Large language models (LLMs) have demonstrated impressive capabilities in various natural language processing tasks. Despite this, their application to information retrieval (IR) tasks is still challenging due to the infrequent occurrence of many IR-specific concepts in natural language. While prompt-based methods can provide task descriptions to LLMs, they often fall short in facilitating a comprehensive understanding and execution of IR tasks, thereby limiting LLMs' applicability. To address this gap, in this work, we explore the potential of instruction tuning to enhance LLMs' proficiency in IR tasks. We introduce a novel instruction tuning dataset, INTERS, encompassing 20 tasks across three fundamental IR categories: query understanding, document understanding, and query-document relationship understanding. The data are derived from 43 distinct datasets with manually written templates. Our empirical results reveal that INTERS significantly boosts the performance of various publicly available LLMs, such as LLaMA, Mistral, and Phi, in IR tasks. Furthermore, we conduct extensive experiments to analyze the effects of instruction design, template diversity, few-shot demonstrations, and the volume of instructions on performance. We make our dataset and the fine-tuned models publicly accessible at https://github.com/DaoD/INTERS.
Urban mobility is on the cusp of transformation with the emergence of shared, connected, and cooperative automated vehicles. Yet, for them to be accepted by customers, trust in their punctuality is vital. Many pilot initiatives operate without a fixed schedule, thus enhancing the importance of reliable arrival time (AT) predictions. This study presents an AT prediction system for autonomous shuttles, utilizing separate models for dwell and running time predictions, validated on real-world data from five cities. Alongside established methods such as XGBoost, we explore the benefits of integrating spatial data using graph neural networks (GNN). To accurately handle the case of a shuttle bypassing a stop, we propose a hierarchical model combining a random forest classifier and a GNN. The results for the final AT prediction are promising, showing low errors even when predicting several stops ahead. Yet, no single model emerges as universally superior, and we provide insights into the characteristics of pilot sites that influence the model selection process. Finally, we identify dwell time prediction as the key determinant in overall AT prediction accuracy when autonomous shuttles are deployed in low-traffic areas or under regulatory speed limits. This research provides insights into the current state of autonomous public transport prediction models and paves the way for more data-informed decision-making as the field advances.
We study the problem of experiment planning with function approximation in contextual bandit problems. In settings where there is a significant overhead to deploying adaptive algorithms -- for example, when the execution of the data collection policies is required to be distributed, or a human in the loop is needed to implement these policies -- producing in advance a set of policies for data collection is paramount. We study the setting where a large dataset of contexts but not rewards is available and may be used by the learner to design an effective data collection strategy. Although when rewards are linear this problem has been well studied, results are still missing for more complex reward models. In this work we propose two experiment planning strategies compatible with function approximation. The first is an eluder planning and sampling procedure that can recover optimality guarantees depending on the eluder dimension of the reward function class. For the second, we show that a uniform sampler achieves competitive optimality rates in the setting where the number of actions is small. We finalize our results introducing a statistical gap fleshing out the fundamental differences between planning and adaptive learning and provide results for planning with model selection.
In recent years, various well-designed algorithms have empowered music platforms to provide content based on one's preferences. Music genres are defined through various aspects, including acoustic features and cultural considerations. Music genre classification works well with content-based filtering, which recommends content based on music similarity to users. Given a considerable dataset, one premise is automatic annotation using machine learning or deep learning methods that can effectively classify audio files. The effectiveness of systems largely depends on feature and model selection, as different architectures and features can facilitate each other and yield different results. In this study, we conduct a comparative study investigating the performances of three models: a proposed convolutional neural network (CNN), the VGG16 with fully connected layers (FC), and an eXtreme Gradient Boosting (XGBoost) approach on different features: 30-second Mel spectrogram and 3-second Mel-frequency cepstral coefficients (MFCCs). The results show that the MFCC XGBoost model outperformed the others. Furthermore, applying data segmentation in the data preprocessing phase can significantly enhance the performance of the CNNs.
