Search

Effective decision-making in partially observable environments demands robust memory management. Despite their success in supervised learning, current deep-learning memory models struggle in reinforcement learning environments that are partially observable and long-term. They fail to efficiently capture relevant past information, adapt flexibly to changing observations, and maintain stable updates over long episodes. We theoretically analyze the limitations of existing memory models within a unified framework and introduce the Stable Hadamard Memory, a novel memory model for reinforcement learning agents. Our model dynamically adjusts memory by erasing no longer needed experiences and reinforcing crucial ones computationally efficiently. To this end, we leverage the Hadamard product for calibrating and updating memory, specifically designed to enhance memory capacity while mitigating numerical and learning challenges. Our approach significantly outperforms state-of-the-art memory-based methods on challenging partially observable benchmarks, such as meta-reinforcement learning, long-horizon credit assignment, and POPGym, demonstrating superior performance in handling long-term and evolving contexts., Comment: Preprint 18 pages

Recently, reinforcement learning (RL) has proved a promising alternative for conventional local heuristics in score-based approaches to learning directed acyclic causal graphs (DAGs) from observational data. However, the intricate acyclicity constraint still challenges the efficient exploration of the vast space of DAGs in existing methods. In this study, we introduce ALIAS (reinforced dAg Learning wIthout Acyclicity conStraints), a novel approach to causal discovery powered by the RL machinery. Our method features an efficient policy for generating DAGs in just a single step with an optimal quadratic complexity, fueled by a novel parametrization of DAGs that directly translates a continuous space to the space of all DAGs, bypassing the need for explicitly enforcing acyclicity constraints. This approach enables us to navigate the search space more effectively by utilizing policy gradient methods and established scoring functions. In addition, we provide compelling empirical evidence for the strong performance of ALIAS in comparison with state-of-the-arts in causal discovery over increasingly difficult experiment conditions on both synthetic and real datasets.

Modern reinforcement learning (RL) struggles to capture real-world cause-and-effect dynamics, leading to inefficient exploration due to extensive trial-and-error actions. While recent efforts to improve agent exploration have leveraged causal discovery, they often make unrealistic assumptions of causal variables in the environments. In this paper, we introduce a novel framework, Variable-Agnostic Causal Exploration for Reinforcement Learning (VACERL), incorporating causal relationships to drive exploration in RL without specifying environmental causal variables. Our approach automatically identifies crucial observation-action steps associated with key variables using attention mechanisms. Subsequently, it constructs the causal graph connecting these steps, which guides the agent towards observation-action pairs with greater causal influence on task completion. This can be leveraged to generate intrinsic rewards or establish a hierarchy of subgoals to enhance exploration efficiency. Experimental results showcase a significant improvement in agent performance in grid-world, 2d games and robotic domains, particularly in scenarios with sparse rewards and noisy actions, such as the notorious Noisy-TV environments.

How can Large Language Models (LLMs) be aligned with human intentions and values? A typical solution is to gather human preference on model outputs and finetune the LLMs accordingly while ensuring that updates do not deviate too far from a reference model. Recent approaches, such as direct preference optimization (DPO), have eliminated the need for unstable and sluggish reinforcement learning optimization by introducing close-formed supervised losses. However, a significant limitation of the current approach is its design for a single reference model only, neglecting to leverage the collective power of numerous pretrained LLMs. To overcome this limitation, we introduce a novel closed-form formulation for direct preference optimization using multiple reference models. The resulting algorithm, Multi-Reference Preference Optimization (MRPO), leverages broader prior knowledge from diverse reference models, substantially enhancing preference learning capabilities compared to the single-reference DPO. Our experiments demonstrate that LLMs finetuned with MRPO generalize better in various preference data, regardless of data scarcity or abundance. Furthermore, MRPO effectively finetunes LLMs to exhibit superior performance in several downstream natural language processing tasks such as GSM8K and TruthfulQA., Comment: 20 pages

We propose Pointer-Augmented Neural Memory (PANM) to help neural networks understand and apply symbol processing to new, longer sequences of data. PANM integrates an external neural memory that uses novel physical addresses and pointer manipulation techniques to mimic human and computer symbol processing abilities. PANM facilitates pointer assignment, dereference, and arithmetic by explicitly using physical pointers to access memory content. Remarkably, it can learn to perform these operations through end-to-end training on sequence data, powering various sequential models. Our experiments demonstrate PANM's exceptional length extrapolating capabilities and improved performance in tasks that require symbol processing, such as algorithmic reasoning and Dyck language recognition. PANM helps Transformer achieve up to 100% generalization accuracy in compositional learning tasks and significantly better results in mathematical reasoning, question answering and machine translation tasks., Comment: Preprint

Large Language Models (LLMs) can perform various natural language processing tasks with suitable instruction prompts. However, designing effective prompts manually is challenging and time-consuming. Existing methods for automatic prompt optimization either lack flexibility or efficiency. In this paper, we propose an effective approach to automatically select the optimal prompt for a given input from a finite set of synthetic candidate prompts. Our approach consists of three steps: (1) clustering the training data and generating candidate prompts for each cluster using an LLM-based prompt generator; (2) synthesizing a dataset of input-prompt-output tuples for training a prompt evaluator to rank the prompts based on their relevance to the input; (3) using the prompt evaluator to select the best prompt for a new input at test time. Our approach balances prompt generality-specificity and eliminates the need for resource-intensive training and inference. It demonstrates competitive performance on zero-shot question-answering datasets: GSM8K, MultiArith, and AQuA., Comment: preprint

