Zhen Zeng

Hanoz Bhathena, Parin Rajesh Jhaveri, Rohan Mittal et al.

Recent advances in multimodal retrieval-augmented generation (MM-RAG) have shifted toward minimal parsing, relying on page-level images for producing retriever embeddings and for answer generation. While efficient, this trend often neglects explicit handling of the rich, structured information in complex enterprise documents, instead depending on pre-trained embeddings or vision-language models to implicitly capture such structure. In this work, we take a more direct approach: MM-BizRAG proactively extracts and represents document structure via a document structure-aware split that dynamically routes documents through orientation-specific ingestion pipelines, applying explicit layout-aware parsing for vertically structured documents (e.g., reports) and holistic page-level representations for horizontally structured documents (e.g., slide decks). A unified LLM-driven artifact transformation pipeline with placeholder-based positional alignment preserves natural reading order, while inference-time multimodal assembly decouples retrieval representations from generation context, enabling richer, more grounded answers without any finetuning requirement. Through experiments on a large, heterogeneous enterprise dataset and two public benchmarks (SlideVQA and FinRAGBench-V), MM-BizRAG consistently outperforms state-of-the-art vision-centric baselines by up to 32% points, with especially strong gains on report-style layouts. Furthermore, we introduce FastRAGEval, a single-call LLM Judge metric for fine-grained generative recall that halves RAGChecker's cost while achieving stronger human alignment.

Leijiang Gu, Zhen Zeng, Feng Li et al.

Existing methods in Multimodal Knowledge Editing (MKE) have advanced the ability to correct outdated or inaccurate knowledge in Multimodal Large Language Models (MLLMs). However, they exhibit a critical limitation: while effectively modifying target factual pairs, they fail to generalize edits to logically related queries and often cause unintended alterations to unrelated but visually or semantically linked information. We identify and formalize two underlying failure modes causing this issue: Causal Misalignment, which confines edits to the specific sample, and Feature Entanglement, which causes unintended alterations to coupled but irrelevant information. To address these issues, we propose Localized and Disentangled Knowledge Editing (LDKE), a new framework that achieves precise and generalized editing by localizing fact-specific model layers and disentangling target-relevant inputs from irrelevant ones. Our approach introduces a Fast Localization module to identify and update critical layers efficiently, along with a Disentanglement Classifier that routes inputs appropriately to preserve unrelated knowledge. Extensive experiments across various benchmarks and MLLMs demonstrate that LDKE achieves superior performance in propagating edits to related contexts while maintaining high locality.

Zhen Zeng, Rachneet Kaur, Suchetha Siddagangappa et al.

Time series forecasting is important across various domains for decision-making. In particular, financial time series such as stock prices can be hard to predict as it is difficult to model short-term and long-term temporal dependencies between data points. Convolutional Neural Networks (CNN) are good at capturing local patterns for modeling short-term dependencies. However, CNNs cannot learn long-term dependencies due to the limited receptive field. Transformers on the other hand are capable of learning global context and long-term dependencies. In this paper, we propose to harness the power of CNNs and Transformers to model both short-term and long-term dependencies within a time series, and forecast if the price would go up, down or remain the same (flat) in the future. In our experiments, we demonstrated the success of the proposed method in comparison to commonly adopted statistical and deep learning methods on forecasting intraday stock price change of S&P 500 constituents.

Huaizhen Tang, Xulong Zhang, Jianzong Wang et al.

Non-parallel many-to-many voice conversion remains an interesting but challenging speech processing task. Recently, AutoVC, a conditional autoencoder based method, achieved excellent conversion results by disentangling the speaker identity and the speech content using information-constraining bottlenecks. However, due to the pure autoencoder training method, it is difficult to evaluate the separation effect of content and speaker identity. In this paper, a novel voice conversion framework, named $\boldsymbol T$ext $\boldsymbol G$uided $\boldsymbol A$utoVC(TGAVC), is proposed to more effectively separate content and timbre from speech, where an expected content embedding produced based on the text transcriptions is designed to guide the extraction of voice content. In addition, the adversarial training is applied to eliminate the speaker identity information in the estimated content embedding extracted from speech. Under the guidance of the expected content embedding and the adversarial training, the content encoder is trained to extract speaker-independent content embedding from speech. Experiments on AIShell-3 dataset show that the proposed model outperforms AutoVC in terms of naturalness and similarity of converted speech.

Yiqiao Jin, Rachneet Kaur, Zhen Zeng et al.

Multi-page visual documents such as manuals, brochures, presentations, and posters convey key information through layout, colors, icons, and cross-slide references. While large language models (LLMs) offer opportunities in document understanding, current systems struggle with complex, multi-page visual documents, particularly in fine-grained reasoning over elements and pages. We introduce SlideAgent, a versatile agentic framework for understanding multi-modal, multi-page, and multi-layout documents, especially slide decks. SlideAgent employs specialized agents and decomposes reasoning into three specialized levels-global, page, and element-to construct a structured, query-agnostic representation that captures both overarching themes and detailed visual or textual cues. During inference, SlideAgent selectively activates specialized agents for multi-level reasoning and integrates their outputs into coherent, context-aware answers. Extensive experiments show that SlideAgent achieves significant improvement over both proprietary (+7.9 overall) and open-source models (+9.8 overall).