This study investigates the short-term asymptotic behavior of the implied volatility surface (IVS), with a particular focus on the at-the-money (ATM) skew and curvature, which are key determinants of the IVS shape and whose are widely concerned in the option market. Departing from conventional process-based models that rely on explicit stochastic differential equations, our work adopts a model-independent, distribution-based approach by imposing cumulant conditions on the asset's log return distribution. Under these broad assumptions, we derive a quadratic expansion for the implied volatility as a function of moneyness for near-the-money options, along with asymptotic expressions for the ATM skew and curvature as the time to maturity tends to zero. These results not only highlight the differences in the properties of ATM asymptotics among different types of models, but also offer a versatile framework for model calibration, including regular stochastic volatility, rough volatility, and discrete-time distribution-based models, using simple moment information. Overall, our findings provide robust analytical tools for the accurate approximation of short-term option prices and the evaluation of model performance against market stylized features.
Recent years have witnessed a surge in the popularity of Machine Learning (ML), applied across diverse domains. However, progress is impeded by the scarcity of training data due to expensive acquisition and privacy legislation. Synthetic data emerges as a solution, but the abundance of released models and limited overview literature pose challenges for decision-making. This work surveys 417 Synthetic Data Generation (SDG) models over the last decade, providing a comprehensive overview of model types, functionality, and improvements. Common attributes are identified, leading to a classification and trend analysis. The findings reveal increased model performance and complexity, with neural network-based approaches prevailing, except for privacy-preserving data generation. Computer vision dominates, with GANs as primary generative models, while diffusion models, transformers, and RNNs compete. Implications from our performance evaluation highlight the scarcity of common metrics and datasets, making comparisons challenging. Additionally, the neglect of training and computational costs in literature necessitates attention in future research. This work serves as a guide for SDG model selection and identifies crucial areas for future exploration.
This paper investigates an under-explored but important problem: given a collection of pre-trained neural networks, predicting their performance on each multi-modal task without fine-tuning them, such as image recognition, referring, captioning, visual question answering, and text question answering. A brute-force approach is to finetune all models on all target datasets, bringing high computational costs. Although recent-advanced approaches employed lightweight metrics to measure models' transferability,they often depend heavily on the prior knowledge of a single task, making them inapplicable in a multi-modal multi-task scenario. To tackle this issue, we propose an efficient multi-task model selector (EMMS), which employs large-scale foundation models to transform diverse label formats such as categories, texts, and bounding boxes of different downstream tasks into a unified noisy label embedding. EMMS can estimate a model's transferability through a simple weighted linear regression, which can be efficiently solved by an alternating minimization algorithm with a convergence guarantee. Extensive experiments on 5 downstream tasks with 24 datasets show that EMMS is fast, effective, and generic enough to assess the transferability of pre-trained models, making it the first model selection method in the multi-task scenario. For instance, compared with the state-of-the-art method LogME enhanced by our label embeddings, EMMS achieves 9.0\%, 26.3\%, 20.1\%, 54.8\%, 12.2\% performance gain on image recognition, referring, captioning, visual question answering, and text question answering, while bringing 5.13x, 6.29x, 3.59x, 6.19x, and 5.66x speedup in wall-clock time, respectively. The code is available at https://github.com/OpenGVLab/Multitask-Model-Selector.
With growing concerns regarding bias and discrimination in predictive models, the AI community has increasingly focused on assessing AI system trustworthiness. Conventionally, trustworthy AI literature relies on the probabilistic framework and calibration as prerequisites for trustworthiness. In this work, we depart from this viewpoint by proposing a novel trust framework inspired by the philosophy literature on trust. We present a precise mathematical definition of trustworthiness, termed $\mathcal{U}$-trustworthiness, specifically tailored for a subset of tasks aimed at maximizing a utility function. We argue that a model's $\mathcal{U}$-trustworthiness is contingent upon its ability to maximize Bayes utility within this task subset. Our first set of results challenges the probabilistic framework by demonstrating its potential to favor less trustworthy models and introduce the risk of misleading trustworthiness assessments. Within the context of $\mathcal{U}$-trustworthiness, we prove that properly-ranked models are inherently $\mathcal{U}$-trustworthy. Furthermore, we advocate for the adoption of the AUC metric as the preferred measure of trustworthiness. By offering both theoretical guarantees and experimental validation, AUC enables robust evaluation of trustworthiness, thereby enhancing model selection and hyperparameter tuning to yield more trustworthy outcomes.