In this paper, we study the "dataset bias" problem from a statistical standpoint, and identify the main cause of the problem as the strong correlation between a class attribute u and a non-class attribute b in the input x, represented by p(u|b) differing significantly from p(u). Since p(u|b) appears as part of the sampling distributions in the standard maximum log-likelihood (MLL) objective, a model trained on a biased dataset via MLL inherently incorporates such correlation into its parameters, leading to poor generalization to unbiased test data. From this observation, we propose to mitigate dataset bias via either weighting the objective of each sample n by \frac{1}{p(u_{n}|b_{n})} or sampling that sample with a weight proportional to \frac{1}{p(u_{n}|b_{n})}. While both methods are statistically equivalent, the former proves more stable and effective in practice. Additionally, we establish a connection between our debiasing approach and causal reasoning, reinforcing our method's theoretical foundation. However, when the bias label is unavailable, computing p(u|b) exactly is difficult. To overcome this challenge, we propose to approximate \frac{1}{p(u|b)} using a biased classifier trained with "bias amplification" losses. Extensive experiments on various biased datasets demonstrate the superiority of our method over existing debiasing techniques in most settings, validating our theoretical analysis.

We propose a systematic training-free method to transform the probability flow of a "linear" stochastic process characterized by the equation X_{t}=a_{t}X_{0}+\sigma_{t}X_{1} into a straight constant-speed (SC) flow, reminiscent of Rectified Flow. This transformation facilitates fast sampling along the original probability flow via the Euler method without training a new model of the SC flow. The flexibility of our approach allows us to extend our transformation to inter-convert two posterior flows of two distinct linear stochastic processes. Moreover, we can easily integrate high-order numerical solvers into the transformed SC flow, further enhancing the sampling accuracy and efficiency. Rigorous theoretical analysis and extensive experimental results substantiate the advantages of our framework. Our code is available at this [https://github.com/clarken92/VFM||link]., Comment: Our code is available at: https://github.com/clarken92/VFM

The accurate prediction of survival times for patients with severe diseases remains a critical challenge despite recent advances in artificial intelligence. This study introduces "SurvTimeSurvival: Survival Analysis On Patients With Multiple Visits/Records", utilizing the Transformer model to not only handle the complexities of time-varying covariates but also covariates data. We also tackle the data sparsity issue common to survival analysis datasets by integrating synthetic data generation into the learning process of our model. We show that our method outperforms state-of-the-art deep learning approaches on both covariates and time-varying covariates datasets. Our approach aims not only to enhance the understanding of individual patient survival trajectories across various medical conditions, thereby improving prediction accuracy, but also to play a pivotal role in designing clinical trials and creating new treatments., Comment: Accepted as Findings Track in Machine Learning For Health (ML4H) 2023

We address catastrophic forgetting issues in graph learning as incoming data transits from one to another graph distribution. Whereas prior studies primarily tackle one setting of graph continual learning such as incremental node classification, we focus on a universal approach wherein each data point in a task can be a node or a graph, and the task varies from node to graph classification. We propose a novel method that enables graph neural networks to excel in this universal setting. Our approach perseveres knowledge about past tasks through a rehearsal mechanism that maintains local and global structure consistency across the graphs. We benchmark our method against various continual learning baselines in real-world graph datasets and achieve significant improvement in average performance and forgetting across tasks., Comment: 16 pages

We present a new computing model for intrinsic rewards in reinforcement learning that addresses the limitations of existing surprise-driven explorations. The reward is the novelty of the surprise rather than the surprise norm. We estimate the surprise novelty as retrieval errors of a memory network wherein the memory stores and reconstructs surprises. Our surprise memory (SM) augments the capability of surprise-based intrinsic motivators, maintaining the agent's interest in exciting exploration while reducing unwanted attraction to unpredictable or noisy observations. Our experiments demonstrate that the SM combined with various surprise predictors exhibits efficient exploring behaviors and significantly boosts the final performance in sparse reward environments, including Noisy-TV, navigation and challenging Atari games., Comment: 17 pages including Appendix

Large language models (LLMs) pretrained on vast source code have achieved prominent progress in code intelligence. However, existing code LLMs have two main limitations in terms of architecture and pretraining tasks. First, they often adopt a specific architecture (encoder-only or decoder-only) or rely on a unified encoder-decoder network for different downstream tasks. The former paradigm is limited by inflexibility in applications while in the latter, the model is treated as a single system for all tasks, leading to suboptimal performance on a subset of tasks. Secondly, they often employ a limited set of pretraining objectives which might not be relevant to some downstream tasks and hence result in substantial performance degrade. To address these limitations, we propose ``CodeT5+'', a family of encoder-decoder LLMs for code in which component modules can be flexibly combined to suit a wide range of downstream code tasks. Such flexibility is enabled by our proposed mixture of pretraining objectives to mitigate the pretrain-finetune discrepancy. These objectives cover span denoising, contrastive learning, text-code matching, and causal LM pretraining tasks, on both unimodal and bimodal multilingual code corpora. Furthermore, we propose to initialize CodeT5+ with frozen off-the-shelf LLMs without training from scratch to efficiently scale up our models, and explore instruction-tuning to align with natural language instructions. We extensively evaluate CodeT5+ on over 20 code-related benchmarks in different settings, including zero-shot, finetuning, and instruction-tuning. We observe state-of-the-art (SoTA) model performance on various code-related tasks, such as code generation and completion, math programming, and text-to-code retrieval tasks. Particularly, our instruction-tuned CodeT5+ 16B achieves new SoTA results on HumanEval code generation task against other open code LLMs., Comment: 26 pages, preprint