Alexander Bienstock, Daniel Escudero, Antigoni Polychroniadou et al.

The prevalence of biometric authentication has been on the rise due to its ease of use and elimination of weak passwords. To date, most biometric authentication systems have been designed for on-device authentication of the device owner (e.g., smartphones and laptops). Recently, biometric authentication systems have started to emerge that are designed to authenticate users against cloud databases storing representations of biometrics for large numbers of users (potentially millions), such as those facilitating biometric payments. However, the use of a large cloud database introduces a significant attack vector, as a breach of the database could lead to the compromise of all enrolled users' sensitive biometric data. Indeed, all such existing systems either do not adequately protect against such a breach, or are impractical to deploy and use due to their high computational overhead. In this work, we present a new biometric authentication system that provides provable security guarantees against data breaches, while remaining scalable and performant. To do so, we marry artificial intelligence with advanced cryptographic techniques in a novel fashion, providing several optimizations along the way. Our work is the first to show that real-world scalable privacy-preserving biometric authentication without auxiliary identifiers is feasible, and we believe that it will spur widespread industrial adoption and further research in this area.

Shanshan Wang, Zhen Zeng, Xun Yang et al.

Cognitive diagnosis is a fundamental yet critical research task in the field of intelligent education, which aims to discover the proficiency level of different students on specific knowledge concepts. Despite the effectiveness of existing efforts, previous methods always considered the mastery level on the whole students, so they still suffer from the Long Tail Effect. A large number of students who have sparse data are performed poorly in the model. To relieve the situation, we proposed a Self-supervised Cognitive Diagnosis (SCD) framework which leverages the self-supervised manner to assist the graph-based cognitive diagnosis, then the performance on those students with sparse data can be improved. Specifically, we came up with a graph confusion method that drops edges under some special rules to generate different sparse views of the graph. By maximizing the consistency of the representation on the same node under different views, the model could be more focused on long-tailed students. Additionally, we proposed an importance-based view generation rule to improve the influence of long-tailed students. Extensive experiments on real-world datasets show the effectiveness of our approach, especially on the students with sparse data.

Rachneet Kaur, Zhen Zeng, Tucker Balch et al.

Recent advancements in language modeling have shown promising results when applied to time series data. In particular, fine-tuning pre-trained large language models (LLMs) for time series classification tasks has achieved state-of-the-art (SOTA) performance on standard benchmarks. However, these LLM-based models have a significant drawback due to the large model size, with the number of trainable parameters in the millions. In this paper, we propose an alternative approach to leveraging the success of language modeling in the time series domain. Instead of fine-tuning LLMs, we utilize a text embedding model to embed time series and then pair the embeddings with a simple classification head composed of convolutional neural networks (CNN) and multilayer perceptron (MLP). We conducted extensive experiments on a well-established time series classification benchmark. We demonstrated LETS-C not only outperforms the current SOTA in classification accuracy but also offers a lightweight solution, using only 14.5% of the trainable parameters on average compared to the SOTA model. Our findings suggest that leveraging text embedding models to encode time series data, combined with a simple yet effective classification head, offers a promising direction for achieving high-performance time series classification while maintaining a lightweight model architecture.

Zhen Zeng, Leijiang Gu, Feng Li et al.

Multimodal Large Language Models (MLLMs), trained primarily on English-centric data, frequently generate culturally inappropriate or misaligned responses in cross-cultural settings. To mitigate this, we introduce the task of cross-cultural knowledge insertion, which focuses on adapting models to specific cultural contexts while preserving their original behavior in other cultures. To facilitate research in this area, we introduce CrossCult-KIBench, a comprehensive evaluation benchmark for assessing both the effectiveness of knowledge insertion and its unintended side effects on non-target cultures. The benchmark includes 9,800 image-grounded cases covering 49 culturally relevant visual scenarios across English, Chinese, and Arabic language-culture groups. It supports evaluation in both single-insert and sequential-insert settings. We also propose Memory-Conditioned Knowledge Insertion (MCKI) as a baseline method. MCKI retrieves relevant cultural knowledge from an external memory using frozen MLLM representations, prepending matched entries as conditional prompts when applicable. Extensive experiments on CrossCult-KIBench reveal that current approaches struggle to balance effective cultural adaptation with behavioral preservation, highlighting a key challenge in developing culturally-aware MLLMs. Our work thus underscores an important research direction for developing more culturally adaptive and responsible MLLMs.

Thomas Cook, Kelly Patel, Sivapriya Vellaichamy et al.

Large Language Models (LLMs) can generate SQL queries from natural language questions but struggle with database-specific schemas and tacit domain knowledge. We introduce a framework for continual learning from human feedback in text-to-SQL, where a learning agent receives natural language feedback to refine queries and distills the revealed knowledge for reuse on future tasks. This distilled knowledge is stored in a structured memory, enabling the agent to improve execution accuracy over time. We design and evaluate multiple variations of a learning agent architecture that vary in how they capture and retrieve past experiences. Experiments on the BIRD benchmark Dev set show that memory-augmented agents, particularly the Procedural Agent, achieve significant accuracy gains and error reduction by leveraging human-in-the-loop feedback. Our results highlight the importance of transforming tacit human expertise into reusable knowledge, paving the way for more adaptive, domain-aware text-to-SQL systems that continually learn from a human-in-the-loop.

