Search

The transformer architecture has catalyzed revolutionary advances in language modeling. However, recent architectural recipes, such as state-space models, have bridged the performance gap. Motivated by this, we examine the benefits of Convolution-Augmented Transformer (CAT) for recall, copying, and length generalization tasks. CAT incorporates convolutional filters in the K/Q/V embeddings of an attention layer. Through CAT, we show that the locality of the convolution synergizes with the global view of the attention. Unlike comparable architectures, such as Mamba or transformer, CAT can provably solve the associative recall (AR) and copying tasks using a single layer while also enjoying guaranteed length generalization. We also establish computational tradeoffs between convolution and attention by characterizing how convolution can mitigate the need for full attention by summarizing the context window and creating salient summary tokens to attend. Evaluations on real datasets corroborate our findings and demonstrate that CAT and its variations indeed enhance the language modeling performance.

Medical Large Language Models (LLMs) such as ClinicalCamel 70B, Llama3-OpenBioLLM 70B have demonstrated impressive performance on a wide variety of medical NLP task.However, there still lacks a large language model (LLM) specifically designed for cancer domain. Moreover, these LLMs typically have billions of parameters, making them computationally expensive for healthcare systems.Thus, in this study, we propose CancerLLM, a model with 7 billion parameters and a Mistral-style architecture, pre-trained on 2,676,642 clinical notes and 515,524 pathology reports covering 17 cancer types, followed by fine-tuning on three cancer-relevant tasks, including cancer phenotypes extraction, cancer diagnosis generation, and cancer treatment plan generation. Our evaluation demonstrated that CancerLLM achieves state-of-the-art results compared to other existing LLMs, with an average F1 score improvement of 8.1\%. Additionally, CancerLLM outperforms other models on two proposed robustness testbeds. This illustrates that CancerLLM can be effectively applied to clinical AI systems, enhancing clinical research and healthcare delivery in the field of cancer.

Large language models (LLM) have demonstrated remarkable capabilities in various biomedical natural language processing (NLP) tasks, leveraging the demonstration within the input context to adapt to new tasks. However, LLM is sensitive to the selection of demonstrations. To address the hallucination issue inherent in LLM, retrieval-augmented LLM (RAL) offers a solution by retrieving pertinent information from an established database. Nonetheless, existing research work lacks rigorous evaluation of the impact of retrieval-augmented large language models on different biomedical NLP tasks. This deficiency makes it challenging to ascertain the capabilities of RAL within the biomedical domain. Moreover, the outputs from RAL are affected by retrieving the unlabeled, counterfactual, or diverse knowledge that is not well studied in the biomedical domain. However, such knowledge is common in the real world. Finally, exploring the self-awareness ability is also crucial for the RAL system. So, in this paper, we systematically investigate the impact of RALs on 5 different biomedical tasks (triple extraction, link prediction, classification, question answering, and natural language inference). We analyze the performance of RALs in four fundamental abilities, including unlabeled robustness, counterfactual robustness, diverse robustness, and negative awareness. To this end, we proposed an evaluation framework to assess the RALs' performance on different biomedical NLP tasks and establish four different testbeds based on the aforementioned fundamental abilities. Then, we evaluate 3 representative LLMs with 3 different retrievers on 5 tasks over 9 datasets.

Large Language Models (LLMs) have swiftly emerged as vital resources for different applications in the biomedical and healthcare domains; however, these models encounter issues such as generating inaccurate information or hallucinations. Retrieval-augmented generation provided a solution for these models to update knowledge and enhance their performance. In contrast to previous retrieval-augmented LMs, which utilize specialized cross-attention mechanisms to help LLM encode retrieved text, BiomedRAG adopts a simpler approach by directly inputting the retrieved chunk-based documents into the LLM. This straightforward design is easily applicable to existing retrieval and language models, effectively bypassing noise information in retrieved documents, particularly in noise-intensive tasks. Moreover, we demonstrate the potential for utilizing the LLM to supervise the retrieval model in the biomedical domain, enabling it to retrieve the document that assists the LM in improving its predictions. Our experiments reveal that with the tuned scorer,\textsc{ BiomedRAG} attains superior performance across 5 biomedical NLP tasks, encompassing information extraction (triple extraction, relation extraction), text classification, link prediction, and question-answering, leveraging over 9 datasets. For instance, in the triple extraction task, \textsc{BiomedRAG} outperforms other triple extraction systems with micro-F1 scores of 81.42 and 88.83 on GIT and ChemProt corpora, respectively.