In this paper we introduce BO-Muse, a new approach to human-AI teaming for the optimization of expensive black-box functions. Inspired by the intrinsic difficulty of extracting expert knowledge and distilling it back into AI models and by observations of human behavior in real-world experimental design, our algorithm lets the human expert take the lead in the experimental process. The human expert can use their domain expertise to its full potential, while the AI plays the role of a muse, injecting novelty and searching for areas of weakness to break the human out of over-exploitation induced by cognitive entrenchment. With mild assumptions, we show that our algorithm converges sub-linearly, at a rate faster than the AI or human alone. We validate our algorithm using synthetic data and with human experts performing real-world experiments., Comment: 34 Pages, 7 Figures and 5 Tables

Social reasoning necessitates the capacity of theory of mind (ToM), the ability to contextualise and attribute mental states to others without having access to their internal cognitive structure. Recent machine learning approaches to ToM have demonstrated that we can train the observer to read the past and present behaviours of other agents and infer their beliefs (including false beliefs about things that no longer exist), goals, intentions and future actions. The challenges arise when the behavioural space is complex, demanding skilful space navigation for rapidly changing contexts for an extended period. We tackle the challenges by equipping the observer with novel neural memory mechanisms to encode, and hierarchical attention to selectively retrieve information about others. The memories allow rapid, selective querying of distal related past behaviours of others to deliberatively reason about their current mental state, beliefs and future behaviours. This results in ToMMY, a theory of mind model that learns to reason while making little assumptions about the underlying mental processes. We also construct a new suite of experiments to demonstrate that memories facilitate the learning process and achieve better theory of mind performance, especially for high-demand false-belief tasks that require inferring through multiple steps of changes., Comment: Accepted for publication at AAAI 2023

The capacity to achieve out-of-distribution (OOD) generalization is a hallmark of human intelligence and yet remains out of reach for machines. This remarkable capability has been attributed to our abilities to make conceptual abstraction and analogy, and to a mechanism known as indirection, which binds two representations and uses one representation to refer to the other. Inspired by these mechanisms, we hypothesize that OOD generalization may be achieved by performing analogy-making and indirection in the functional space instead of the data space as in current methods. To realize this, we design FINE (Functional Indirection Neural Estimator), a neural framework that learns to compose functions that map data input to output on-the-fly. FINE consists of a backbone network and a trainable semantic memory of basis weight matrices. Upon seeing a new input-output data pair, FINE dynamically constructs the backbone weights by mixing the basis weights. The mixing coefficients are indirectly computed through querying a separate corresponding semantic memory using the data pair. We demonstrate empirically that FINE can strongly improve out-of-distribution generalization on IQ tasks that involve geometric transformations. In particular, we train FINE and competing models on IQ tasks using images from the MNIST, Omniglot and CIFAR100 datasets and test on tasks with unseen image classes from one or different datasets and unseen transformation rules. FINE not only achieves the best performance on all tasks but also is able to adapt to small-scale data scenarios., Comment: Accepted paper at NeurIPS 2022

Successful Artificial Intelligence systems often require numerous labeled data to extract information from document images. In this paper, we investigate the problem of improving the performance of Artificial Intelligence systems in understanding document images, especially in cases where training data is limited. We address the problem by proposing a novel finetuning method using reinforcement learning. Our approach treats the Information Extraction model as a policy network and uses policy gradient training to update the model to maximize combined reward functions that complement the traditional cross-entropy losses. Our experiments on four datasets using labels and expert feedback demonstrate that our finetuning mechanism consistently improves the performance of a state-of-the-art information extractor, especially in the small training data regime., Comment: Accepted to ICONIP 2022

We introduce LAVIS, an open-source deep learning library for LAnguage-VISion research and applications. LAVIS aims to serve as a one-stop comprehensive library that brings recent advancements in the language-vision field accessible for researchers and practitioners, as well as fertilizing future research and development. It features a unified interface to easily access state-of-the-art image-language, video-language models and common datasets. LAVIS supports training, evaluation and benchmarking on a rich variety of tasks, including multimodal classification, retrieval, captioning, visual question answering, dialogue and pre-training. In the meantime, the library is also highly extensible and configurable, facilitating future development and customization. In this technical report, we describe design principles, key components and functionalities of the library, and also present benchmarking results across common language-vision tasks. The library is available at: https://github.com/salesforce/LAVIS., Comment: Preprint of LAVIS technical report

