TinyVQA: Compact Multimodal Deep Neural Network for Visual Question Answering on Resource-Constrained Devices

Traditional machine learning models often require powerful hardware, making them unsuitable for deployment on resource-limited devices. Tiny Machine Learning (tinyML) has emerged as a promising approach for running machine learning models on these devices, but integrating multiple data modalities into tinyML models still remains a challenge due to increased complexity, latency, and power consumption. This paper proposes TinyVQA, a novel multimodal deep neural network for visual question answering tasks that can be deployed on resource-constrained tinyML hardware. TinyVQA leverages a supervised attention-based model to learn how to answer questions about images using both vision and language modalities. Distilled knowledge from the supervised attention-based VQA model trains the memory aware compact TinyVQA model and low bit-width quantization technique is employed to further compress the model for deployment on tinyML devices. The TinyVQA model was evaluated on the FloodNet dataset, which is used for post-disaster damage assessment. The compact model achieved an accuracy of 79.5%, demonstrating the effectiveness of TinyVQA for real-world applications. Additionally, the model was deployed on a Crazyflie 2.0 drone, equipped with an AI deck and GAP8 microprocessor. The TinyVQA model achieved low latencies of 56 ms and consumes 693 mW power while deployed on the tiny drone, showcasing its suitability for resource-constrained embedded systems.

Squeezed Edge YOLO: Onboard Object Detection on Edge Devices

Demand for efficient onboard object detection is increasing due to its key role in autonomous navigation. However, deploying object detection models such as YOLO on resource constrained edge devices is challenging due to the high computational requirements of such models. In this paper, an compressed object detection model named Squeezed Edge YOLO is examined. This model is compressed and optimized to kilobytes of parameters in order to fit onboard such edge devices. To evaluate Squeezed Edge YOLO, two use cases - human and shape detection - are used to show the model accuracy and performance. Moreover, the model is deployed onboard a GAP8 processor with 8 RISC-V cores and an NVIDIA Jetson Nano with 4GB of memory. Experimental results show Squeezed Edge YOLO model size is optimized by a factor of 8x which leads to 76% improvements in energy efficiency and 3.3x faster throughout.

LLM Augmented Hierarchical Agents

Solving long-horizon, temporally-extended tasks using Reinforcement Learning (RL) is challenging, compounded by the common practice of learning without prior knowledge (or tabula rasa learning). Humans can generate and execute plans with temporally-extended actions and quickly learn to perform new tasks because we almost never solve problems from scratch. We want autonomous agents to have this same ability. Recently, LLMs have been shown to encode a tremendous amount of knowledge about the world and to perform impressive in-context learning and reasoning. However, using LLMs to solve real world problems is hard because they are not grounded in the current task. In this paper we exploit the planning capabilities of LLMs while using RL to provide learning from the environment, resulting in a hierarchical agent that uses LLMs to solve long-horizon tasks. Instead of completely relying on LLMs, they guide a high-level policy, making learning significantly more sample efficient. This approach is evaluated in simulation environments such as MiniGrid, SkillHack, and Crafter, and on a real robot arm in block manipulation tasks. We show that agents trained using our approach outperform other baselines methods and, once trained, don't need access to LLMs during deployment.

ReProHRL: Towards Multi-Goal Navigation in the Real World using Hierarchical Agents

Robots have been successfully used to perform tasks with high precision. In real-world environments with sparse rewards and multiple goals, learning is still a major challenge and Reinforcement Learning (RL) algorithms fail to learn good policies. Training in simulation environments and then fine-tuning in the real world is a common approach. However, adapting to the real-world setting is a challenge. In this paper, we present a method named Ready for Production Hierarchical RL (ReProHRL) that divides tasks with hierarchical multi-goal navigation guided by reinforcement learning. We also use object detectors as a pre-processing step to learn multi-goal navigation and transfer it to the real world. Empirical results show that the proposed ReProHRL method outperforms the state-of-the-art baseline in simulation and real-world environments in terms of both training time and performance. Although both methods achieve a 100% success rate in a simple environment for single goal-based navigation, in a more complex environment and multi-goal setting, the proposed method outperforms the baseline by 18% and 5%, respectively. For the real-world implementation and proof of concept demonstration, we deploy the proposed method on a nano-drone named Crazyflie with a front camera to perform multi-goal navigation experiments.

Towards an Interpretable Hierarchical Agent Framework using Semantic Goals

Learning to solve long horizon temporally extended tasks with reinforcement learning has been a challenge for several years now. We believe that it is important to leverage both the hierarchical structure of complex tasks and to use expert supervision whenever possible to solve such tasks. This work introduces an interpretable hierarchical agent framework by combining planning and semantic goal directed reinforcement learning. We assume access to certain spatial and haptic predicates and construct a simple and powerful semantic goal space. These semantic goal representations are more interpretable, making expert supervision and intervention easier. They also eliminate the need to write complex, dense reward functions thereby reducing human engineering effort. We evaluate our framework on a robotic block manipulation task and show that it performs better than other methods, including both sparse and dense reward functions. We also suggest some next steps and discuss how this framework makes interaction and collaboration with humans easier.