Fine-tuning Pre-trained protein language models (PLMs) has emerged as a prominent strategy for enhancing downstream prediction tasks, often outperforming traditional supervised learning approaches. As a widely applied powerful technique in natural language processing, employing Parameter-Efficient Fine-Tuning techniques could potentially enhance the performance of PLMs. However, the direct transfer to life science tasks is non-trivial due to the different training strategies and data forms. To address this gap, we introduce SES-Adapter, a simple, efficient, and scalable adapter method for enhancing the representation learning of PLMs. SES-Adapter incorporates PLM embeddings with structural sequence embeddings to create structure-aware representations. We show that the proposed method is compatible with different PLM architectures and across diverse tasks. Extensive evaluations are conducted on 2 types of folding structures with notable quality differences, 9 state-of-the-art baselines, and 9 benchmark datasets across distinct downstream tasks. Results show that compared to vanilla PLMs, SES-Adapter improves downstream task performance by a maximum of 11% and an average of 3%, with significantly accelerated training speed by a maximum of 1034% and an average of 362%, the convergence rate is also improved by approximately 2 times. Moreover, positive optimization is observed even with low-quality predicted structures. The source code for SES-Adapter is available at https://github.com/tyang816/SES-Adapter., Comment: 30 pages, 4 figures, 8 tables

Zero-shot link prediction (ZSLP) on knowledge graphs aims at automatically identifying relations between given entities. Existing methods primarily employ auxiliary information to predict tail entity given head entity and its relation, yet face challenges due to the occasional unavailability of such detailed information and the inherent simplicity of predicting tail entities based on semantic similarities. Even though Large Language Models (LLMs) offer a promising solution to predict unobserved relations between the head and tail entity in a zero-shot manner, their performance is still restricted due to the inability to leverage all the (exponentially many) paths' information between two entities, which are critical in collectively indicating their relation types. To address this, in this work, we introduce a Condensed Transition Graph Framework for Zero-Shot Link Prediction (CTLP), which encodes all the paths' information in linear time complexity to predict unseen relations between entities, attaining both efficiency and information preservation. Specifically, we design a condensed transition graph encoder with theoretical guarantees on its coverage, expressiveness, and efficiency. It is learned by a transition graph contrastive learning strategy. Subsequently, we design a soft instruction tuning to learn and map the all-path embedding to the input of LLMs. Experimental results show that our proposed CTLP method achieves state-of-the-art performance on three standard ZSLP datasets

Accurately modeling the protein fitness landscapes holds great importance for protein engineering. Recently, due to their capacity and representation ability, pre-trained protein language models have achieved state-of-the-art performance in predicting protein fitness without experimental data. However, their predictions are limited in accuracy as well as interpretability. Furthermore, such deep learning models require abundant labeled training examples for performance improvements, posing a practical barrier. In this work, we introduce FSFP, a training strategy that can effectively optimize protein language models under extreme data scarcity. By combining the techniques of meta-transfer learning, learning to rank, and parameter-efficient fine-tuning, FSFP can significantly boost the performance of various protein language models using merely tens of labeled single-site mutants from the target protein. The experiments across 87 deep mutational scanning datasets underscore its superiority over both unsupervised and supervised approaches, revealing its potential in facilitating AI-guided protein design.

Modern classification problems exhibit heterogeneities across individual classes: Each class may have unique attributes, such as sample size, label quality, or predictability (easy vs difficult), and variable importance at test-time. Without care, these heterogeneities impede the learning process, most notably, when optimizing fairness objectives. Confirming this, under a gaussian mixture setting, we show that the optimal SVM classifier for balanced accuracy needs to be adaptive to the class attributes. This motivates us to propose CAP: An effective and general method that generates a class-specific learning strategy (e.g. hyperparameter) based on the attributes of that class. This way, optimization process better adapts to heterogeneities. CAP leads to substantial improvements over the naive approach of assigning separate hyperparameters to each class. We instantiate CAP for loss function design and post-hoc logit adjustment, with emphasis on label-imbalanced problems. We show that CAP is competitive with prior art and its flexibility unlocks clear benefits for fairness objectives beyond balanced accuracy. Finally, we evaluate CAP on problems with label noise as well as weighted test objectives to showcase how CAP can jointly adapt to different heterogeneities., Comment: 15 pages, 8 figures