Program synthesis or code generation aims to generate a program that satisfies a problem specification. Recent approaches using large-scale pretrained language models (LMs) have shown promising results, yet they have some critical limitations. In particular, they often follow a standard supervised fine-tuning procedure to train a code generation model only from the pairs of natural-language problem descriptions and ground-truth programs. Such paradigm largely ignores some important but potentially useful signals in the problem specification such as unit tests, which thus often results in poor performance when solving complex unseen coding tasks. To address the limitations, we propose "CodeRL", a new framework for program synthesis tasks through pretrained LMs and deep reinforcement learning (RL). Specifically, during training, we treat the code-generating LM as an actor network, and introduce a critic network that is trained to predict the functional correctness of generated programs and provide dense feedback signals to the actor. During inference, we introduce a new generation procedure with a critical sampling strategy that allows a model to automatically regenerate programs based on feedback from example unit tests and critic scores. For the model backbones, we extended the encoder-decoder architecture of CodeT5 with enhanced learning objectives, larger model sizes, and better pretraining data. Our method not only achieves new SOTA results on the challenging APPS benchmark, but also shows strong zero-shot transfer capability with new SOTA results on the simpler MBPP benchmark., Comment: An earlier version of the work was accepted to NeurIPS 2022

Designed for tracking user goals in dialogues, a dialogue state tracker is an essential component in a dialogue system. However, the research of dialogue state tracking has largely been limited to unimodality, in which slots and slot values are limited by knowledge domains (e.g. restaurant domain with slots of restaurant name and price range) and are defined by specific database schema. In this paper, we propose to extend the definition of dialogue state tracking to multimodality. Specifically, we introduce a novel dialogue state tracking task to track the information of visual objects that are mentioned in video-grounded dialogues. Each new dialogue utterance may introduce a new video segment, new visual objects, or new object attributes, and a state tracker is required to update these information slots accordingly. We created a new synthetic benchmark and designed a novel baseline, Video-Dialogue Transformer Network (VDTN), for this task. VDTN combines both object-level features and segment-level features and learns contextual dependencies between videos and dialogues to generate multimodal dialogue states. We optimized VDTN for a state generation task as well as a self-supervised video understanding task which recovers video segment or object representations. Finally, we trained VDTN to use the decoded states in a response prediction task. Together with comprehensive ablation and qualitative analysis, we discovered interesting insights towards building more capable multimodal dialogue systems., Comment: Accepted at NAACL 2022 (Oral)

We introduce OmniXAI (short for Omni eXplainable AI), an open-source Python library of eXplainable AI (XAI), which offers omni-way explainable AI capabilities and various interpretable machine learning techniques to address the pain points of understanding and interpreting the decisions made by machine learning (ML) in practice. OmniXAI aims to be a one-stop comprehensive library that makes explainable AI easy for data scientists, ML researchers and practitioners who need explanation for various types of data, models and explanation methods at different stages of ML process (data exploration, feature engineering, model development, evaluation, and decision-making, etc). In particular, our library includes a rich family of explanation methods integrated in a unified interface, which supports multiple data types (tabular data, images, texts, time-series), multiple types of ML models (traditional ML in Scikit-learn and deep learning models in PyTorch/TensorFlow), and a range of diverse explanation methods including "model-specific" and "model-agnostic" ones (such as feature-attribution explanation, counterfactual explanation, gradient-based explanation, etc). For practitioners, the library provides an easy-to-use unified interface to generate the explanations for their applications by only writing a few lines of codes, and also a GUI dashboard for visualization of different explanations for more insights about decisions. In this technical report, we present OmniXAI's design principles, system architectures, and major functionalities, and also demonstrate several example use cases across different types of data, tasks, and models., Comment: Github repo: https://github.com/salesforce/OmniXAI

We introduce a constrained optimization method for policy gradient reinforcement learning, which uses a virtual trust region to regulate each policy update. In addition to using the proximity of one single old policy as the normal trust region, we propose forming a second trust region through another virtual policy representing a wide range of past policies. We then enforce the new policy to stay closer to the virtual policy, which is beneficial if the old policy performs poorly. More importantly, we propose a mechanism to automatically build the virtual policy from a memory of past policies, providing a new capability for dynamically learning appropriate virtual trust regions during the optimization process. Our proposed method, dubbed Memory-Constrained Policy Optimization (MCPO), is examined in diverse environments, including robotic locomotion control, navigation with sparse rewards and Atari games, consistently demonstrating competitive performance against recent on-policy constrained policy gradient methods., Comment: Preprint, 22 pages

Machine learning of Theory of Mind (ToM) is essential to build social agents that co-live with humans and other agents. This capacity, once acquired, will help machines infer the mental states of others from observed contextual action trajectories, enabling future prediction of goals, intention, actions and successor representations. The underlying mechanism for such a prediction remains unclear, however. Inspired by the observation that humans often infer the character traits of others, then use it to explain behaviour, we propose a new neural ToM architecture that learns to generate a latent trait vector of an actor from the past trajectories. This trait vector then multiplicatively modulates the prediction mechanism via a `fast weights' scheme in the prediction neural network, which reads the current context and predicts the behaviour. We empirically show that the fast weights provide a good inductive bias to model the character traits of agents and hence improves mindreading ability. On the indirect assessment of false-belief understanding, the new ToM model enables more efficient helping behaviours., Comment: Accepted for publication at AAMAS 2022

