Deepanway Ghosal

Deepanway Ghosal, Yew Ken Chia, Navonil Majumder et al. · deepmind

Recently, the release of INSTRUCTEVAL has provided valuable insights into the performance of large language models (LLMs) that utilize encoder-decoder or decoder-only architecture. Interestingly, despite being introduced four years ago, T5-based LLMs, such as FLAN-T5, continue to outperform the latest decoder-based LLMs, such as LLAMA and VICUNA, on tasks that require general problem-solving skills. This performance discrepancy can be attributed to three key factors: (1) Pre-training data, (2) Backbone architecture, and (3) Instruction dataset. In this technical report, our main focus is on investigating the impact of the third factor by leveraging VICUNA, a large language model based on LLAMA, which has undergone fine-tuning on ChatGPT conversations. To achieve this objective, we fine-tuned VICUNA using a customized instruction dataset collection called FLANMINI. This collection includes a subset of the large-scale instruction dataset known as FLAN, as well as various code-related datasets and conversational datasets derived from ChatGPT/GPT-4. This dataset comprises a large number of tasks that demand problem-solving skills. Our experimental findings strongly indicate that the enhanced problem-solving abilities of our model, FLACUNA, are obtained through fine-tuning VICUNA on the FLAN dataset, leading to significant improvements across numerous benchmark datasets in INSTRUCTEVAL. FLACUNA is publicly available at https://huggingface.co/declare-lab/flacuna-13b-v1.0.

Deepanway Ghosal, Navonil Majumder, Roy Ka-Wei Lee et al. · deepmind

Visual question answering (VQA) is the task of answering questions about an image. The task assumes an understanding of both the image and the question to provide a natural language answer. VQA has gained popularity in recent years due to its potential applications in a wide range of fields, including robotics, education, and healthcare. In this paper, we focus on knowledge-augmented VQA, where answering the question requires commonsense knowledge, world knowledge, and reasoning about ideas and concepts not present in the image. We propose a multimodal framework that uses language guidance (LG) in the form of rationales, image captions, scene graphs, etc to answer questions more accurately. We benchmark our method on the multi-choice question-answering task of the A-OKVQA, Science-QA, VSR, and IconQA datasets using CLIP and BLIP models. We show that the use of language guidance is a simple but powerful and effective strategy for visual question answering. Our language guidance improves the performance of CLIP by 7.6% and BLIP-2 by 4.8% in the challenging A-OKVQA dataset. We also observe consistent improvement in performance on the Science-QA, VSR, and IconQA datasets when using the proposed language guidances. The implementation of LG-VQA is publicly available at https:// github.com/declare-lab/LG-VQA.

Deepanway Ghosal, Navonil Majumder, Rada Mihalcea et al. · deepmind

We propose a simple refactoring of multi-choice question answering (MCQA) tasks as a series of binary classifications. The MCQA task is generally performed by scoring each (question, answer) pair normalized over all the pairs, and then selecting the answer from the pair that yield the highest score. For n answer choices, this is equivalent to an n-class classification setup where only one class (true answer) is correct. We instead show that classifying (question, true answer) as positive instances and (question, false answer) as negative instances is significantly more effective across various models and datasets. We show the efficacy of our proposed approach in different tasks -- abductive reasoning, commonsense question answering, science question answering, and sentence completion. Our DeBERTa binary classification model reaches the top or close to the top performance on public leaderboards for these tasks. The source code of the proposed approach is available at https://github.com/declare-lab/TEAM.

Deepanway Ghosal, Navonil Majumder, Ambuj Mehrish et al. · deepmind

The immense scale of the recent large language models (LLM) allows many interesting properties, such as, instruction- and chain-of-thought-based fine-tuning, that has significantly improved zero- and few-shot performance in many natural language processing (NLP) tasks. Inspired by such successes, we adopt such an instruction-tuned LLM Flan-T5 as the text encoder for text-to-audio (TTA) generation -- a task where the goal is to generate an audio from its textual description. The prior works on TTA either pre-trained a joint text-audio encoder or used a non-instruction-tuned model, such as, T5. Consequently, our latent diffusion model (LDM)-based approach TANGO outperforms the state-of-the-art AudioLDM on most metrics and stays comparable on the rest on AudioCaps test set, despite training the LDM on a 63 times smaller dataset and keeping the text encoder frozen. This improvement might also be attributed to the adoption of audio pressure level-based sound mixing for training set augmentation, whereas the prior methods take a random mix.