Zhen Zeng, William Watson, Nicole Cho et al.

The rapidly evolving field of Robotic Process Automation (RPA) has made significant strides in automating repetitive processes, yet its effectiveness diminishes in scenarios requiring spontaneous or unpredictable tasks demanded by users. This paper introduces a novel approach, FlowMind, leveraging the capabilities of Large Language Models (LLMs) such as Generative Pretrained Transformer (GPT), to address this limitation and create an automatic workflow generation system. In FlowMind, we propose a generic prompt recipe for a lecture that helps ground LLM reasoning with reliable Application Programming Interfaces (APIs). With this, FlowMind not only mitigates the common issue of hallucinations in LLMs, but also eliminates direct interaction between LLMs and proprietary data or code, thus ensuring the integrity and confidentiality of information - a cornerstone in financial services. FlowMind further simplifies user interaction by presenting high-level descriptions of auto-generated workflows, enabling users to inspect and provide feedback effectively. We also introduce NCEN-QA, a new dataset in finance for benchmarking question-answering tasks from N-CEN reports on funds. We used NCEN-QA to evaluate the performance of workflows generated by FlowMind against baseline and ablation variants of FlowMind. We demonstrate the success of FlowMind, the importance of each component in the proposed lecture recipe, and the effectiveness of user interaction and feedback in FlowMind.

Elizabeth Fons, Rachneet Kaur, Soham Palande et al.

Large Language Models (LLMs) offer the potential for automatic time series analysis and reporting, which is a critical task across many domains, spanning healthcare, finance, climate, energy, and many more. In this paper, we propose a framework for rigorously evaluating the capabilities of LLMs on time series understanding, encompassing both univariate and multivariate forms. We introduce a comprehensive taxonomy of time series features, a critical framework that delineates various characteristics inherent in time series data. Leveraging this taxonomy, we have systematically designed and synthesized a diverse dataset of time series, embodying the different outlined features, each accompanied by textual descriptions. This dataset acts as a solid foundation for assessing the proficiency of LLMs in comprehending time series. Our experiments shed light on the strengths and limitations of state-of-the-art LLMs in time series understanding, revealing which features these models readily comprehend effectively and where they falter. In addition, we uncover the sensitivity of LLMs to factors including the formatting of the data, the position of points queried within a series and the overall time series length.

Zhen Zeng, Rachneet Kaur, Suchetha Siddagangappa et al.

Time series forecasting plays a crucial role in decision-making across various domains, but it presents significant challenges. Recent studies have explored image-driven approaches using computer vision models to address these challenges, often employing lineplots as the visual representation of time series data. In this paper, we propose a novel approach that uses time-frequency spectrograms as the visual representation of time series data. We introduce the use of a vision transformer for multimodal learning, showcasing the advantages of our approach across diverse datasets from different domains. To evaluate its effectiveness, we compare our method against statistical baselines (EMA and ARIMA), a state-of-the-art deep learning-based approach (DeepAR), other visual representations of time series data (lineplot images), and an ablation study on using only the time series as input. Our experiments demonstrate the benefits of utilizing spectrograms as a visual representation for time series data, along with the advantages of employing a vision transformer for simultaneous learning in both the time and frequency domains.

Gaurav Verma, Rachneet Kaur, Nishan Srishankar et al. · gatech

State-of-the-art multimodal web agents, powered by Multimodal Large Language Models (MLLMs), can autonomously execute many web tasks by processing user instructions and interacting with graphical user interfaces (GUIs). Current strategies for building web agents rely on (i) the generalizability of underlying MLLMs and their steerability via prompting, and (ii) large-scale fine-tuning of MLLMs on web-related tasks. However, web agents still struggle to automate tasks on unseen websites and domains, limiting their applicability to enterprise-specific and proprietary platforms. Beyond generalization from large-scale pre-training and fine-tuning, we propose building agents for few-shot adaptability using human demonstrations. We introduce the AdaptAgent framework that enables both proprietary and open-weights multimodal web agents to adapt to new websites and domains using few human demonstrations (up to 2). Our experiments on two popular benchmarks -- Mind2Web & VisualWebArena -- show that using in-context demonstrations (for proprietary models) or meta-adaptation demonstrations (for meta-learned open-weights models) boosts task success rate by 3.36% to 7.21% over non-adapted state-of-the-art models, corresponding to a relative increase of 21.03% to 65.75%. Furthermore, our additional analyses (a) show the effectiveness of multimodal demonstrations over text-only ones, (b) shed light on the influence of different data selection strategies during meta-learning on the generalization of the agent, and (c) demonstrate the effect of number of few-shot examples on the web agent's success rate. Overall, our results unlock a complementary axis for developing widely applicable multimodal web agents beyond large-scale pre-training and fine-tuning, emphasizing few-shot adaptability.

William Watson, Nicole Cho, Nishan Srishankar et al.

