reaction to different inputs, while supporting other predictors and constructing a final prediction jointly. We empirically show that the model size is scalable to up to 0.8 billion parameters ($\approx$400-layer LSTM). The proposed method is evaluated on our proprietary E-Commerce (100K) and the public M5(30K) datasets, and it outperforms existing forecasting models with a significant margin. In addition, we observe that our framework generalizes well to unseen data points when evaluated in a zeroshot fashion on downstream datasets. Last but not least, we present extensive qualitative and quantitative studies to analyze how the proposed model outperforms baseline models and differs from conventional ensemble approaches. The code is available at https://github.com/young-j-parld22-ICDM-Forchestra.

We present a hierarchical planning and control framework that enables an agent to perform various tasks and adapt to a new task flexibly. Rather than learning an individual policy for each particular task, the proposed framework, DISH, distills a hierarchical policy from a set of tasks by representation and reinforcement learning. The framework is based on the idea of latent variable models that represent high-dimensional observations using low-dimensional latent variables. The resulting policy consists of two levels of hierarchy: (i) a planning module that reasons a sequence of latent intentions that would lead to an optimistic future and (ii) a feedback control policy, shared across the tasks, that executes the inferred intention. Because the planning is performed in low-dimensional latent space, the learned policy can immediately be used to solve or adapt to new tasks without additional training. We demonstrate the proposed framework can learn compact representations (3- and 1-dimensional latent states and commands for a humanoid with 197- and 36-dimensional state features and actions) while solving a small number of imitation tasks, and the resulting policy is directly applicable to other types of tasks, i.e., navigation in cluttered environments.

Temporal set prediction is becoming increasingly important as many companies employ recommender systems in their online businesses, e.g., personalized purchase prediction of shopping baskets. While most previous techniques have focused on leveraging a user’s history, the study of combining it with others’ histories remains untapped potential. This paper proposes Global-Local Item Embedding (GLOIE) that learns to utilize the temporal properties of sets across whole users as well as within a user by coining the names as global and local information to distinguish the two temporal patterns. GLOIE uses Variational Autoencoder (VAE) and dynamic graph-based model to capture global and local information and then applies attention to integrate resulting item embeddings. Additionally, we propose to use Tweedie output for the decoder of VAE as it can easily model zero-inflated and long-tailed distribution, which is more suitable for several real-world data distributions than Gaussian or multinomial counterparts. When evaluated on three public benchmarks, our algorithm consistently outperforms previous state-of-the-art methods in most ranking metrics.

We present a representation learning algorithm that learns a low-dimensional latent dynamical system from high-dimensional sequential raw data, e.g., video. The framework builds upon recent advances in amortized inference methods that use both an inference network and a refinement procedure to output samples from a variational distribution given an observation sequence, and takes advantage of the duality between control and inference to approximately solve the intractable inference problem using the path integral control approach. The learned dynamical model can be used to predict and plan the future states; we also present the efficient planning method that exploits the learned low-dimensional latent dynamics. Numerical experiments show that the proposed path-integral control based variational inference method leads to tighter lower bounds in statistical model learning of sequential data. Supplementary video: https://youtu.be/xCp35crUoLQ

BACKGROUND: Protein-protein interactions (PPIs) play key roles in various cellular functions. In addition, some critical inter-species interactions such as host-pathogen interactions and pathogenicity occur through PPIs. Phytopathogenic bacteria infect hosts through attachment to host tissue, enzyme secretion, exopolysaccharides production, toxins release, iron acquisition, and effector proteins secretion. Many such mechanisms involve some kind of protein-protein interaction in hosts. Our first aim was to predict the whole protein interaction pairs (interactome) of Xanthomonas oryzae pathovar oryzae (Xoo) that is an important pathogenic bacterium that causes bacterial blight (BB) in rice. We developed a detection protocol to find possibly interacting proteins in its host using whole genome PPI prediction algorithms. The second aim was to build a DB server and a bioinformatic procedure for finding target proteins in Xoo for developing pesticides that block host-pathogen protein interactions within critical biochemical pathways. DESCRIPTION: A PPI network in Xoo proteome was predicted by bioinformatics algorithms: PSIMAP, PEIMAP, and iPfam. We present the resultant species specific interaction network and host-pathogen interaction, XooNET. It is a comprehensive predicted initial PPI data for Xoo. XooNET can be used by experimentalists to pick up protein targets for blocking pathological interactions. XooNET uses most of the major types of PPI algorithms. They are: 1) Protein Structural Interactome MAP (PSIMAP), a method using structural domain of SCOP, 2) Protein Experimental Interactome MAP (PEIMAP), a common method using public resources of experimental protein interaction information such as HPRD, BIND, DIP, MINT, IntAct, and BioGrid, and 3) Domain-domain interactions, a method using Pfam domains such as iPfam. Additionally, XooNET provides information on network properties of the Xoo interactome. CONCLUSION: XooNET is an open and free public database server for protein interaction information for Xoo. It contains 4,538 proteins and 26,932 possible interactions consisting of 18,503 (PSIMAP), 3,118 (PEIMAP), and 8,938 (iPfam) pairs. In addition, XooNET provides 3,407 possible interaction pairs between two sets of proteins; 141 Xoo proteins that are predicted as membrane proteins and rice proteomes. The resultant interacting partners of a query protein can be easily retrieved by users as well as the interaction networks in graphical web interfaces. XooNET is freely available from http://bioportal.kobic.kr/XooNET/.