Trojan attacks on deep neural networks are both dangerous and surreptitious. Over the past few years, Trojan attacks have advanced from using only a single input-agnostic trigger and targeting only one class to using multiple, input-specific triggers and targeting multiple classes. However, Trojan defenses have not caught up with this development. Most defense methods still make inadequate assumptions about Trojan triggers and target classes, thus, can be easily circumvented by modern Trojan attacks. To deal with this problem, we propose two novel "filtering" defenses called Variational Input Filtering (VIF) and Adversarial Input Filtering (AIF) which leverage lossy data compression and adversarial learning respectively to effectively purify potential Trojan triggers in the input at run time without making assumptions about the number of triggers/target classes or the input dependence property of triggers. In addition, we introduce a new defense mechanism called "Filtering-then-Contrasting" (FtC) which helps avoid the drop in classification accuracy on clean data caused by "filtering", and combine it with VIF/AIF to derive new defenses of this kind. Extensive experimental results and ablation studies show that our proposed defenses significantly outperform well-known baseline defenses in mitigating five advanced Trojan attacks including two recent state-of-the-art while being quite robust to small amounts of training data and large-norm triggers., Comment: Accepted to ECCV 2022

We introduce a novel training procedure for policy gradient methods wherein episodic memory is used to optimize the hyperparameters of reinforcement learning algorithms on-the-fly. Unlike other hyperparameter searches, we formulate hyperparameter scheduling as a standard Markov Decision Process and use episodic memory to store the outcome of used hyperparameters and their training contexts. At any policy update step, the policy learner refers to the stored experiences, and adaptively reconfigures its learning algorithm with the new hyperparameters determined by the memory. This mechanism, dubbed as Episodic Policy Gradient Training (EPGT), enables an episodic learning process, and jointly learns the policy and the learning algorithm's hyperparameters within a single run. Experimental results on both continuous and discrete environments demonstrate the advantage of using the proposed method in boosting the performance of various policy gradient algorithms., Comment: 19 pages

Episodic control enables sample efficiency in reinforcement learning by recalling past experiences from an episodic memory. We propose a new model-based episodic memory of trajectories addressing current limitations of episodic control. Our memory estimates trajectory values, guiding the agent towards good policies. Built upon the memory, we construct a complementary learning model via a dynamic hybrid control unifying model-based, episodic and habitual learning into a single architecture. Experiments demonstrate that our model allows significantly faster and better learning than other strong reinforcement learning agents across a variety of environments including stochastic and non-Markovian settings., Comment: 26 pages

The optimistic nature of the Q-learning target leads to an overestimation bias, which is an inherent problem associated with standard $Q-$learning. Such a bias fails to account for the possibility of low returns, particularly in risky scenarios. However, the existence of biases, whether overestimation or underestimation, need not necessarily be undesirable. In this paper, we analytically examine the utility of biased learning, and show that specific types of biases may be preferable, depending on the scenario. Based on this finding, we design a novel reinforcement learning algorithm, Balanced Q-learning, in which the target is modified to be a convex combination of a pessimistic and an optimistic term, whose associated weights are determined online, analytically. We prove the convergence of this algorithm in a tabular setting, and empirically demonstrate its superior learning performance in various environments., Comment: 26 pages, 11 figures

Sample-efficient generalisation of reinforcement learning approaches have always been a challenge, especially, for complex scenes with many components. In this work, we introduce Plug and Play Markov Decision Processes, an object-based representation that allows zero-shot integration of new objects from known object classes. This is achieved by representing the global transition dynamics as a union of local transition functions, each with respect to one active object in the scene. Transition dynamics from an object class can be pre-learnt and thus would be ready to use in a new environment. Each active object is also endowed with its reward function. Since there is no central reward function, addition or removal of objects can be handled efficiently by only updating the reward functions of objects involved. A new transfer learning mechanism is also proposed to adapt reward function in such cases. Experiments show that our representation can achieve sample-efficiency in a variety of set-ups.

Transfer in reinforcement learning is usually achieved through generalisation across tasks. Whilst many studies have investigated transferring knowledge when the reward function changes, they have assumed that the dynamics of the environments remain consistent. Many real-world RL problems require transfer among environments with different dynamics. To address this problem, we propose an approach based on successor features in which we model successor feature functions with Gaussian Processes permitting the source successor features to be treated as noisy measurements of the target successor feature function. Our theoretical analysis proves the convergence of this approach as well as the bounded error on modelling successor feature functions with Gaussian Processes in environments with both different dynamics and rewards. We demonstrate our method on benchmark datasets and show that it outperforms current baselines.