Legal contracts in the custody and fund services domain govern critical aspects such as key provider responsibilities, fee schedules, and indemnification rights. However, it is challenging for an off-the-shelf Large Language Model (LLM) to ingest these contracts due to the lengthy unstructured streams of text, limited LLM context windows, and complex legal jargon. To address these challenges, we introduce LAW (Legal Agentic Workflows for Custody and Fund Services Contracts). LAW features a modular design that responds to user queries by orchestrating a suite of domain-specific tools and text agents. Our experiments demonstrate that LAW, by integrating multiple specialized agents and tools, significantly outperforms the baseline. LAW excels particularly in complex tasks such as calculating a contract's termination date, surpassing the baseline by 92.9% points. Furthermore, LAW offers a cost-effective alternative to traditional fine-tuned legal LLMs by leveraging reusable, domain-specific tools.

Zhen Zeng, Leijiang Gu, Xun Yang et al.

Knowledge editing aims to efficiently and cost-effectively correct inaccuracies and update outdated information. Recently, there has been growing interest in extending knowledge editing from Large Language Models (LLMs) to Multimodal Large Language Models (MLLMs), which integrate both textual and visual information, introducing additional editing complexities. Existing multimodal knowledge editing works primarily focus on text-oriented, coarse-grained scenarios, failing to address the unique challenges posed by multimodal contexts. In this paper, we propose a visual-oriented, fine-grained multimodal knowledge editing task that targets precise editing in images with multiple interacting entities. We introduce the Fine-Grained Visual Knowledge Editing (FGVEdit) benchmark to evaluate this task. Moreover, we propose a Multimodal Scope Classifier-based Knowledge Editor (MSCKE) framework. MSCKE leverages a multimodal scope classifier that integrates both visual and textual information to accurately identify and update knowledge related to specific entities within images. This approach ensures precise editing while preserving irrelevant information, overcoming the limitations of traditional text-only editing methods. Extensive experiments on the FGVEdit benchmark demonstrate that MSCKE outperforms existing methods, showcasing its effectiveness in solving the complex challenges of multimodal knowledge editing.

Elizabeth Fons, Elena Kochkina, Rachneet Kaur et al.

This paper explores the potential of large language models (LLMs) to generate financial reports from time series data. We propose a framework encompassing prompt engineering, model selection, and evaluation. We introduce an automated highlighting system to categorize information within the generated reports, differentiating between insights derived directly from time series data, stemming from financial reasoning, and those reliant on external knowledge. This approach aids in evaluating the factual grounding and reasoning capabilities of the models. Our experiments, utilizing both data from the real stock market indices and synthetic time series, demonstrate the capability of LLMs to produce coherent and informative financial reports.

Elizabeth Fons, Rachneet Kaur, Zhen Zeng et al.

While image captioning has gained significant attention, the potential of captioning time-series images, prevalent in areas like finance and healthcare, remains largely untapped. Existing time-series captioning methods typically offer generic, domain-agnostic descriptions of time-series shapes and struggle to adapt to new domains without substantial retraining. To address these limitations, we introduce TADACap, a retrieval-based framework to generate domain-aware captions for time-series images, capable of adapting to new domains without retraining. Building on TADACap, we propose a novel retrieval strategy that retrieves diverse image-caption pairs from a target domain database, namely TADACap-diverse. We benchmarked TADACap-diverse against state-of-the-art methods and ablation variants. TADACap-diverse demonstrates comparable semantic accuracy while requiring significantly less annotation effort.

Kausik Lakkaraju, Rachneet Kaur, Parisa Zehtabi et al.

Foundation Models (FMs) have improved time series forecasting in various sectors, such as finance, but their vulnerability to input disturbances can hinder their adoption by stakeholders, such as investors and analysts. To address this, we propose a causally grounded rating framework to study the robustness of Foundational Models for Time Series (FMTS) with respect to input perturbations. We evaluate our approach to the stock price prediction problem, a well-studied problem with easily accessible public data, evaluating six state-of-the-art (some multi-modal) FMTS across six prominent stocks spanning three industries. The ratings proposed by our framework effectively assess the robustness of FMTS and also offer actionable insights for model selection and deployment. Within the scope of our study, we find that (1) multi-modal FMTS exhibit better robustness and accuracy compared to their uni-modal versions and, (2) FMTS pre-trained on time series forecasting task exhibit better robustness and forecasting accuracy compared to general-purpose FMTS pre-trained across diverse settings. Further, to validate our framework's usability, we conduct a user study showcasing FMTS prediction errors along with our computed ratings. The study confirmed that our ratings reduced the difficulty for users in comparing the robustness of different systems.

Zhen Zeng, Leijiang Gu, Zhangling Duan et al.

Multimodal Large Language Models (MLLMs) achieve remarkable capabilities but can inadvertently memorize privacy-sensitive information. Although existing unlearning methods can remove such knowledge, they fail to achieve benign forgetting because they often degrade the model's general image understanding performance. To address this, we propose the Sculpted Memory Forgetting Adapter (SMFA), which confines forgetting to targeted memory regions while preserving overall capabilities. SMFA first fine-tunes the model to replace sensitive responses with refusals, yielding a memory forgetting adapter, and then applies a retaining anchor-guided masking mechanism to prevent interference with unrelated knowledge and understanding ability. To systematically evaluate selective MLLM unlearning, we introduce S-MLLMUn Bench, the first benchmark designed to jointly assess the removal of sensitive knowledge and retention of general visual understanding. Extensive experiments show that, unlike prior methods, SMFA achieves precise and controllable unlearning while maintaining the model's foundational image understanding.