Deepanway Ghosal, Siqi Shen, Navonil Majumder et al. · deepmind

This paper addresses the problem of dialogue reasoning with contextualized commonsense inference. We curate CICERO, a dataset of dyadic conversations with five types of utterance-level reasoning-based inferences: cause, subsequent event, prerequisite, motivation, and emotional reaction. The dataset contains 53,105 of such inferences from 5,672 dialogues. We use this dataset to solve relevant generative and discriminative tasks: generation of cause and subsequent event; generation of prerequisite, motivation, and listener's emotional reaction; and selection of plausible alternatives. Our results ascertain the value of such dialogue-centric commonsense knowledge datasets. It is our hope that CICERO will open new research avenues into commonsense-based dialogue reasoning.

Siqi Shen, Deepanway Ghosal, Navonil Majumder et al. · deepmind

Contextual commonsense inference is the task of generating various types of explanations around the events in a dyadic dialogue, including cause, motivation, emotional reaction, and others. Producing a coherent and non-trivial explanation requires awareness of the dialogue's structure and of how an event is grounded in the context. In this work, we create CICEROv2, a dataset consisting of 8,351 instances from 2,379 dialogues, containing multiple human-written answers for each contextual commonsense inference question, representing a type of explanation on cause, subsequent event, motivation, and emotional reaction. We show that the inferences in CICEROv2 are more semantically diverse than other contextual commonsense inference datasets. To solve the inference task, we propose a collection of pre-training objectives, including concept denoising and utterance sorting to prepare a pre-trained model for the downstream contextual commonsense inference task. Our results show that the proposed pre-training objectives are effective at adapting the pre-trained T5-Large model for the contextual commonsense inference task.

Deepanway Ghosal, Somak Aditya, Monojit Choudhury · deepmind

The Natural Language Inference (NLI) task often requires reasoning over multiple steps to reach the conclusion. While the necessity of generating such intermediate steps (instead of a summary explanation) has gained popular support, it is unclear how to generate such steps without complete end-to-end supervision and how such generated steps can be further utilized. In this work, we train a sequence-to-sequence model to generate only the next step given an NLI premise and hypothesis pair (and previous steps); then enhance it with external knowledge and symbolic search to generate intermediate steps with only next-step supervision. We show the correctness of such generated steps through automated and human verification. Furthermore, we show that such generated steps can help improve end-to-end NLI task performance using simple data augmentation strategies, across multiple public NLI datasets.

Aileen Cheng, Alon Jacovi, Amir Globerson et al.

We introduce The FACTS Leaderboard, an online leaderboard suite and associated set of benchmarks that comprehensively evaluates the ability of language models to generate factually accurate text across diverse scenarios. The suite provides a holistic measure of factuality by aggregating the performance of models on four distinct sub-leaderboards: (1) FACTS Multimodal, which measures the factuality of responses to image-based questions; (2) FACTS Parametric, which assesses models' world knowledge by answering closed-book factoid questions from internal parameters; (3) FACTS Search, which evaluates factuality in information-seeking scenarios, where the model must use a search API; and (4) FACTS Grounding (v2), which evaluates whether long-form responses are grounded in provided documents, featuring significantly improved judge models. Each sub-leaderboard employs automated judge models to score model responses, and the final suite score is an average of the four components, designed to provide a robust and balanced assessment of a model's overall factuality. The FACTS Leaderboard Suite will be actively maintained, containing both public and private splits to allow for external participation while guarding its integrity. It can be found at https://www.kaggle.com/benchmarks/google/facts .