Biomedical triple extraction systems aim to automatically extract biomedical entities and relations between entities. The exploration of applying large language models (LLM) to triple extraction is still relatively unexplored. In this work, we mainly focus on sentence-level biomedical triple extraction. Furthermore, the absence of a high-quality biomedical triple extraction dataset impedes the progress in developing robust triple extraction systems. To address these challenges, initially, we compare the performance of various large language models. Additionally, we present GIT, an expert-annotated biomedical triple extraction dataset that covers a wider range of relation types., Comment: this is the onging work

Large protein language models are adept at capturing the underlying evolutionary information in primary structures, offering significant practical value for protein engineering. Compared to natural language models, protein amino acid sequences have a smaller data volume and a limited combinatorial space. Choosing an appropriate vocabulary size to optimize the pre-trained model is a pivotal issue. Moreover, despite the wealth of benchmarks and studies in the natural language community, there remains a lack of a comprehensive benchmark for systematically evaluating protein language model quality. Given these challenges, PETA trained language models with 14 different vocabulary sizes under three tokenization methods. It conducted thousands of tests on 33 diverse downstream datasets to assess the models' transfer learning capabilities, incorporating two classification heads and three random seeds to mitigate potential biases. Extensive experiments indicate that vocabulary sizes between 50 and 200 optimize the model, whereas sizes exceeding 800 detrimentally affect the model's representational performance. Our code, model weights and datasets are available at https://github.com/ginnm/ProteinPretraining., Comment: 46 pages, 4figures, 9 tables

Reinforcement learning (RL) has emerged as a powerful approach for tackling complex medical decision-making problems such as treatment planning, personalized medicine, and optimizing the scheduling of surgeries and appointments. It has gained significant attention in the field of Natural Language Processing (NLP) due to its ability to learn optimal strategies for tasks such as dialogue systems, machine translation, and question-answering. This paper presents a review of the RL techniques in NLP, highlighting key advancements, challenges, and applications in healthcare. The review begins by visualizing a roadmap of machine learning and its applications in healthcare. And then it explores the integration of RL with NLP tasks. We examined dialogue systems where RL enables the learning of conversational strategies, RL-based machine translation models, question-answering systems, text summarization, and information extraction. Additionally, ethical considerations and biases in RL-NLP systems are addressed.

Neural Machine Translation (NMT) models have become successful, but their performance remains poor when translating on new domains with a limited number of data. In this paper, we present a novel approach Epi-Curriculum to address low-resource domain adaptation (DA), which contains a new episodic training framework along with denoised curriculum learning. Our episodic training framework enhances the model's robustness to domain shift by episodically exposing the encoder/decoder to an inexperienced decoder/encoder. The denoised curriculum learning filters the noised data and further improves the model's adaptability by gradually guiding the learning process from easy to more difficult tasks. Experiments on English-German and English-Romanian translation show that: (i) Epi-Curriculum improves both model's robustness and adaptability in seen and unseen domains; (ii) Our episodic training framework enhances the encoder and decoder's robustness to domain shift.

Generative large language models (LLMs) have shown great success in various applications, including question-answering (QA) and dialogue systems. However, in specialized domains like traditional Chinese medical QA, these models may perform unsatisfactorily without fine-tuning on domain-specific datasets. To address this, we introduce MedChatZH, a dialogue model designed specifically for traditional Chinese medical QA. Our model is pre-trained on Chinese traditional medical books and fine-tuned with a carefully curated medical instruction dataset. It outperforms several solid baselines on a real-world medical dialogue dataset. We release our model, code, and dataset on https://github.com/tyang816/MedChatZH to facilitate further research in the domain of traditional Chinese medicine and LLMs., Comment: 7 pages, 3 figures