Rachneet Kaur, Nishan Srishankar, Zhen Zeng et al.

Recent multimodal LLMs have shown promise in chart-based visual question answering, but their performance declines sharply on unannotated charts, those requiring precise visual interpretation rather than relying on textual shortcuts. To address this, we introduce ChartAgent, a novel agentic framework that explicitly performs visual reasoning directly within the chart's spatial domain. Unlike textual chain-of-thought reasoning, ChartAgent iteratively decomposes queries into visual subtasks and actively manipulates and interacts with chart images through specialized actions such as drawing annotations, cropping regions (e.g., segmenting pie slices, isolating bars), and localizing axes, using a library of chart-specific vision tools to fulfill each subtask. This iterative reasoning process closely mirrors human cognitive strategies for chart comprehension. ChartAgent achieves state-of-the-art accuracy on the ChartBench and ChartX benchmarks, surpassing prior methods by up to 16.07% absolute gain overall and 17.31% on unannotated, numerically intensive queries. Furthermore, our analyses show that ChartAgent is (a) effective across diverse chart types, (b) achieve the highest scores across varying visual and reasoning complexity levels, and (c) serves as a plug-and-play framework that boosts performance across diverse underlying LLMs. Our work is among the first to demonstrate visually grounded reasoning for chart understanding using tool-augmented multimodal agents.

Tomas de la Rosa, Sriram Gopalakrishnan, Alberto Pozanco et al.

Travel planning is a complex task that involves generating a sequence of actions related to visiting places subject to constraints and maximizing some user satisfaction criteria. Traditional approaches rely on problem formulation in a given formal language, extracting relevant travel information from web sources, and use an adequate problem solver to generate a valid solution. As an alternative, recent Large Language Model (LLM) based approaches directly output plans from user requests using language. Although LLMs possess extensive travel domain knowledge and provide high-level information like points of interest and potential routes, current state-of-the-art models often generate plans that lack coherence, fail to satisfy constraints fully, and do not guarantee the generation of high-quality solutions. We propose TRIP-PAL, a hybrid method that combines the strengths of LLMs and automated planners, where (i) LLMs get and translate travel information and user information into data structures that can be fed into planners; and (ii) automated planners generate travel plans that guarantee constraint satisfaction and optimize for users' utility. Our experiments across various travel scenarios show that TRIP-PAL outperforms an LLM when generating travel plans.

Kausik Lakkaraju, Rachneet Kaur, Zhen Zeng et al.

AI systems are notorious for their fragility; minor input changes can potentially cause major output swings. When such systems are deployed in critical areas like finance, the consequences of their uncertain behavior could be severe. In this paper, we focus on multi-modal time-series forecasting, where imprecision due to noisy or incorrect data can lead to erroneous predictions, impacting stakeholders such as analysts, investors, and traders. Recently, it has been shown that beyond numeric data, graphical transformations can be used with advanced visual models to achieve better performance. In this context, we introduce a rating methodology to assess the robustness of Multi-Modal Time-Series Forecasting Models (MM-TSFM) through causal analysis, which helps us understand and quantify the isolated impact of various attributes on the forecasting accuracy of MM-TSFM. We apply our novel rating method on a variety of numeric and multi-modal forecasting models in a large experimental setup (six input settings of control and perturbations, ten data distributions, time series from six leading stocks in three industries over a year of data, and five time-series forecasters) to draw insights on robust forecasting models and the context of their strengths. Within the scope of our study, our main result is that multi-modal (numeric + visual) forecasting, which was found to be more accurate than numeric forecasting in previous studies, can also be more robust in diverse settings. Our work will help different stakeholders of time-series forecasting understand the models` behaviors along trust (robustness) and accuracy dimensions to select an appropriate model for forecasting using our rating method, leading to improved decision-making.

Alphonsus Adu-Bredu, Nikhil Devraj, Pin-Han Lin et al.

For autonomous service robots to successfully perform long horizon tasks in the real world, they must act intelligently in partially observable environments. Most Task and Motion Planning approaches assume full observability of their state space, making them ineffective in stochastic and partially observable domains that reflect the uncertainties in the real world. We propose an online planning and execution approach for performing long horizon tasks in partially observable domains. Given the robot's belief and a plan skeleton composed of symbolic actions, our approach grounds each symbolic action by inferring continuous action parameters needed to execute the plan successfully. To achieve this, we formulate the problem of joint inference of action parameters as a Hybrid Constraint Satisfaction Problem (H-CSP) and solve the H-CSP using Belief Propagation. The robot executes the resulting parameterized actions, updates its belief of the world and replans when necessary. Our approach is able to efficiently solve partially observable tasks in a realistic kitchen simulation environment. Our approach outperformed an adaptation of the state-of-the-art method across our experiments.

Xiao Li, Yidong Du, Zhen Zeng et al.