Pengfei Hong, Navonil Majumder, Deepanway Ghosal et al. · deepmind

Recent advancements in Large Language Models (LLMs) have showcased striking results on existing logical reasoning benchmarks, with some models even surpassing human performance. However, the true depth of their competencies and robustness in reasoning tasks remains an open question. To this end, in this paper, we focus on two popular reasoning tasks: arithmetic reasoning and code generation. Particularly, we introduce (i) a general ontology of perturbations for math and coding questions, (ii) a semi-automatic method to apply these perturbations, and (iii) two datasets, GSMORE and HUMANEVAL-CORE, respectively, of perturbed math and coding problems to probe LLM capabilities in numeric reasoning and coding tasks. Through comprehensive evaluations of both closed-source and open-source LLMs, we show a significant performance drop across all the models against the perturbed questions, suggesting that the current LLMs lack robust problem solving skills and structured reasoning abilities in many areas, as defined by our ontology. We open-source the datasets and source codes at: https://github.com/declare-lab/LLM-ReasoningTest.

Vernon Y. H. Toh, Deepanway Ghosal, Soujanya Poria · deepmind

Large language models (LLMs) have shown increasing competence in solving mathematical reasoning problems. However, many open-source LLMs still struggle with errors in calculation and semantic understanding during intermediate reasoning steps. In this work, we introduce Prove, a simple yet effective framework that leverages translated programs derived from natural language solutions as a verification mechanism to filter out potentially incorrect reasoning paths before aggregating final answers. Unlike vanilla majority voting, our approach filters out solutions whose corresponding program output is inconsistent with the generated solution, aggregating only those that pass verification. We conducted extensive experiments using 13 open-source LLMs from various model families and sizes, ranging from 0.5B to 13B parameters, across eight mathematical benchmarks. Our results show that Prove consistently outperforms vanilla majority voting as a heuristic for solving mathematical reasoning tasks across all model sizes and datasets, achieving improvements of up to 18% on GSM8K and 8% on MATH-500. Our codes are available at https://github.com/declare-lab/prove.

Vernon Y. H. Toh, Yew Ken Chia, Deepanway Ghosal et al. · deepmind

The releases of OpenAI's o-[n] series, such as o1, o3, and o4-mini, mark a significant paradigm shift in Large Language Models towards advanced reasoning capabilities. Notably, models like o3 have demonstrated strong performance on benchmarks like the Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI). However, this benchmark is limited to symbolic patterns, whereas humans often perceive and reason about multimodal scenarios involving both vision and language data. Thus, there is an urgent need to investigate advanced reasoning capabilities in multimodal tasks. To this end, we track the evolution of the GPT-[n] and o-[n] series models (including o1, o3, and o4-mini) on challenging multimodal puzzles from PuzzleVQA and AlgoPuzzleVQA, which demand fine-grained visual perception. Our results reveal that o-[n] series, particularly later iterations like o3 and o4-mini, significantly outperform the GPT-[n] series and show strong scalability in multimodal reasoning. Nonetheless, despite these substantial advancements and the superior capabilities demonstrated by the o-[n] series, our findings highlight that even these leading models face persistent challenges. Difficulties are particularly evident in tasks requiring precise visual perception, robust compositional reasoning across multiple visual attributes, and solving complex algorithmic or highly combinatorial puzzles, indicating critical areas for future AGI development. We plan to continuously track new models in the series and update our results in this paper accordingly. All resources used in this evaluation are openly available at https://github.com/declare-lab/LLM-PuzzleTest.

Koushik Viswanadha, Deepanway Ghosal, Somak Aditya · deepmind

Logical reasoning is a key task for artificial intelligence due to it's role in major downstream tasks such as Question Answering, Summarization. Recent methods in improving the reasoning ability of LLMs fall short in correctly converting a natural language reasoning problem to an equivalent logical formulation, which hinders the framework's overall ability to reason. Towards this, we propose to use finetuning on a preference optimization dataset to learn to parse and represent a natural language problem as a whole to a consistent logical program by 1) introducing a new supervised and preference optimization dataset LogicPO, and 2) adopting popular techniques such as Direct Preference Optimization (DPO), Kahneman-Tversky optimization (KTO) to finetune open-source LLMs. Our best model with Phi-3.5 consistently outperforms GPT-3.5-turbo's (8-shot) by producing 10% more logically correct and with 14% less syntax errors. Through the framework and our improved evaluation metrics, we offer a promising direction in improving the logical reasoning of LLMs by better representing them in their logical formulations.