Designing protein mutants of both high stability and activity is a critical yet challenging task in protein engineering. Here, we introduce Pro-PRIME, a deep learning zero-shot model, which can suggest protein mutants of improved stability and activity without any prior experimental mutagenesis data. By leveraging temperature-guided language modelling, Pro-PRIME demonstrated superior predictive power compared to current state-of-the-art models on the public mutagenesis dataset over 33 proteins. Furthermore, we carried out wet experiments to test Pro-PRIME on five distinct proteins to engineer certain physicochemical properties, including thermal stability, rates of RNA polymerization and DNA cleavage, hydrolase activity, antigen-antibody binding affinity, or even the nonnatural properties, e.g., the ability to polymerize non-natural nucleic acid or resilience to extreme alkaline conditions. Surprisingly, about 40% AI-designed mutants show better performance than the one before mutation for all five proteins studied and for all properties targeted for engineering. Hence, Pro-PRIME demonstrates the general applicability in protein engineering., Comment: arXiv admin note: text overlap with arXiv:2304.03780

The growth and diversity of machine learning applications motivate a rethinking of learning with mobile and edge devices. How can we address diverse client goals and learn with scarce heterogeneous data? While federated learning aims to address these issues, it has challenges hindering a unified solution. Large transformer models have been shown to work across a variety of tasks achieving remarkable few-shot adaptation. This raises the question: Can clients use a single general-purpose model, rather than custom models for each task, while obeying device and network constraints? In this work, we investigate pretrained transformers (PTF) to achieve these on-device learning goals and thoroughly explore the roles of model size and modularity, where the latter refers to adaptation through modules such as prompts or adapters. Focusing on federated learning, we demonstrate that: (1) Larger scale shrinks the accuracy gaps between alternative approaches and improves heterogeneity robustness. Scale allows clients to run more local SGD epochs which can significantly reduce the number of communication rounds. At the extreme, clients can achieve respectable accuracy locally highlighting the potential of fully-local learning. (2) Modularity, by design, enables $>$100$\times$ less communication in bits. Surprisingly, it also boosts the generalization capability of local adaptation methods and the robustness of smaller PTFs. Finally, it enables clients to solve multiple unrelated tasks simultaneously using a single PTF, whereas full updates are prone to catastrophic forgetting. These insights on scale and modularity motivate a new federated learning approach we call "You Only Load Once" (FedYolo): The clients load a full PTF model once and all future updates are accomplished through communication-efficient modules with limited catastrophic-forgetting, where each task is assigned to its own module., Comment: 20 pages, 18 figures

Recent advancements in language models (LMs) have led to the emergence of powerful models such as Small LMs (e.g., T5) and Large LMs (e.g., GPT-4). These models have demonstrated exceptional capabilities across a wide range of tasks, such as name entity recognition (NER) in the general domain. (We define SLMs as pre-trained models with fewer parameters compared to models like GPT-3/3.5/4, such as T5, BERT, and others.) Nevertheless, their efficacy in the medical section remains uncertain and the performance of medical NER always needs high accuracy because of the particularity of the field. This paper aims to provide a thorough investigation to compare the performance of LMs in medical few-shot NER and answer How far is LMs from 100\% Few-shot NER in Medical Domain, and moreover to explore an effective entity recognizer to help improve the NER performance. Based on our extensive experiments conducted on 16 NER models spanning from 2018 to 2023, our findings clearly indicate that LLMs outperform SLMs in few-shot medical NER tasks, given the presence of suitable examples and appropriate logical frameworks. Despite the overall superiority of LLMs in few-shot medical NER tasks, it is important to note that they still encounter some challenges, such as misidentification, wrong template prediction, etc. Building on previous findings, we introduce a simple and effective method called \textsc{RT} (Retrieving and Thinking), which serves as retrievers, finding relevant examples, and as thinkers, employing a step-by-step reasoning process. Experimental results show that our proposed \textsc{RT} framework significantly outperforms the strong open baselines on the two open medical benchmark datasets, Comment: Published as a journal paper at Journal of the American Medical Informatics Association (JAMIA). arXiv admin note: text overlap with arXiv:2305.18624

Stochastic approximation with multiple coupled sequences (MSA) has found broad applications in machine learning as it encompasses a rich class of problems including bilevel optimization (BLO), multi-level compositional optimization (MCO), and reinforcement learning (specifically, actor-critic methods). However, designing provably-efficient federated algorithms for MSA has been an elusive question even for the special case of double sequence approximation (DSA). Towards this goal, we develop FedMSA which is the first federated algorithm for MSA, and establish its near-optimal communication complexity. As core novelties, (i) FedMSA enables the provable estimation of hypergradients in BLO and MCO via local client updates, which has been a notable bottleneck in prior theory, and (ii) our convergence guarantees are sensitive to the heterogeneity-level of the problem. We also incorporate momentum and variance reduction techniques to achieve further acceleration leading to near-optimal rates. Finally, we provide experiments that support our theory and demonstrate the empirical benefits of FedMSA. As an example, FedMSA enables order-of-magnitude savings in communication rounds compared to prior federated BLO schemes.