Intelligence necessitates memory. Without memory, humans fail to perform various nontrivial tasks such as reading novels, playing games or solving maths. As the ultimate goal of machine learning is to derive intelligent systems that learn and act automatically just like human, memory construction for machine is inevitable. Artificial neural networks model neurons and synapses in the brain by interconnecting computational units via weights, which is a typical class of machine learning algorithms that resembles memory structure. Their descendants with more complicated modeling techniques (a.k.a deep learning) have been successfully applied to many practical problems and demonstrated the importance of memory in the learning process of machinery systems. Recent progresses on modeling memory in deep learning have revolved around external memory constructions, which are highly inspired by computational Turing models and biological neuronal systems. Attention mechanisms are derived to support acquisition and retention operations on the external memory. Despite the lack of theoretical foundations, these approaches have shown promises to help machinery systems reach a higher level of intelligence. The aim of this thesis is to advance the understanding on memory and attention in deep learning. Its contributions include: (i) presenting a collection of taxonomies for memory, (ii) constructing new memory-augmented neural networks (MANNs) that support multiple control and memory units, (iii) introducing variability via memory in sequential generative models, (iv) searching for optimal writing operations to maximise the memorisation capacity in slot-based memory networks, and (v) simulating the Universal Turing Machine via Neural Stored-program Memory-a new kind of external memory for neural networks., Comment: PHD Thesis

Video-grounded dialogue systems aim to integrate video understanding and dialogue understanding to generate responses that are relevant to both the dialogue and video context. Most existing approaches employ deep learning models and have achieved remarkable performance, given the relatively small datasets available. However, the results are partly accomplished by exploiting biases in the datasets rather than developing multimodal reasoning, resulting in limited generalization. In this paper, we propose a novel approach of Compositional Counterfactual Contrastive Learning ($C^3$) to develop contrastive training between factual and counterfactual samples in video-grounded dialogues. Specifically, we design factual/counterfactual sampling based on the temporal steps in videos and tokens in dialogues and propose contrastive loss functions that exploit object-level or action-level variance. Different from prior approaches, we focus on contrastive hidden state representations among compositional output tokens to optimize the representation space in a generation setting. We achieved promising performance gains on the Audio-Visual Scene-Aware Dialogues (AVSD) benchmark and showed the benefits of our approach in grounding video and dialogue context., Comment: 24th Meeting of the Special Interest Group on Discourse and Dialogue (SIGDIAL)

Neural module networks (NMN) have achieved success in image-grounded tasks such as Visual Question Answering (VQA) on synthetic images. However, very limited work on NMN has been studied in the video-grounded dialogue tasks. These tasks extend the complexity of traditional visual tasks with the additional visual temporal variance and language cross-turn dependencies. Motivated by recent NMN approaches on image-grounded tasks, we introduce Video-grounded Neural Module Network (VGNMN) to model the information retrieval process in video-grounded language tasks as a pipeline of neural modules. VGNMN first decomposes all language components in dialogues to explicitly resolve any entity references and detect corresponding action-based inputs from the question. The detected entities and actions are used as parameters to instantiate neural module networks and extract visual cues from the video. Our experiments show that VGNMN can achieve promising performance on a challenging video-grounded dialogue benchmark as well as a video QA benchmark., Comment: Accepted at NAACL 2022 (Oral)

Compared to traditional visual question answering, video-grounded dialogues require additional reasoning over dialogue context to answer questions in a multi-turn setting. Previous approaches to video-grounded dialogues mostly use dialogue context as a simple text input without modelling the inherent information flows at the turn level. In this paper, we propose a novel framework of Reasoning Paths in Dialogue Context (PDC). PDC model discovers information flows among dialogue turns through a semantic graph constructed based on lexical components in each question and answer. PDC model then learns to predict reasoning paths over this semantic graph. Our path prediction model predicts a path from the current turn through past dialogue turns that contain additional visual cues to answer the current question. Our reasoning model sequentially processes both visual and textual information through this reasoning path and the propagated features are used to generate the answer. Our experimental results demonstrate the effectiveness of our method and provide additional insights on how models use semantic dependencies in a dialogue context to retrieve visual cues., Comment: Accepted at ICLR (International Conference on Learning Representations) 2021

A video-grounded dialogue system is required to understand both dialogue, which contains semantic dependencies from turn to turn, and video, which contains visual cues of spatial and temporal scene variations. Building such dialogue systems is a challenging problem, involving various reasoning types on both visual and language inputs. Existing benchmarks do not have enough annotations to thoroughly analyze dialogue systems and understand their capabilities and limitations in isolation. These benchmarks are also not explicitly designed to minimise biases that models can exploit without actual reasoning. To address these limitations, in this paper, we present DVD, a Diagnostic Dataset for Video-grounded Dialogues. The dataset is designed to contain minimal biases and has detailed annotations for the different types of reasoning over the spatio-temporal space of video. Dialogues are synthesized over multiple question turns, each of which is injected with a set of cross-turn semantic relationships. We use DVD to analyze existing approaches, providing interesting insights into their abilities and limitations. In total, DVD is built from $11k$ CATER synthetic videos and contains $10$ instances of $10$-round dialogues for each video, resulting in more than $100k$ dialogues and $1M$ question-answer pairs. Our code and dataset are publicly available at https://github.com/facebookresearch/DVDialogues., Comment: 20 pages, 14 figures, 8 tables