Deepanway Ghosal, Preksha Nema, Aravindan Raghuveer

The task of table summarization involves generating text that both succinctly and accurately represents the table or a specific set of highlighted cells within a table. While significant progress has been made in table to text generation techniques, models still mostly generate descriptive summaries, which reiterates the information contained within the table in sentences. Through analysis of popular table to text benchmarks (ToTTo (Parikh et al., 2020 and InfoTabs (Gupta et al., 2020) we observe that in order to generate the ideal summary, multiple types of reasoning is needed coupled with access to knowledge beyond the scope of the table. To address this gap, we propose ReTAG, a table and reasoning aware model that uses vector-quantization to infuse different types of analytical reasoning into the output. ReTAG achieves 2.2%, 2.9% improvement on the PARENT metric in the relevant slice of ToTTo and InfoTabs for the table to text generation task over state of the art baselines. Through human evaluation, we observe that output from ReTAG is upto 12% more faithful and analytical compared to a strong table-aware model. To the best of our knowledge, ReTAG is the first model that can controllably use multiple reasoning methods within a structure-aware sequence to sequence model to surpass state of the art performance in multiple table to text tasks. We extend (and open source 35.6K analytical, 55.9k descriptive instances) the ToTTo, InfoTabs datasets with the reasoning categories used in each reference sentences.

Deepanway Ghosal, Navonil Majumder, Rada Mihalcea et al.

Sentence order prediction is the task of finding the correct order of sentences in a randomly ordered document. Correctly ordering the sentences requires an understanding of coherence with respect to the chronological sequence of events described in the text. Document-level contextual understanding and commonsense knowledge centered around these events are often essential in uncovering this coherence and predicting the exact chronological order. In this paper, we introduce STaCK -- a framework based on graph neural networks and temporal commonsense knowledge to model global information and predict the relative order of sentences. Our graph network accumulates temporal evidence using knowledge of `past' and `future' and formulates sentence ordering as a constrained edge classification problem. We report results on five different datasets, and empirically show that the proposed method is naturally suitable for order prediction. The implementation of this work is publicly available at: https://github.com/declare-lab/sentence-ordering.

Navonil Majumder, Deepanway Ghosal, Devamanyu Hazarika et al.

The majority of existing methods for empathetic response generation rely on the emotion of the context to generate empathetic responses. However, empathy is much more than generating responses with an appropriate emotion. It also often entails subtle expressions of understanding and personal resonance with the situation of the other interlocutor. Unfortunately, such qualities are difficult to quantify and the datasets lack the relevant annotations. To address this issue, in this paper we propose an approach that relies on exemplars to cue the generative model on fine stylistic properties that signal empathy to the interlocutor. To this end, we employ dense passage retrieval to extract relevant exemplary responses from the training set. Three elements of human communication -- emotional presence, interpretation, and exploration, and sentiment are additionally introduced using synthetic labels to guide the generation towards empathy. The human evaluation is also extended by these elements of human communication. We empirically show that these approaches yield significant improvements in empathetic response quality in terms of both automated and human-evaluated metrics. The implementation is available at https://github.com/declare-lab/exemplary-empathy.

Soujanya Poria, Navonil Majumder, Devamanyu Hazarika et al.

We address the problem of recognizing emotion cause in conversations, define two novel sub-tasks of this problem, and provide a corresponding dialogue-level dataset, along with strong Transformer-based baselines. The dataset is available at https://github.com/declare-lab/RECCON. Introduction: Recognizing the cause behind emotions in text is a fundamental yet under-explored area of research in NLP. Advances in this area hold the potential to improve interpretability and performance in affect-based models. Identifying emotion causes at the utterance level in conversations is particularly challenging due to the intermingling dynamics among the interlocutors. Method: We introduce the task of Recognizing Emotion Cause in CONversations with an accompanying dataset named RECCON, containing over 1,000 dialogues and 10,000 utterance cause-effect pairs. Furthermore, we define different cause types based on the source of the causes, and establish strong Transformer-based baselines to address two different sub-tasks on this dataset: causal span extraction and causal emotion entailment. Result: Our Transformer-based baselines, which leverage contextual pre-trained embeddings, such as RoBERTa, outperform the state-of-the-art emotion cause extraction approaches Conclusion: We introduce a new task highly relevant for (explainable) emotion-aware artificial intelligence: recognizing emotion cause in conversations, provide a new highly challenging publicly available dialogue-level dataset for this task, and give strong baseline results on this dataset.