Recent years have seen a lot of progress in algorithms for learning parameters of spreading dynamics from both full and partial data. Some of the remaining challenges include model selection under the scenarios of unknown network structure, noisy data, missing observations in time, as well as an efficient incorporation of prior information to minimize the number of samples required for an accurate learning. Here, we introduce a universal learning method based on scalable dynamic message-passing technique that addresses these challenges often encountered in real data. The algorithm leverages available prior knowledge on the model and on the data, and reconstructs both network structure and parameters of a spreading model. We show that a linear computational complexity of the method with the key model parameters makes the algorithm scalable to large network instances.
Recommender systems aim to recommend the most suitable items to users from a large number of candidates. Their computation cost grows as the number of user requests and the complexity of services (or models) increases. Under the limitation of computation resources (CRs), how to make a trade-off between computation cost and business revenue becomes an essential question. The existing studies focus on dynamically allocating CRs in queue truncation scenarios (i.e., allocating the size of candidates), and formulate the CR allocation problem as an optimization problem with constraints. Some of them focus on single-phase CR allocation, and others focus on multi-phase CR allocation but introduce some assumptions about queue truncation scenarios. However, these assumptions do not hold in other scenarios, such as retrieval channel selection and prediction model selection. Moreover, existing studies ignore the state transition process of requests between different phases, limiting the effectiveness of their approaches. This paper proposes a Reinforcement Learning (RL) based Multi-Phase Computation Allocation approach (RL-MPCA), which aims to maximize the total business revenue under the limitation of CRs. RL-MPCA formulates the CR allocation problem as a Weakly Coupled MDP problem and solves it with an RL-based approach. Specifically, RL-MPCA designs a novel deep Q-network to adapt to various CR allocation scenarios, and calibrates the Q-value by introducing multiple adaptive Lagrange multipliers (adaptive-$\lambda$) to avoid violating the global CR constraints. Finally, experiments on the offline simulation environment and online real-world recommender system validate the effectiveness of our approach.
Multi-modal (vision-language) models, such as CLIP, are replacing traditional supervised pre-training models (e.g., ImageNet-based pre-training) as the new generation of visual foundation models. These models with robust and aligned semantic representations learned from billions of internet image-text pairs and can be applied to various downstream tasks in a zero-shot manner. However, in some fine-grained domains like medical imaging and remote sensing, the performance of multi-modal foundation models often leaves much to be desired. Consequently, many researchers have begun to explore few-shot adaptation methods for these models, gradually deriving three main technical approaches: 1) prompt-based methods, 2) adapter-based methods, and 3) external knowledge-based methods. Nevertheless, this rapidly developing field has produced numerous results without a comprehensive survey to systematically organize the research progress. Therefore, in this survey, we introduce and analyze the research advancements in few-shot adaptation methods for multi-modal models, summarizing commonly used datasets and experimental setups, and comparing the results of different methods. In addition, due to the lack of reliable theoretical support for existing methods, we derive the few-shot adaptation generalization error bound for multi-modal models. The theorem reveals that the generalization error of multi-modal foundation models is constrained by three factors: domain gap, model capacity, and sample size. Based on this, we propose three possible solutions from the following aspects: 1) adaptive domain generalization, 2) adaptive model selection, and 3) adaptive knowledge utilization.
Large language models (LLMs) have achieved huge success for their general knowledge and ability to solve a wide spectrum of tasks in natural language processing (NLP). Due to their impressive abilities, LLMs have shed light on potential inter-discipline applications to foster scientific discoveries of a specific domain by using artificial intelligence (AI for science, AI4S). In the meantime, utilizing NLP techniques in geoscience research and practice is wide and convoluted, contributing from knowledge extraction and document classification to question answering and knowledge discovery. In this work, we take the initial step to leverage LLM for science, through a rather straightforward approach. We try to specialize an LLM into geoscience, by further pre-training the model with a vast amount of texts in geoscience, as well as supervised fine-tuning (SFT) the resulting model with our custom collected instruction tuning dataset. These efforts result in a model GeoGalactica consisting of 30 billion parameters. To our best knowledge, it is the largest language model for the geoscience domain. More specifically, GeoGalactica is from further pre-training of Galactica. We train GeoGalactica over a geoscience-related text corpus containing 65 billion tokens, preserving as the largest geoscience-specific text corpus. Then we fine-tune the model with 1 million pairs of instruction-tuning data consisting of questions that demand professional geoscience knowledge to answer. In this technical report, we will illustrate in detail all aspects of GeoGalactica, including data collection, data cleaning, base model selection, pre-training, SFT, and evaluation. We open-source our data curation tools and the checkpoints of GeoGalactica during the first 3/4 of pre-training.