Video-grounded dialogues are very challenging due to (i) the complexity of videos which contain both spatial and temporal variations, and (ii) the complexity of user utterances which query different segments and/or different objects in videos over multiple dialogue turns. However, existing approaches to video-grounded dialogues often focus on superficial temporal-level visual cues, but neglect more fine-grained spatial signals from videos. To address this drawback, we propose Bi-directional Spatio-Temporal Learning (BiST), a vision-language neural framework for high-resolution queries in videos based on textual cues. Specifically, our approach not only exploits both spatial and temporal-level information, but also learns dynamic information diffusion between the two feature spaces through spatial-to-temporal and temporal-to-spatial reasoning. The bidirectional strategy aims to tackle the evolving semantics of user queries in the dialogue setting. The retrieved visual cues are used as contextual information to construct relevant responses to the users. Our empirical results and comprehensive qualitative analysis show that BiST achieves competitive performance and generates reasonable responses on a large-scale AVSD benchmark. We also adapt our BiST models to the Video QA setting, and substantially outperform prior approaches on the TGIF-QA benchmark.

Artificial Neural Networks are uniquely adroit at machine learning by processing data through a network of artificial neurons. The inter-neuronal connection weights represent the learnt Neural Program that instructs the network on how to compute the data. However, without an external memory to store Neural Programs, they are restricted to only one, overwriting learnt programs when trained on new data. This is functionally equivalent to a special-purpose computer. Here we design Neurocoder, an entirely new class of general-purpose conditional computational machines in which the neural network "codes" itself in a data-responsive way by composing relevant programs from a set of shareable, modular programs. This can be considered analogous to building Lego structures from simple Lego bricks. Notably, our bricks change their shape through learning. External memory is used to create, store and retrieve modular programs. Like today's stored-program computers, Neurocoder can now access diverse programs to process different data. Unlike manually crafted computer programs, Neurocoder creates programs through training. Integrating Neurocoder into current neural architectures, we demonstrate new capacity to learn modular programs, handle severe pattern shifts and remember old programs as new ones are learnt, and show substantial performance improvement in solving object recognition, playing video games and continual learning tasks. Such integration with Neurocoder increases the computation capability of any current neural network and endows it with entirely new capacity to reuse simple programs to build complex ones. For the first time a Neural Program is treated as a datum in memory, paving the ways for modular, recursive and procedural neural programming., Comment: 22 pages, preprint

Pre-trained language models have shown remarkable success in improving various downstream NLP tasks due to their ability to capture dependencies in textual data and generate natural responses. In this paper, we leverage the power of pre-trained language models for improving video-grounded dialogue, which is very challenging and involves complex features of different dynamics: (1) Video features which can extend across both spatial and temporal dimensions; and (2) Dialogue features which involve semantic dependencies over multiple dialogue turns. We propose a framework by extending GPT-2 models to tackle these challenges by formulating video-grounded dialogue tasks as a sequence-to-sequence task, combining both visual and textual representation into a structured sequence, and fine-tuning a large pre-trained GPT-2 network. Our framework allows fine-tuning language models to capture dependencies across multiple modalities over different levels of information: spatio-temporal level in video and token-sentence level in dialogue context. We achieve promising improvement on the Audio-Visual Scene-Aware Dialogues (AVSD) benchmark from DSTC7, which supports a potential direction in this line of research., Comment: Accepted at ACL 2020 (Short Paper)

Building an end-to-end conversational agent for multi-domain task-oriented dialogues has been an open challenge for two main reasons. First, tracking dialogue states of multiple domains is non-trivial as the dialogue agent must obtain complete states from all relevant domains, some of which might have shared slots among domains as well as unique slots specifically for one domain only. Second, the dialogue agent must also process various types of information across domains, including dialogue context, dialogue states, and database, to generate natural responses to users. Unlike the existing approaches that are often designed to train each module separately, we propose "UniConv" -- a novel unified neural architecture for end-to-end conversational systems in multi-domain task-oriented dialogues, which is designed to jointly train (i) a Bi-level State Tracker which tracks dialogue states by learning signals at both slot and domain level independently, and (ii) a Joint Dialogue Act and Response Generator which incorporates information from various input components and models dialogue acts and target responses simultaneously. We conduct comprehensive experiments in dialogue state tracking, context-to-text, and end-to-end settings on the MultiWOZ2.1 benchmark, achieving superior performance over competitive baselines., Comment: Accepted The 2020 Conference on Empirical Methods in Natural Language Processing

Audio-Visual Scene-Aware Dialog (AVSD) is an extension from Video Question Answering (QA) whereby the dialogue agent is required to generate natural language responses to address user queries and carry on conversations. This is a challenging task as it consists of video features of multiple modalities, including text, visual, and audio features. The agent also needs to learn semantic dependencies among user utterances and system responses to make coherent conversations with humans. In this work, we describe our submission to the AVSD track of the 8th Dialogue System Technology Challenge. We adopt dot-product attention to combine text and non-text features of input video. We further enhance the generation capability of the dialogue agent by adopting pointer networks to point to tokens from multiple source sequences in each generation step. Our systems achieve high performance in automatic metrics and obtain 5th and 6th place in human evaluation among all submissions., Comment: Accepted at DSTC Workshop at AAAI 2020