Deepanway Ghosal, Navonil Majumder, Alexander Gelbukh et al.

In this paper, we address the task of utterance level emotion recognition in conversations using commonsense knowledge. We propose COSMIC, a new framework that incorporates different elements of commonsense such as mental states, events, and causal relations, and build upon them to learn interactions between interlocutors participating in a conversation. Current state-of-the-art methods often encounter difficulties in context propagation, emotion shift detection, and differentiating between related emotion classes. By learning distinct commonsense representations, COSMIC addresses these challenges and achieves new state-of-the-art results for emotion recognition on four different benchmark conversational datasets. Our code is available at https://github.com/declare-lab/conv-emotion.

Navonil Majumder, Pengfei Hong, Shanshan Peng et al.

Current approaches to empathetic response generation view the set of emotions expressed in the input text as a flat structure, where all the emotions are treated uniformly. We argue that empathetic responses often mimic the emotion of the user to a varying degree, depending on its positivity or negativity and content. We show that the consideration of this polarity-based emotion clusters and emotional mimicry results in improved empathy and contextual relevance of the response as compared to the state-of-the-art. Also, we introduce stochasticity into the emotion mixture that yields emotionally more varied empathetic responses than the previous work. We demonstrate the importance of these factors to empathetic response generation using both automatic- and human-based evaluations. The implementation of MIME is publicly available at https://github.com/declare-lab/MIME.

Deepanway Ghosal, Navonil Majumder, Rada Mihalcea et al.

The recent abundance of conversational data on the Web and elsewhere calls for effective NLP systems for dialog understanding. Complete utterance-level understanding often requires context understanding, defined by nearby utterances. In recent years, a number of approaches have been proposed for various utterance-level dialogue understanding tasks. Most of these approaches account for the context for effective understanding. In this paper, we explore and quantify the role of context for different aspects of a dialogue, namely emotion, intent, and dialogue act identification, using state-of-the-art dialog understanding methods as baselines. Specifically, we employ various perturbations to distort the context of a given utterance and study its impact on the different tasks and baselines. This provides us with insights into the fundamental contextual controlling factors of different aspects of a dialogue. Such insights can inspire more effective dialogue understanding models, and provide support for future text generation approaches. The implementation pertaining to this work is available at https://github.com/declare-lab/dialogue-understanding.

Navonil Majumder, Chia-Yu Hung, Deepanway Ghosal et al. · deepmind

Generative multimodal content is increasingly prevalent in much of the content creation arena, as it has the potential to allow artists and media personnel to create pre-production mockups by quickly bringing their ideas to life. The generation of audio from text prompts is an important aspect of such processes in the music and film industry. Many of the recent diffusion-based text-to-audio models focus on training increasingly sophisticated diffusion models on a large set of datasets of prompt-audio pairs. These models do not explicitly focus on the presence of concepts or events and their temporal ordering in the output audio with respect to the input prompt. Our hypothesis is focusing on how these aspects of audio generation could improve audio generation performance in the presence of limited data. As such, in this work, using an existing text-to-audio model Tango, we synthetically create a preference dataset where each prompt has a winner audio output and some loser audio outputs for the diffusion model to learn from. The loser outputs, in theory, have some concepts from the prompt missing or in an incorrect order. We fine-tune the publicly available Tango text-to-audio model using diffusion-DPO (direct preference optimization) loss on our preference dataset and show that it leads to improved audio output over Tango and AudioLDM2, in terms of both automatic- and manual-evaluation metrics.

Yew Ken Chia, Vernon Toh Yan Han, Deepanway Ghosal et al. · deepmind

Large multimodal models extend the impressive capabilities of large language models by integrating multimodal understanding abilities. However, it is not clear how they can emulate the general intelligence and reasoning ability of humans. As recognizing patterns and abstracting concepts are key to general intelligence, we introduce PuzzleVQA, a collection of 2000 puzzle instances based on abstract patterns. With this dataset, we evaluate large multimodal models with abstract patterns based on fundamental concepts, including colors, numbers, sizes, and shapes. Through our experiments on state-of-the-art large multimodal models, we find that they are not able to generalize well to simple abstract patterns. Notably, GPT-4V achieves a score of 46.4% on single-concept puzzles, which shows that state-of-the-art models struggle on our dataset. To diagnose the reasoning challenges in large multimodal models, we progressively guide the models with our ground truth reasoning explanations for visual perception, inductive reasoning, and deductive reasoning. Our systematic analysis finds that the main bottlenecks of GPT-4V are weaker visual perception and inductive reasoning abilities. Through this work, we hope to shed light on the limitations of large multimodal models and how they can better emulate human cognitive processes in the future. Our data and code are available at https://puzzlevqa.github.io