We aim for domestic robots to perform long-term indoor service. Under the object-level scene dynamics induced by daily human activities, a robot needs to robustly localize itself in the environment subject to scene uncertainties. Previous works have addressed visual-based localization in static environments, yet the object-level scene dynamics challenge existing methods for the long-term deployment of the robot. This paper proposes a SEmantic understANding Network (SeanNet) architecture that enables an effective learning process with coupled visual and semantic inputs. With a dataset that contains object dynamics, we propose a cascaded contrastive learning scheme to train the SeanNet for learning a vector scene embedding. Subsequently, we can measure the similarity between the current observed scene and the target scene, whereby enables robust localization under object-level dynamics. In our experiments, we benchmark SeanNet against state-of-the-art image-encoding networks (baselines) on scene similarity measures. The SeanNet architecture with the proposed training method can achieve an 85.02\% accuracy which is higher than baselines. We further integrate the SeanNet and the other networks as the localizers into a visual navigation application. We demonstrate that SeanNet achieves higher success rates compared to the baselines.

Naftali Cohen, Srijan Sood, Zhen Zeng et al.

In this work, we address time-series forecasting as a computer vision task. We capture input data as an image and train a model to produce the subsequent image. This approach results in predicting distributions as opposed to pointwise values. To assess the robustness and quality of our approach, we examine various datasets and multiple evaluation metrics. Our experiments show that our forecasting tool is effective for cyclic data but somewhat less for irregular data such as stock prices. Importantly, when using image-based evaluation metrics, we find our method to outperform various baselines, including ARIMA, and a numerical variation of our deep learning approach.

Zhen Zeng, Tucker Balch, Manuela Veloso

Time series forecasting is essential for decision making in many domains. In this work, we address the challenge of predicting prices evolution among multiple potentially interacting financial assets. A solution to this problem has obvious importance for governments, banks, and investors. Statistical methods such as Auto Regressive Integrated Moving Average (ARIMA) are widely applied to these problems. In this paper, we propose to approach economic time series forecasting of multiple financial assets in a novel way via video prediction. Given past prices of multiple potentially interacting financial assets, we aim to predict the prices evolution in the future. Instead of treating the snapshot of prices at each time point as a vector, we spatially layout these prices in 2D as an image, such that we can harness the power of CNNs in learning a latent representation for these financial assets. Thus, the history of these prices becomes a sequence of images, and our goal becomes predicting future images. We build on a state-of-the-art video prediction method for forecasting future images. Our experiments involve the prediction task of the price evolution of nine financial assets traded in U.S. stock markets. The proposed method outperforms baselines including ARIMA, Prophet, and variations of the proposed method, demonstrating the benefits of harnessing the power of CNNs in the problem of economic time series forecasting.

Zhen Zeng, Jianzong Wang, Ning Cheng et al.

Recent neural vocoders usually use a WaveNet-like network to capture the long-term dependencies of the waveform, but a large number of parameters are required to obtain good modeling capabilities. In this paper, an efficient network, named location-variable convolution, is proposed to model the dependencies of waveforms. Different from the use of unified convolution kernels in WaveNet to capture the dependencies of arbitrary waveforms, location-variable convolutions utilizes a kernel predictor to generate multiple sets of convolution kernels based on the mel-spectrum, where each set of convolution kernels is used to perform convolution operations on the associated waveform intervals. Combining WaveGlow and location-variable convolutions, an efficient vocoder, named MelGlow, is designed. Experiments on the LJSpeech dataset show that MelGlow achieves better performance than WaveGlow at small model sizes, which verifies the effectiveness and potential optimization space of location-variable convolutions.

Aolan Sun, Jianzong Wang, Ning Cheng et al.

This paper introduces a graphical representation approach of prosody boundary (GraphPB) in the task of Chinese speech synthesis, intending to parse the semantic and syntactic relationship of input sequences in a graphical domain for improving the prosody performance. The nodes of the graph embedding are formed by prosodic words, and the edges are formed by the other prosodic boundaries, namely prosodic phrase boundary (PPH) and intonation phrase boundary (IPH). Different Graph Neural Networks (GNN) like Gated Graph Neural Network (GGNN) and Graph Long Short-term Memory (G-LSTM) are utilised as graph encoders to exploit the graphical prosody boundary information. Graph-to-sequence model is proposed and formed by a graph encoder and an attentional decoder. Two techniques are proposed to embed sequential information into the graph-to-sequence text-to-speech model. The experimental results show that this proposed approach can encode the phonetic and prosody rhythm of an utterance. The mean opinion score (MOS) of these GNN models shows comparative results with the state-of-the-art sequence-to-sequence models with better performance in the aspect of prosody. This provides an alternative approach for prosody modelling in end-to-end speech synthesis.

Alphonsus Adu-Bredu, Zhen Zeng, Neha Pusalkar et al.

Recent advances in computational perception have significantly improved the ability of autonomous robots to perform state estimation with low entropy. Such advances motivate a reconsideration of robot decision-making under uncertainty. Current approaches to solving sequential decision-making problems model states as inhabiting the extremes of the perceptual entropy spectrum. As such, these methods are either incapable of overcoming perceptual errors or asymptotically inefficient in solving problems with low perceptual entropy. With low entropy perception in mind, we aim to explore a happier medium that balances computational efficiency with the forms of uncertainty we now observe from modern robot perception. We propose an approach for efficient task planning for goal-directed robot reasoning. Our approach combines belief space representation with the fast, goal-directed features of classical planning to efficiently plan for low entropy goal-directed reasoning tasks. We compare our approach with current classical planning and belief space planning approaches by solving low entropy goal-directed grocery packing tasks in simulation. Our approach outperforms these approaches in planning time, execution time, and task success rate in our simulation experiments. We also demonstrate our approach on a real world grocery packing task with physical robot.