Contrastive learning has become a popular solution for few-shot Name Entity Recognization (NER). The conventional configuration strives to reduce the distance between tokens with the same labels and increase the distance between tokens with different labels. The effect of this setup may, however, in the medical domain, there are a lot of entities annotated as OUTSIDE (O), and they are undesirably pushed apart to other entities that are not labeled as OUTSIDE (O) by the current contrastive learning method end up with a noisy prototype for the semantic representation of the label, though there are many OUTSIDE (O) labeled entities are relevant to the labeled entities. To address this challenge, we propose a novel method named Weighted Prototypical Contrastive Learning for Medical Few Shot Named Entity Recognization (W-PROCER). Our approach primarily revolves around constructing the prototype-based contractive loss and weighting network. These components play a crucial role in assisting the model in differentiating the negative samples from OUTSIDE (O) tokens and enhancing the discrimination ability of contrastive learning. Experimental results show that our proposed W-PROCER framework significantly outperforms the strong baselines on the three medical benchmark datasets., Comment: Under Review

Open domain entity state tracking aims to predict reasonable state changes of entities (i.e., [attribute] of [entity] was [before_state] and [after_state] afterwards) given the action descriptions. It's important to many reasoning tasks to support human everyday activities. However, it's challenging as the model needs to predict an arbitrary number of entity state changes caused by the action while most of the entities are implicitly relevant to the actions and their attributes as well as states are from open vocabularies. To tackle these challenges, we propose a novel end-to-end Knowledge Informed framework for open domain Entity State Tracking, namely KIEST, which explicitly retrieves the relevant entities and attributes from external knowledge graph (i.e., ConceptNet) and incorporates them to autoregressively generate all the entity state changes with a novel dynamic knowledge grained encoder-decoder framework. To enforce the logical coherence among the predicted entities, attributes, and states, we design a new constraint decoding strategy and employ a coherence reward to improve the decoding process. Experimental results show that our proposed KIEST framework significantly outperforms the strong baselines on the public benchmark dataset OpenPI., Comment: Published as a conference paper at SIGIR 2023

We introduce TemPL, a novel deep learning approach for zero-shot prediction of protein stability and activity, harnessing temperature-guided language modeling. By assembling an extensive dataset of 96 million sequence-host bacterial strain optimal growth temperatures (OGTs) and {\Delta}Tm data for point mutations under consistent experimental conditions, we effectively compared TemPL with state-of-the-art models. Notably, TemPL demonstrated superior performance in predicting protein stability. An ablation study was conducted to elucidate the influence of OGT prediction and language modeling modules on TemPL's performance, revealing the importance of integrating both components. Consequently, TemPL offers considerable promise for protein engineering applications, facilitating the design of mutation sequences with enhanced stability and activity., Comment: This project has been terminated

Deep learning has been widely used for protein engineering. However, it is limited by the lack of sufficient experimental data to train an accurate model for predicting the functional fitness of high-order mutants. Here, we develop SESNet, a supervised deep-learning model to predict the fitness for protein mutants by leveraging both sequence and structure information, and exploiting attention mechanism. Our model integrates local evolutionary context from homologous sequences, the global evolutionary context encoding rich semantic from the universal protein sequence space and the structure information accounting for the microenvironment around each residue in a protein. We show that SESNet outperforms state-of-the-art models for predicting the sequence-function relationship on 26 deep mutational scanning datasets. More importantly, we propose a data augmentation strategy by leveraging the data from unsupervised models to pre-train our model. After that, our model can achieve strikingly high accuracy in prediction of the fitness of protein mutants, especially for the higher order variants (> 4 mutation sites), when finetuned by using only a small number of experimental mutation data (<50). The strategy proposed is of great practical value as the required experimental effort, i.e., producing a few tens of experimental mutation data on a given protein, is generally affordable by an ordinary biochemical group and can be applied on almost any protein.