Deepanway Ghosal, Vernon Toh Yan Han, Chia Yew Ken et al. · deepmind

This paper introduces the novel task of multimodal puzzle solving, framed within the context of visual question-answering. We present a new dataset, AlgoPuzzleVQA designed to challenge and evaluate the capabilities of multimodal language models in solving algorithmic puzzles that necessitate both visual understanding, language understanding, and complex algorithmic reasoning. We create the puzzles to encompass a diverse array of mathematical and algorithmic topics such as boolean logic, combinatorics, graph theory, optimization, search, etc., aiming to evaluate the gap between visual data interpretation and algorithmic problem-solving skills. The dataset is generated automatically from code authored by humans. All our puzzles have exact solutions that can be found from the algorithm without tedious human calculations. It ensures that our dataset can be scaled up arbitrarily in terms of reasoning complexity and dataset size. Our investigation reveals that large language models (LLMs) such as GPT4V and Gemini exhibit limited performance in puzzle-solving tasks. We find that their performance is near random in a multi-choice question-answering setup for a significant number of puzzles. The findings emphasize the challenges of integrating visual, language, and algorithmic knowledge for solving complex reasoning problems.

Qi Sun, Pengfei Hong, Tej Deep Pala et al. · deepmind

Traditional reinforcement learning-based robotic control methods are often task-specific and fail to generalize across diverse environments or unseen objects and instructions. Visual Language Models (VLMs) demonstrate strong scene understanding and planning capabilities but lack the ability to generate actionable policies tailored to specific robotic embodiments. To address this, Visual-Language-Action (VLA) models have emerged, yet they face challenges in long-horizon spatial reasoning and grounded task planning. In this work, we propose the Embodied Multimodal Action Model with Grounded Chain of Thought and Look-ahead Spatial Reasoning, Emma-X. Emma-X leverages our constructed hierarchical embodiment dataset based on BridgeV2, containing 60,000 robot manipulation trajectories auto-annotated with grounded task reasoning and spatial guidance. Additionally, we introduce a trajectory segmentation strategy based on gripper states and motion trajectories, which can help mitigate hallucination in grounding subtask reasoning generation. Experimental results demonstrate that Emma-X achieves superior performance over competitive baselines, particularly in real-world robotic tasks requiring spatial reasoning.

Jinjie Ni, Yifan Song, Deepanway Ghosal et al. · deepmind

Perceiving and generating diverse modalities are crucial for AI models to effectively learn from and engage with real-world signals, necessitating reliable evaluations for their development. We identify two major issues in current evaluations: (1) inconsistent standards, shaped by different communities with varying protocols and maturity levels; and (2) significant query, grading, and generalization biases. To address these, we introduce MixEval-X, the first any-to-any, real-world benchmark designed to optimize and standardize evaluations across diverse input and output modalities. We propose multi-modal benchmark mixture and adaptation-rectification pipelines to reconstruct real-world task distributions, ensuring evaluations generalize effectively to real-world use cases. Extensive meta-evaluations show our approach effectively aligns benchmark samples with real-world task distributions. Meanwhile, MixEval-X's model rankings correlate strongly with that of crowd-sourced real-world evaluations (up to 0.98) while being much more efficient. We provide comprehensive leaderboards to rerank existing models and organizations and offer insights to enhance understanding of multi-modal evaluations and inform future research.