Recent efforts in Dialogue State Tracking (DST) for task-oriented dialogues have progressed toward open-vocabulary or generation-based approaches where the models can generate slot value candidates from the dialogue history itself. These approaches have shown good performance gain, especially in complicated dialogue domains with dynamic slot values. However, they fall short in two aspects: (1) they do not allow models to explicitly learn signals across domains and slots to detect potential dependencies among (domain, slot) pairs; and (2) existing models follow auto-regressive approaches which incur high time cost when the dialogue evolves over multiple domains and multiple turns. In this paper, we propose a novel framework of Non-Autoregressive Dialog State Tracking (NADST) which can factor in potential dependencies among domains and slots to optimize the models towards better prediction of dialogue states as a complete set rather than separate slots. In particular, the non-autoregressive nature of our method not only enables decoding in parallel to significantly reduce the latency of DST for real-time dialogue response generation, but also detect dependencies among slots at token level in addition to slot and domain level. Our empirical results show that our model achieves the state-of-the-art joint accuracy across all domains on the MultiWOZ 2.1 corpus, and the latency of our model is an order of magnitude lower than the previous state of the art as the dialogue history extends over time., Comment: Accepted at ICLR 2020. International Conference on Learning Representations (2020)

Heretofore, neural networks with external memory are restricted to single memory with lossy representations of memory interactions. A rich representation of relationships between memory pieces urges a high-order and segregated relational memory. In this paper, we propose to separate the storage of individual experiences (item memory) and their occurring relationships (relational memory). The idea is implemented through a novel Self-attentive Associative Memory (SAM) operator. Found upon outer product, SAM forms a set of associative memories that represent the hypothetical high-order relationships between arbitrary pairs of memory elements, through which a relational memory is constructed from an item memory. The two memories are wired into a single sequential model capable of both memorization and relational reasoning. We achieve competitive results with our proposed two-memory model in a diversity of machine learning tasks, from challenging synthetic problems to practical testbeds such as geometry, graph, reinforcement learning, and question answering., Comment: 19 pages

An important aspect of health monitoring is effective logging of food consumption. This can help management of diet-related diseases like obesity, diabetes, and even cardiovascular diseases. Moreover, food logging can help fitness enthusiasts, and people who wanting to achieve a target weight. However, food-logging is cumbersome, and requires not only taking additional effort to note down the food item consumed regularly, but also sufficient knowledge of the food item consumed (which is difficult due to the availability of a wide variety of cuisines). With increasing reliance on smart devices, we exploit the convenience offered through the use of smart phones and propose a smart-food logging system: FoodAI, which offers state-of-the-art deep-learning based image recognition capabilities. FoodAI has been developed in Singapore and is particularly focused on food items commonly consumed in Singapore. FoodAI models were trained on a corpus of 400,000 food images from 756 different classes. In this paper we present extensive analysis and insights into the development of this system. FoodAI has been deployed as an API service and is one of the components powering Healthy 365, a mobile app developed by Singapore's Heath Promotion Board. We have over 100 registered organizations (universities, companies, start-ups) subscribing to this service and actively receive several API requests a day. FoodAI has made food logging convenient, aiding smart consumption and a healthy lifestyle., Comment: Published at KDD 2019 (Applied Data Science track). Demo is accessible at https://foodai.org/

Neural networks powered with external memory simulate computer behaviors. These models, which use the memory to store data for a neural controller, can learn algorithms and other complex tasks. In this paper, we introduce a new memory to store weights for the controller, analogous to the stored-program memory in modern computer architectures. The proposed model, dubbed Neural Stored-program Memory, augments current memory-augmented neural networks, creating differentiable machines that can switch programs through time, adapt to variable contexts and thus resemble the Universal Turing Machine. A wide range of experiments demonstrate that the resulting machines not only excel in classical algorithmic problems, but also have potential for compositional, continual, few-shot learning and question-answering tasks., Comment: 27 pages

We propose a new deep recurrent neural network (RNN) architecture for sequential signal reconstruction. Our network is designed by unfolding the iterations of the proximal gradient method that solves the l1-l1 minimization problem. As such, our network leverages by design that signals have a sparse representation and that the difference between consecutive signal representations is also sparse. We evaluate the proposed model in the task of reconstructing video frames from compressive measurements and show that it outperforms several state-of-the-art RNN models.

Memory-augmented neural networks consisting of a neural controller and an external memory have shown potentials in long-term sequential learning. Current RAM-like memory models maintain memory accessing every timesteps, thus they do not effectively leverage the short-term memory held in the controller. We hypothesize that this scheme of writing is suboptimal in memory utilization and introduces redundant computation. To validate our hypothesis, we derive a theoretical bound on the amount of information stored in a RAM-like system and formulate an optimization problem that maximizes the bound. The proposed solution dubbed Uniform Writing is proved to be optimal under the assumption of equal timestep contributions. To relax this assumption, we introduce modifications to the original solution, resulting in a solution termed Cached Uniform Writing. This method aims to balance between maximizing memorization and forgetting via overwriting mechanisms. Through an extensive set of experiments, we empirically demonstrate the advantages of our solutions over other recurrent architectures, claiming the state-of-the-arts in various sequential modeling tasks., Comment: Accepted to ICLR'19, oral presentation