Tree-based models have been successfully applied to a wide variety of tasks, including time series forecasting. They are increasingly in demand and widely accepted because of their comparatively high level of interpretability. However, many of them suffer from the overfitting problem, which limits their application in real-world decision-making. This problem becomes even more severe in online-forecasting settings where time series observations are incrementally acquired, and the distributions from which they are drawn may keep changing over time. In this context, we propose a novel method for the online selection of tree-based models using the TreeSHAP explainability method in the task of time series forecasting. We start with an arbitrary set of different tree-based models. Then, we outline a performance-based ranking with a coherent design to make TreeSHAP able to specialize the tree-based forecasters across different regions in the input time series. In this framework, adequate model selection is performed online, adaptively following drift detection in the time series. In addition, explainability is supported on three levels, namely online input importance, model selection, and model output explanation. An extensive empirical study on various real-world datasets demonstrates that our method achieves excellent or on-par results in comparison to the state-of-the-art approaches as well as several baselines.
Devising procedures for downstream task-oriented generative model selections is an unresolved problem of practical importance. Existing studies focused on the utility of a single family of generative models. They provided limited insights on how synthetic data practitioners select the best family generative models for synthetic training tasks given a specific combination of machine learning model class and performance metric. In this paper, we approach the downstream task-oriented generative model selections problem in the case of training fraud detection models and investigate the best practice given different combinations of model interpretability and model performance constraints. Our investigation supports that, while both Neural Network(NN)-based and Bayesian Network(BN)-based generative models are both good to complete synthetic training task under loose model interpretability constrain, the BN-based generative models is better than NN-based when synthetic training fraud detection model under strict model interpretability constrain. Our results provides practical guidance for machine learning practitioner who is interested in replacing their training dataset from real to synthetic, and shed lights on more general downstream task-oriented generative model selection problems.
Structured data in the form of tabular datasets contain features that are distinct and discrete, with varying individual and relative importances to the target. Combinations of one or more features may be more predictive and meaningful than simple individual feature contributions. R's mixed effect linear models library allows users to provide such interactive feature combinations in the model design. However, given many features and possible interactions to select from, model selection becomes an exponentially difficult task. We aim to automate the model selection process for predictions on tabular datasets incorporating feature interactions while keeping computational costs small. The framework includes two distinct approaches for feature selection: a Priority-based Random Grid Search and a Greedy Search method. The Priority-based approach efficiently explores feature combinations using prior probabilities to guide the search. The Greedy method builds the solution iteratively by adding or removing features based on their impact. Experiments on synthetic demonstrate the ability to effectively capture predictive feature combinations.
Despite numerous years of research into the merits and trade-offs of various model selection criteria, obtaining robust results that elucidate the behavior of cross-validation remains a challenging endeavor. In this paper, we highlight the inherent limitations of cross-validation when employed to discern the structure of a Gaussian graphical model. We provide finite-sample bounds on the probability that the Lasso estimator for the neighborhood of a node within a Gaussian graphical model, optimized using a prediction oracle, misidentifies the neighborhood. Our results pertain to both undirected and directed acyclic graphs, encompassing general, sparse covariance structures. To support our theoretical findings, we conduct an empirical investigation of this inconsistency by contrasting our outcomes with other commonly used information criteria through an extensive simulation study. Given that many algorithms designed to learn the structure of graphical models require hyperparameter selection, the precise calibration of this hyperparameter is paramount for accurately estimating the inherent structure. Consequently, our observations shed light on this widely recognized practical challenge.