Srijan Sood, Zhen Zeng, Naftali Cohen et al.

Time series forecasting is essential for agents to make decisions. Traditional approaches rely on statistical methods to forecast given past numeric values. In practice, end-users often rely on visualizations such as charts and plots to reason about their forecasts. Inspired by practitioners, we re-imagine the topic by creating a novel framework to produce visual forecasts, similar to the way humans intuitively do. In this work, we leverage advances in deep learning to extend the field of time series forecasting to a visual setting. We capture input data as an image and train a model to produce the subsequent image. This approach results in predicting distributions as opposed to pointwise values. We examine various synthetic and real datasets with diverse degrees of complexity. Our experiments show that visual forecasting is effective for cyclic data but somewhat less for irregular data such as stock price. Importantly, when using image-based evaluation metrics, we find the proposed visual forecasting method to outperform various numerical baselines, including ARIMA and a numerical variation of our method. We demonstrate the benefits of incorporating vision-based approaches in forecasting tasks -- both for the quality of the forecasts produced, as well as the metrics that can be used to evaluate them.

Xiaotong Chen, Kaizhi Zheng, Zhen Zeng et al.

In order to enable robust operation in unstructured environments, robots should be able to generalize manipulation actions to novel object instances. For example, to pour and serve a drink, a robot should be able to recognize novel containers which afford the task. Most importantly, robots should be able to manipulate these novel containers to fulfill the task. To achieve this, we aim to provide robust and generalized perception of object affordances and their associated manipulation poses for reliable manipulation. In this work, we combine the notions of affordance and category-level pose, and introduce the Affordance Coordinate Frame (ACF). With ACF, we represent each object class in terms of individual affordance parts and the compatibility between them, where each part is associated with a part category-level pose for robot manipulation. In our experiments, we demonstrate that ACF outperforms state-of-the-art methods for object detection, as well as category-level pose estimation for object parts. We further demonstrate the applicability of ACF to robot manipulation tasks through experiments in a simulated environment.

Zhen Zeng, Jianzong Wang, Ning Cheng et al.

Recent neural speech synthesis systems have gradually focused on the control of prosody to improve the quality of synthesized speech, but they rarely consider the variability of prosody and the correlation between prosody and semantics together. In this paper, a prosody learning mechanism is proposed to model the prosody of speech based on TTS system, where the prosody information of speech is extracted from the melspectrum by a prosody learner and combined with the phoneme sequence to reconstruct the mel-spectrum. Meanwhile, the sematic features of text from the pre-trained language model is introduced to improve the prosody prediction results. In addition, a novel self-attention structure, named as local attention, is proposed to lift this restriction of input text length, where the relative position information of the sequence is modeled by the relative position matrices so that the position encodings is no longer needed. Experiments on English and Mandarin show that speech with more satisfactory prosody has obtained in our model. Especially in Mandarin synthesis, our proposed model outperforms baseline model with a MOS gap of 0.08, and the overall naturalness of the synthesized speech has been significantly improved.

Zhen Zeng, Adrian Röfer, Odest Chadwicke Jenkins

We aim for mobile robots to function in a variety of common human environments. Such robots need to be able to reason about the locations of previously unseen target objects. Landmark objects can help this reasoning by narrowing down the search space significantly. More specifically, we can exploit background knowledge about common spatial relations between landmark and target objects. For example, seeing a table and knowing that cups can often be found on tables aids the discovery of a cup. Such correlations can be expressed as distributions over possible pairing relationships of objects. In this paper, we propose an active visual object search strategy method through our introduction of the Semantic Linking Maps (SLiM) model. SLiM simultaneously maintains the belief over a target object's location as well as landmark objects' locations, while accounting for probabilistic inter-object spatial relations. Based on SLiM, we describe a hybrid search strategy that selects the next best view pose for searching for the target object based on the maintained belief. We demonstrate the efficiency of our SLiM-based search strategy through comparative experiments in simulated environments. We further demonstrate the real-world applicability of SLiM-based search in scenarios with a Fetch mobile manipulation robot.

Zhen Zeng, Jianzong Wang, Ning Cheng et al.

Targeting at both high efficiency and performance, we propose AlignTTS to predict the mel-spectrum in parallel. AlignTTS is based on a Feed-Forward Transformer which generates mel-spectrum from a sequence of characters, and the duration of each character is determined by a duration predictor.Instead of adopting the attention mechanism in Transformer TTS to align text to mel-spectrum, the alignment loss is presented to consider all possible alignments in training by use of dynamic programming. Experiments on the LJSpeech dataset show that our model achieves not only state-of-the-art performance which outperforms Transformer TTS by 0.03 in mean option score (MOS), but also a high efficiency which is more than 50 times faster than real-time.

Aolan Sun, Jianzong Wang, Ning Cheng et al.