With the development of machine learning and data science, data sharing is very common between companies and research institutes to avoid data scarcity. However, sharing original datasets that contain private information can cause privacy leakage. A reliable solution is to utilize private synthetic datasets which preserve statistical information from original datasets. In this paper, we propose MC-GEN, a privacy-preserving synthetic data generation method under differential privacy guarantee for machine learning classification tasks. MC-GEN applies multi-level clustering and differential private generative model to improve the utility of synthetic data. In the experimental evaluation, we evaluated the effects of parameters and the effectiveness of MC-GEN. The results showed that MC-GEN can achieve significant effectiveness under certain privacy guarantees on multiple classification tasks. Moreover, we compare MC-GEN with three existing methods. The results showed that MC-GEN outperforms other methods in terms of utility.

Standard federated optimization methods successfully apply to stochastic problems with single-level structure. However, many contemporary ML problems -- including adversarial robustness, hyperparameter tuning, and actor-critic -- fall under nested bilevel programming that subsumes minimax and compositional optimization. In this work, we propose \fedblo: A federated alternating stochastic gradient method to address general nested problems. We establish provable convergence rates for \fedblo in the presence of heterogeneous data and introduce variations for bilevel, minimax, and compositional optimization. \fedblo introduces multiple innovations including federated hypergradient computation and variance reduction to address inner-level heterogeneity. We complement our theory with experiments on hyperparameter \& hyper-representation learning and minimax optimization that demonstrate the benefits of our method in practice. Code is available at https://github.com/ucr-optml/FedNest., Comment: ICML 2022 (accepted as a long presentation), 34 pages, 6 figures

Due to the incompleteness of knowledge graphs (KGs), zero-shot link prediction (ZSLP) which aims to predict unobserved relations in KGs has attracted recent interest from researchers. A common solution is to use textual features of relations (e.g., surface name or textual descriptions) as auxiliary information to bridge the gap between seen and unseen relations. Current approaches learn an embedding for each word token in the text. These methods lack robustness as they suffer from the out-of-vocabulary (OOV) problem. Meanwhile, models built on character n-grams have the capability of generating expressive representations for OOV words. Thus, in this paper, we propose a Hierarchical N-Gram framework for Zero-Shot Link Prediction (HNZSLP), which considers the dependencies among character n-grams of the relation surface name for ZSLP. Our approach works by first constructing a hierarchical n-gram graph on the surface name to model the organizational structure of n-grams that leads to the surface name. A GramTransformer, based on the Transformer is then presented to model the hierarchical n-gram graph to construct the relation embedding for ZSLP. Experimental results show the proposed HNZSLP achieved state-of-the-art performance on two ZSLP datasets., Comment: Published as a conference paper at EMNLP Findings 2022

Building query graphs from natural language questions is an important step in complex question answering over knowledge graph (Complex KGQA). In general, a question can be correctly answered if its query graph is built correctly and the right answer is then retrieved by issuing the query graph against the KG. Therefore, this paper focuses on query graph generation from natural language questions. Existing approaches for query graph generation ignore the semantic structure of a question, resulting in a large number of noisy query graph candidates that undermine prediction accuracies. In this paper, we define six semantic structures from common questions in KGQA and develop a novel Structure-BERT to predict the semantic structure of a question. By doing so, we can first filter out noisy candidate query graphs, and then rank the remaining candidates with a BERT-based ranking model. Extensive experiments on two popular benchmarks MetaQA and WebQuestionsSP (WSP) demonstrate the effectiveness of our method as compared to state-of-the-arts., Comment: Published as a conference paper at COLING 2022

In continual learning (CL), the goal is to design models that can learn a sequence of tasks without catastrophic forgetting. While there is a rich set of techniques for CL, relatively little understanding exists on how representations built by previous tasks benefit new tasks that are added to the network. To address this, we study the problem of continual representation learning (CRL) where we learn an evolving representation as new tasks arrive. Focusing on zero-forgetting methods where tasks are embedded in subnetworks (e.g., PackNet), we first provide experiments demonstrating CRL can significantly boost sample efficiency when learning new tasks. To explain this, we establish theoretical guarantees for CRL by providing sample complexity and generalization error bounds for new tasks by formalizing the statistical benefits of previously-learned representations. Our analysis and experiments also highlight the importance of the order in which we learn the tasks. Specifically, we show that CL benefits if the initial tasks have large sample size and high "representation diversity". Diversity ensures that adding new tasks incurs small representation mismatch and can be learned with few samples while training only few additional nonzero weights. Finally, we ask whether one can ensure each task subnetwork to be efficient during inference time while retaining the benefits of representation learning. To this end, we propose an inference-efficient variation of PackNet called Efficient Sparse PackNet (ESPN) which employs joint channel & weight pruning. ESPN embeds tasks in channel-sparse subnets requiring up to 80% less FLOPs to compute while approximately retaining accuracy and is very competitive with a variety of baselines. In summary, this work takes a step towards data and compute-efficient CL with a representation learning perspective. GitHub page: https://github.com/ucr-optml/CtRL