TinyM$^2$Net: A Flexible System Algorithm Co-designed Multimodal Learning Framework for Tiny Devices

With the emergence of Artificial Intelligence (AI), new attention has been given to implement AI algorithms on resource constrained tiny devices to expand the application domain of IoT. Multimodal Learning has recently become very popular with the classification task due to its impressive performance for both image and audio event classification. This paper presents TinyM$^2$Net -- a flexible system algorithm co-designed multimodal learning framework for resource constrained tiny devices. The framework was designed to be evaluated on two different case-studies: COVID-19 detection from multimodal audio recordings and battle field object detection from multimodal images and audios. In order to compress the model to implement on tiny devices, substantial network architecture optimization and mixed precision quantization were performed (mixed 8-bit and 4-bit). TinyM$^2$Net shows that even a tiny multimodal learning model can improve the classification performance than that of any unimodal frameworks. The most compressed TinyM$^2$Net achieves 88.4% COVID-19 detection accuracy (14.5% improvement from unimodal base model) and 96.8\% battle field object detection accuracy (3.9% improvement from unimodal base model). Finally, we test our TinyM$^2$Net models on a Raspberry Pi 4 to see how they perform when deployed to a resource constrained tiny device.

A Fast Network Exploration Strategy to Profile Low Energy Consumption for Keyword Spotting

Keyword Spotting nowadays is an integral part of speech-oriented user interaction targeted for smart devices. To this extent, neural networks are extensively used for their flexibility and high accuracy. However, coming up with a suitable configuration for both accuracy requirements and hardware deployment is a challenge. We propose a regression-based network exploration technique that considers the scaling of the network filters ($s$) and quantization ($q$) of the network layers, leading to a friendly and energy-efficient configuration for FPGA hardware implementation. We experiment with different combinations of $\mathcal{NN}\scriptstyle\langle q,\,s\rangle \displaystyle$ on the FPGA to profile the energy consumption of the deployed network so that the user can choose the most energy-efficient network configuration promptly. Our accelerator design is deployed on the Xilinx AC 701 platform and has at least 2.1$\times$ and 4$\times$ improvements on energy and energy efficiency results, respectively, compared to recent hardware implementations for keyword spotting.

Automatic Goal Generation using Dynamical Distance Learning

Reinforcement Learning (RL) agents can learn to solve complex sequential decision making tasks by interacting with the environment. However, sample efficiency remains a major challenge. In the field of multi-goal RL, where agents are required to reach multiple goals to solve complex tasks, improving sample efficiency can be especially challenging. On the other hand, humans or other biological agents learn such tasks in a much more strategic way, following a curriculum where tasks are sampled with increasing difficulty level in order to make gradual and efficient learning progress. In this work, we propose a method for automatic goal generation using a dynamical distance function (DDF) in a self-supervised fashion. DDF is a function which predicts the dynamical distance between any two states within a markov decision process (MDP). With this, we generate a curriculum of goals at the appropriate difficulty level to facilitate efficient learning throughout the training process. We evaluate this approach on several goal-conditioned robotic manipulation and navigation tasks, and show improvements in sample efficiency over a baseline method which only uses random goal sampling.

Interactive Hierarchical Guidance using Language

Reinforcement learning has been successful in many tasks ranging from robotic control, games, energy management etc. In complex real world environments with sparse rewards and long task horizons, sample efficiency is still a major challenge. Most complex tasks can be easily decomposed into high-level planning and low level control. Therefore, it is important to enable agents to leverage the hierarchical structure and decompose bigger tasks into multiple smaller sub-tasks. We introduce an approach where we use language to specify sub-tasks and a high-level planner issues language commands to a low level controller. The low-level controller executes the sub-tasks based on the language commands. Our experiments show that this method is able to solve complex long horizon planning tasks with limited human supervision. Using language has added benefit of interpretability and ability for expert humans to take over the high-level planning task and provide language commands if necessary.

Neural Networks for Pulmonary Disease Diagnosis using Auditory and Demographic Information

Pulmonary diseases impact millions of lives globally and annually. The recent outbreak of the pandemic of the COVID-19, a novel pulmonary infection, has more than ever brought the attention of the research community to the machine-aided diagnosis of respiratory problems. This paper is thus an effort to exploit machine learning for classification of respiratory problems and proposes a framework that employs as much correlated information (auditory and demographic information in this work) as a dataset provides to increase the sensitivity and specificity of a diagnosing system. First, we use deep convolutional neural networks (DCNNs) to process and classify a publicly released pulmonary auditory dataset, and then we take advantage of the existing demographic information within the dataset and show that the accuracy of the pulmonary classification increases by 5% when trained on the auditory information in conjunction with the demographic information. Since the demographic data can be extracted using computer vision, we suggest using another parallel DCNN to estimate the demographic information of the subject under test visioned by the processing computer. Lastly, as a proposition to bring the healthcare system to users' fingertips, we measure deployment characteristics of the auditory DCNN model onto processing components of an NVIDIA TX2 development board.