The challenge in biomarker discovery using machine learning from omics data lies in the abundance of molecular features but scarcity of samples. Most feature selection methods in machine learning require evaluating various sets of features (models) to determine the most effective combination. This process, typically conducted using a validation dataset, involves testing different feature sets to optimize the model's performance. Evaluations have performance estimation error and when the selection involves many models the best ones are almost certainly overestimated. Biomarker identification with feature selection methods can be addressed as a multi-objective problem with trade-offs between predictive ability and parsimony in the number of features. Genetic algorithms are a popular tool for multi-objective optimization but they evolve numerous solutions thus are prone to overestimation. Methods have been proposed to reduce the overestimation after a model has already been selected in single-objective problems, but no algorithm existed capable of reducing the overestimation during the optimization, improving model selection, or applied in the more general multi-objective domain. We propose DOSA-MO, a novel multi-objective optimization wrapper algorithm that learns how the original estimation, its variance, and the feature set size of the solutions predict the overestimation. DOSA-MO adjusts the expectation of the performance during the optimization, improving the composition of the solution set. We verify that DOSA-MO improves the performance of a state-of-the-art genetic algorithm on left-out or external sample sets, when predicting cancer subtypes and/or patient overall survival, using three transcriptomics datasets for kidney and breast cancer.
Inverse reinforcement learning (IRL) usually assumes the reward function model is pre-specified as a weighted sum of features and estimates the weighting parameters only. However, how to select features and determine a proper reward model is nontrivial and experience-dependent. A simplistic model is less likely to contain the ideal reward function, while a model with high complexity leads to substantial computation cost and potential overfitting. This paper addresses this trade-off in the model selection for IRL problems by introducing the structural risk minimization (SRM) framework from statistical learning. SRM selects an optimal reward function class from a hypothesis set minimizing both estimation error and model complexity. To formulate an SRM scheme for IRL, we estimate the policy gradient from given demonstration as the empirical risk, and establish the upper bound of Rademacher complexity as the model penalty of hypothesis function classes. The SRM learning guarantee is further presented. In particular, we provide the explicit form for the linear weighted sum setting. Simulations demonstrate the performance and efficiency of our algorithm.
Neural marked temporal point processes have been a valuable addition to the existing toolbox of statistical parametric models for continuous-time event data. These models are useful for sequences where each event is associated with a single item (a single type of event or a "mark") -- but such models are not suited for the practical situation where each event is associated with a set of items. In this work, we develop a general framework for modeling set-valued data in continuous-time, compatible with any intensity-based recurrent neural point process model. In addition, we develop inference methods that can use such models to answer probabilistic queries such as "the probability of item $A$ being observed before item $B$," conditioned on sequence history. Computing exact answers for such queries is generally intractable for neural models due to both the continuous-time nature of the problem setting and the combinatorially-large space of potential outcomes for each event. To address this, we develop a class of importance sampling methods for querying with set-based sequences and demonstrate orders-of-magnitude improvements in efficiency over direct sampling via systematic experiments with four real-world datasets. We also illustrate how to use this framework to perform model selection using likelihoods that do not involve one-step-ahead prediction.
Human annotations are vital to supervised learning, yet annotators often disagree on the correct label, especially as annotation tasks increase in complexity. A strategy to improve label quality is to ask multiple annotators to label the same item and aggregate their labels. Many aggregation models have been proposed for categorical or numerical annotation tasks, but far less work has considered more complex annotation tasks involving open-ended, multivariate, or structured responses. While a variety of bespoke models have been proposed for specific tasks, our work is the first to introduce aggregation methods that generalize across many diverse complex tasks, including sequence labeling, translation, syntactic parsing, ranking, bounding boxes, and keypoints. This generality is achieved by devising a task-agnostic method to model distances between labels rather than the labels themselves. This article extends our prior work with investigation of three new research questions. First, how do complex annotation properties impact aggregation accuracy? Second, how should a task owner navigate the many modeling choices to maximize aggregation accuracy? Finally, what diagnoses can verify that aggregation models are specified correctly for the given data? To understand how various factors impact accuracy and to inform model selection, we conduct simulation studies and experiments on real, complex datasets. Regarding testing, we introduce unit tests for aggregation models and present a suite of such tests to ensure that a given model is not mis-specified and exhibits expected behavior. Beyond investigating these research questions above, we discuss the foundational concept of annotation complexity, present a new aggregation model as a bridge between traditional models and our own, and contribute a new semi-supervised learning method for complex label aggregation that outperforms prior work.