This paper leverages the graph-to-sequence method in neural text-to-speech (GraphTTS), which maps the graph embedding of the input sequence to spectrograms. The graphical inputs consist of node and edge representations constructed from input texts. The encoding of these graphical inputs incorporates syntax information by a GNN encoder module. Besides, applying the encoder of GraphTTS as a graph auxiliary encoder (GAE) can analyse prosody information from the semantic structure of texts. This can remove the manual selection of reference audios process and makes prosody modelling an end-to-end procedure. Experimental analysis shows that GraphTTS outperforms the state-of-the-art sequence-to-sequence models by 0.24 in Mean Opinion Score (MOS). GAE can adjust the pause, ventilation and tones of synthesised audios automatically. This experimental conclusion may give some inspiration to researchers working on improving speech synthesis prosody.

Zhen Zeng, Yuefeng Lu, Judong Shen et al.

Precision medicine is becoming a focus in medical research recently, as its implementation brings values to all stakeholders in the healthcare system. Various statistical methodologies have been developed tackling problems in different aspects of this field, e.g., assessing treatment heterogeneity, identifying patient subgroups, or building treatment decision models. However, there is a lack of new tools devoted to selecting and prioritizing predictive biomarkers. We propose a novel tree-based ensemble method, random interaction forest (RIF), to generate predictive importance scores and prioritize candidate biomarkers for constructing refined treatment decision models. RIF was evaluated by comparing with the conventional random forest and univariable regression methods and showed favorable properties under various simulation scenarios. We applied the proposed RIF method to a biomarker dataset from two phase III clinical trials of bezlotoxumab on $\textit{Clostridium difficile}$ infection recurrence and obtained biologically meaningful results.

Zhen Zeng, Yunwen Zhou, Odest Chadwicke Jenkins et al.

We present a filtering-based method for semantic mapping to simultaneously detect objects and localize their 6 degree-of-freedom pose. For our method, called Contextual Temporal Mapping (or CT-Map), we represent the semantic map as a belief over object classes and poses across an observed scene. Inference for the semantic mapping problem is then modeled in the form of a Conditional Random Field (CRF). CT-Map is a CRF that considers two forms of relationship potentials to account for contextual relations between objects and temporal consistency of object poses, as well as a measurement potential on observations. A particle filtering algorithm is then proposed to perform inference in the CT-Map model. We demonstrate the efficacy of the CT-Map method with a Michigan Progress Fetch robot equipped with a RGB-D sensor. Our results demonstrate that the particle filtering based inference of CT-Map provides improved object detection and pose estimation with respect to baseline methods that treat observations as independent samples of a scene.

Zhen Zeng, Zheming Zhou, Zhiqiang Sui et al.

We present the Semantic Robot Programming (SRP) paradigm as a convergence of robot programming by demonstration and semantic mapping. In SRP, a user can directly program a robot manipulator by demonstrating a snapshot of their intended goal scene in workspace. The robot then parses this goal as a scene graph comprised of object poses and inter-object relations, assuming known object geometries. Task and motion planning is then used to realize the user's goal from an arbitrary initial scene configuration. Even when faced with different initial scene configurations, SRP enables the robot to seamlessly adapt to reach the user's demonstrated goal. For scene perception, we propose the Discriminatively-Informed Generative Estimation of Scenes and Transforms (DIGEST) method to infer the initial and goal states of the world from RGBD images. The efficacy of SRP with DIGEST perception is demonstrated for the task of tray-setting with a Michigan Progress Fetch robot. Scene perception and task execution are evaluated with a public household occlusion dataset and our cluttered scene dataset.

Zhiqiang Sui, Zheming Zhou, Zhen Zeng et al.

In order to perform autonomous sequential manipulation tasks, perception in cluttered scenes remains a critical challenge for robots. In this paper, we propose a probabilistic approach for robust sequential scene estimation and manipulation - Sequential Scene Understanding and Manipulation(SUM). SUM considers uncertainty due to discriminative object detection and recognition in the generative estimation of the most likely object poses maintained over time to achieve a robust estimation of the scene under heavy occlusions and unstructured environment. Our method utilizes candidates from discriminative object detector and recognizer to guide the generative process of sampling scene hypothesis, and each scene hypotheses is evaluated against the observations. Also SUM maintains beliefs of scene hypothesis over robot physical actions for better estimation and against noisy detections. We conduct extensive experiments to show that our approach is able to perform robust estimation and manipulation.

Zhen Zeng, Benjamin Kuipers

We aim to enable robot to learn object manipulation by imitation. Given external observations of demonstrations on object manipulations, we believe that two underlying problems to address in learning by imitation is 1) segment a given demonstration into skills that can be individually learned and reused, and 2) formulate the correct RL (Reinforcement Learning) problem that only considers the relevant aspects of each skill so that the policy for each skill can be effectively learned. Previous works made certain progress in this direction, but none has taken private information into account. The public information is the information that is available in the external observations of demonstration, and the private information is the information that are only available to the agent that executes the actions, such as tactile sensations. Our contribution is that we provide a method for the robot to automatically segment the demonstration of object manipulations into multiple skills, and formulate the correct RL problem for each skill, and automatically decide whether the private information is an important aspect of each skill based on interaction with the world. Our experiment shows that our robot learns to pick up a block, and stack it onto another block by imitating an observed demonstration. The evaluation is based on 1) whether the demonstration is reasonably segmented, 2) whether the correct RL problems are formulated, 3) and whether a good policy is learned.