Deep learning often requires a large amount of data. In real-world applications, e.g., healthcare applications, the data collected by a single organization (e.g., hospital) is often limited, and the majority of massive and diverse data is often segregated across multiple organizations. As such, it motivates the researchers to conduct distributed deep learning, where the data user would like to build DL models using the data segregated across multiple different data owners. However, this could lead to severe privacy concerns due to the sensitive nature of the data, thus the data owners would be hesitant and reluctant to participate. We propose LDP-DL, a privacy-preserving distributed deep learning framework via local differential privacy and knowledge distillation, where each data owner learns a teacher model using its own (local) private dataset, and the data user learns a student model to mimic the output of the ensemble of the teacher models. In the experimental evaluation, a comprehensive comparison has been made among our proposed approach (i.e., LDP-DL), DP-SGD, PATE and DP-FL, using three popular deep learning benchmark datasets (i.e., CIFAR10, MNIST and FashionMNIST). The experimental results show that LDP-DL consistently outperforms the other competitors in terms of privacy budget and model accuracy., Comment: 10 pages, 6 figures, 1 table. Submitted to IEEE Transactions on Knowledge and Data Engineering

Imbalanced datasets are commonplace in modern machine learning problems. The presence of under-represented classes or groups with sensitive attributes results in concerns about generalization and fairness. Such concerns are further exacerbated by the fact that large capacity deep nets can perfectly fit the training data and appear to achieve perfect accuracy and fairness during training, but perform poorly during test. To address these challenges, we propose AutoBalance, a bi-level optimization framework that automatically designs a training loss function to optimize a blend of accuracy and fairness-seeking objectives. Specifically, a lower-level problem trains the model weights, and an upper-level problem tunes the loss function by monitoring and optimizing the desired objective over the validation data. Our loss design enables personalized treatment for classes/groups by employing a parametric cross-entropy loss and individualized data augmentation schemes. We evaluate the benefits and performance of our approach for the application scenarios of imbalanced and group-sensitive classification. Extensive empirical evaluations demonstrate the benefits of AutoBalance over state-of-the-art approaches. Our experimental findings are complemented with theoretical insights on loss function design and the benefits of train-validation split. All code is available open-source.

Neural Architecture Search (NAS) is a popular method for automatically designing optimized architectures for high-performance deep learning. In this approach, it is common to use bilevel optimization where one optimizes the model weights over the training data (inner problem) and various hyperparameters such as the configuration of the architecture over the validation data (outer problem). This paper explores the statistical aspects of such problems with train-validation splits. In practice, the inner problem is often overparameterized and can easily achieve zero loss. Thus, a-priori it seems impossible to distinguish the right hyperparameters based on training loss alone which motivates a better understanding of the role of train-validation split. To this aim this work establishes the following results. (1) We show that refined properties of the validation loss such as risk and hyper-gradients are indicative of those of the true test loss. This reveals that the outer problem helps select the most generalizable model and prevent overfitting with a near-minimal validation sample size. This is established for continuous search spaces which are relevant for differentiable schemes. Extensions to transfer learning are developed in terms of the mismatch between training & validation distributions. (2) We establish generalization bounds for NAS problems with an emphasis on an activation search problem. When optimized with gradient-descent, we show that the train-validation procedure returns the best (model, architecture) pair even if all architectures can perfectly fit the training data to achieve zero error. (3) Finally, we highlight connections between NAS, multiple kernel learning, and low-rank matrix learning. The latter leads to novel insights where the solution of the outer problem can be accurately learned via efficient spectral methods to achieve near-minimal risk., Comment: ICML 2021