Aritra Dutta, Swapnanil Mukherjee, Deepanway Ghosal et al. · deepmind

Commonsense visual-question answering often hinges on knowledge that is missing from the image or the question. Small vision-language models (sVLMs) such as ViLT, VisualBERT and FLAVA therefore lag behind their larger generative counterparts. To study the effect of careful commonsense knowledge integration on sVLMs, we present an end-to-end framework (NLKI) that (i) retrieves natural language facts, (ii) prompts an LLM to craft natural language explanations, and (iii) feeds both signals to sVLMs respectively across two commonsense VQA datasets (CRIC, AOKVQA) and a visual-entailment dataset (e-SNLI-VE). Facts retrieved using a fine-tuned ColBERTv2 and an object information-enriched prompt yield explanations that largely cut down hallucinations, while lifting the end-to-end answer accuracy by up to 7% (across 3 datasets), making FLAVA and other models in NLKI match or exceed medium-sized VLMs such as Qwen-2 VL-2B and SmolVLM-2.5B. As these benchmarks contain 10-25% label noise, additional finetuning using noise-robust losses (such as symmetric cross entropy and generalised cross entropy) adds another 2.5% in CRIC, and 5.5% in AOKVQA. Our findings expose when LLM-based commonsense knowledge beats retrieval from commonsense knowledge bases, how noise-aware training stabilises small models in the context of external knowledge augmentation, and why parameter-efficient commonsense reasoning is now within reach for 250M models.

Zhifeng Kong, Sang-gil Lee, Deepanway Ghosal et al.

It is an open challenge to obtain high quality training data, especially captions, for text-to-audio models. Although prior methods have leveraged \textit{text-only language models} to augment and improve captions, such methods have limitations related to scale and coherence between audio and captions. In this work, we propose an audio captioning pipeline that uses an \textit{audio language model} to synthesize accurate and diverse captions for audio at scale. We leverage this pipeline to produce a dataset of synthetic captions for AudioSet, named \texttt{AF-AudioSet}, and then evaluate the benefit of pre-training text-to-audio models on these synthetic captions. Through systematic evaluations on AudioCaps and MusicCaps, we find leveraging our pipeline and synthetic captions leads to significant improvements on audio generation quality, achieving a new \textit{state-of-the-art}.

Deepanway Ghosal, Pengfei Hong, Siqi Shen et al.

Commonsense inference to understand and explain human language is a fundamental research problem in natural language processing. Explaining human conversations poses a great challenge as it requires contextual understanding, planning, inference, and several aspects of reasoning including causal, temporal, and commonsense reasoning. In this work, we introduce CIDER -- a manually curated dataset that contains dyadic dialogue explanations in the form of implicit and explicit knowledge triplets inferred using contextual commonsense inference. Extracting such rich explanations from conversations can be conducive to improving several downstream applications. The annotated triplets are categorized by the type of commonsense knowledge present (e.g., causal, conditional, temporal). We set up three different tasks conditioned on the annotated dataset: Dialogue-level Natural Language Inference, Span Extraction, and Multi-choice Span Selection. Baseline results obtained with transformer-based models reveal that the tasks are difficult, paving the way for promising future research. The dataset and the baseline implementations are publicly available at https://cider-task.github.io/cider/.

Abhinaba Roy, Deepanway Ghosal, Erik Cambria et al.

Zero shot learning -- the problem of training and testing on a completely disjoint set of classes -- relies greatly on its ability to transfer knowledge from train classes to test classes. Traditionally semantic embeddings consisting of human defined attributes (HA) or distributed word embeddings (DWE) are used to facilitate this transfer by improving the association between visual and semantic embeddings. In this paper, we take advantage of explicit relations between nodes defined in ConceptNet, a commonsense knowledge graph, to generate commonsense embeddings of the class labels by using a graph convolution network-based autoencoder. Our experiments performed on three standard benchmark datasets surpass the strong baselines when we fuse our commonsense embeddings with existing semantic embeddings i.e. HA and DWE.

Hui Chen, Deepanway Ghosal, Navonil Majumder et al.