Active learning is the set of techniques for intelligently labeling large unlabeled datasets to reduce the labeling effort. In parallel, recent developments in self-supervised and semi-supervised learning (S4L) provide powerful techniques, based on data-augmentation, contrastive learning, and self-training, that enable superior utilization of unlabeled data which led to a significant reduction in required labeling in the standard machine learning benchmarks. A natural question is whether these paradigms can be unified to obtain superior results. To this aim, this paper provides a novel algorithmic framework integrating self-supervised pretraining, active learning, and consistency-regularized self-training. We conduct extensive experiments with our framework on CIFAR10 and CIFAR100 datasets. These experiments enable us to isolate and assess the benefits of individual components which are evaluated using state-of-the-art methods (e.g.~Core-Set, VAAL, simCLR, FixMatch). Our experiments reveal two key insights: (i) Self-supervised pre-training significantly improves semi-supervised learning, especially in the few-label regime, (ii) The benefit of active learning is undermined and subsumed by S4L techniques. Specifically, we fail to observe any additional benefit of state-of-the-art active learning algorithms when combined with state-of-the-art S4L techniques.

Model pruning is an essential procedure for building compact and computationally-efficient machine learning models. A key feature of a good pruning algorithm is that it accurately quantifies the relative importance of the model weights. While model pruning has a rich history, we still don't have a full grasp of the pruning mechanics even for relatively simple problems involving linear models or shallow neural nets. In this work, we provide a principled exploration of pruning by building on a natural notion of importance. For linear models, we show that this notion of importance is captured by covariance scaling which connects to the well-known Hessian-based pruning. We then derive asymptotic formulas that allow us to precisely compare the performance of different pruning methods. For neural networks, we demonstrate that the importance can be at odds with larger magnitudes and proper initialization is critical for magnitude-based pruning. Specifically, we identify settings in which weights become more important despite becoming smaller, which in turn leads to a catastrophic failure of magnitude-based pruning. Our results also elucidate that implicit regularization in the form of Hessian structure has a catalytic role in identifying the important weights, which dictate the pruning performance., Comment: 28 pages

Short text is becoming more and more popular on the web, such as Chat Message, SMS and Product Reviews. Accurately classifying short text is an important and challenging task. A number of studies have difficulties in addressing this problem because of the word ambiguity and data sparsity. To address this issue, we propose a knowledge powered attention with similarity matrix based convolutional neural network (KASM) model, which can compute comprehensive information by utilizing the knowledge and deep neural network. We use knowledge graph (KG) to enrich the semantic representation of short text, specially, the information of parent-entity is introduced in our model. Meanwhile, we consider the word interaction in the literal-level between short text and the representation of label, and utilize similarity matrix based convolutional neural network (CNN) to extract it. For the purpose of measuring the importance of knowledge, we introduce the attention mechanisms to choose the important information. Experimental results on five standard datasets show that our model significantly outperforms state-of-the-art methods., Comment: there is an error in this paper

Princeton WordNet (PWN) is a lexicon-semantic network based on cognitive linguistics, which promotes the development of natural language processing. Based on PWN, five Chinese wordnets have been developed to solve the problems of syntax and semantics. They include: Northeastern University Chinese WordNet (NEW), Sinica Bilingual Ontological WordNet (BOW), Southeast University Chinese WordNet (SEW), Taiwan University Chinese WordNet (CWN), Chinese Open WordNet (COW). By using them, we found that these word networks have low accuracy and coverage, and cannot completely portray the semantic network of PWN. So we decided to make a new Chinese wordnet called Multi-Fusion Chinese Wordnet (MCW) to make up those shortcomings. The key idea is to extend the SEW with the help of Oxford bilingual dictionary and Xinhua bilingual dictionary, and then correct it. More specifically, we used machine learning and manual adjustment in our corrections. Two standards were formulated to help our work. We conducted experiments on three tasks including relatedness calculation, word similarity and word sense disambiguation for the comparison of lemma's accuracy, at the same time, coverage also was compared. The results indicate that MCW can benefit from coverage and accuracy via our method. However, it still has room for improvement, especially with lemmas. In the future, we will continue to enhance the accuracy of MCW and expand the concepts in it., Comment: 7 pages. CICLing 2019: International Conference on Computational Linguistics and Intelligent Text Processing