Persuasion aims at forming one's opinion and action via a series of persuasive messages containing persuader's strategies. Due to its potential application in persuasive dialogue systems, the task of persuasive strategy recognition has gained much attention lately. Previous methods on user intent recognition in dialogue systems adopt recurrent neural network (RNN) or convolutional neural network (CNN) to model context in conversational history, neglecting the tactic history and intra-speaker relation. In this paper, we demonstrate the limitations of a Transformer-based approach coupled with Conditional Random Field (CRF) for the task of persuasive strategy recognition. In this model, we leverage inter- and intra-speaker contextual semantic features, as well as label dependencies to improve the recognition. Despite extensive hyper-parameter optimizations, this architecture fails to outperform the baseline methods. We observe two negative results. Firstly, CRF cannot capture persuasive label dependencies, possibly as strategies in persuasive dialogues do not follow any strict grammar or rules as the cases in Named Entity Recognition (NER) or part-of-speech (POS) tagging. Secondly, the Transformer encoder trained from scratch is less capable of capturing sequential information in persuasive dialogues than Long Short-Term Memory (LSTM). We attribute this to the reason that the vanilla Transformer encoder does not efficiently consider relative position information of sequence elements.

Deepanway Ghosal, Maheshkumar H. Kolekar

Visual interest & affect prediction is a very interesting area of research in the area of computer vision. In this paper, we propose a transfer learning and attention mechanism based neural network model to predict visual interest & affective dimensions in digital photos. Learning the multi-dimensional affects is addressed through a multi-task learning framework. With various experiments we show the effectiveness of the proposed approach. Evaluation of our model on the benchmark dataset shows large improvement over current state-of-the-art systems.

Deepanway Ghosal, Devamanyu Hazarika, Abhinaba Roy et al.

Cross-domain sentiment analysis has received significant attention in recent years, prompted by the need to combat the domain gap between different applications that make use of sentiment analysis. In this paper, we take a novel perspective on this task by exploring the role of external commonsense knowledge. We introduce a new framework, KinGDOM, which utilizes the ConceptNet knowledge graph to enrich the semantics of a document by providing both domain-specific and domain-general background concepts. These concepts are learned by training a graph convolutional autoencoder that leverages inter-domain concepts in a domain-invariant manner. Conditioning a popular domain-adversarial baseline method with these learned concepts helps improve its performance over state-of-the-art approaches, demonstrating the efficacy of our proposed framework.

Deepanway Ghosal, Navonil Majumder, Soujanya Poria et al.

Emotion recognition in conversation (ERC) has received much attention, lately, from researchers due to its potential widespread applications in diverse areas, such as health-care, education, and human resources. In this paper, we present Dialogue Graph Convolutional Network (DialogueGCN), a graph neural network based approach to ERC. We leverage self and inter-speaker dependency of the interlocutors to model conversational context for emotion recognition. Through the graph network, DialogueGCN addresses context propagation issues present in the current RNN-based methods. We empirically show that this method alleviates such issues, while outperforming the current state of the art on a number of benchmark emotion classification datasets.

Md Shad Akhtar, Dushyant Singh Chauhan, Deepanway Ghosal et al.

Related tasks often have inter-dependence on each other and perform better when solved in a joint framework. In this paper, we present a deep multi-task learning framework that jointly performs sentiment and emotion analysis both. The multi-modal inputs (i.e., text, acoustic and visual frames) of a video convey diverse and distinctive information, and usually do not have equal contribution in the decision making. We propose a context-level inter-modal attention framework for simultaneously predicting the sentiment and expressed emotions of an utterance. We evaluate our proposed approach on CMU-MOSEI dataset for multi-modal sentiment and emotion analysis. Evaluation results suggest that multi-task learning framework offers improvement over the single-task framework. The proposed approach reports new state-of-the-art performance for both sentiment analysis and emotion analysis.

Md Shad Akhtar, Deepanway Ghosal, Asif Ekbal et al.

In this paper, through multi-task ensemble framework we address three problems of emotion and sentiment analysis i.e. "emotion classification & intensity", "valence, arousal & dominance for emotion" and "valence & arousal} for sentiment". The underlying problems cover two granularities (i.e. coarse-grained and fine-grained) and a diverse range of domains (i.e. tweets, Facebook posts, news headlines, blogs, letters etc.). The ensemble model aims to leverage the learned representations of three deep learning models (i.e. CNN, LSTM and GRU) and a hand-crafted feature representation for the predictions. Experimental results on the benchmark datasets show the efficacy of our proposed multi-task ensemble frameworks. We obtain the performance improvement of 2-3 points on an average over single-task systems for most of the problems and domains.