Multimodal learning

<h2 id="motivation">Motivation</h2>
Data usually comes with different modalities which carry different information. For example, it is very common to caption an image to convey the information not presented in the image itself. Similarly, sometimes it is more straightforward to use an image to describe information which may not be obvious from text. As a result, if different words appear in similar images, then these words likely describe the same thing. Conversely, if a word is used to describe seemingly dissimilar images, then these images may represent the same object. Thus, in cases dealing with multi-modal data, it is important to use a model which is able to jointly represent the information such that the model can capture the combined information from different modalities.

<h2 id="multimodal-transformers">Multimodal transformers</h2>
This section is an excerpt from <a href="/facts/Transformer_(deep_learning_architecture)/cDbjx6a8">Transformer (deep learning architecture) § Multimodality</a>.[<a href="https://en.wikipedia.org/w/index.php?title=Transformer_(deep_learning_architecture)&action=edit">edit</a>]
Transformers can also be used/adapted for modalities (input or output) beyond just text, usually by finding a way to "tokenize" the modality.
Multimodal models can either be trained from scratch, or by finetuning. A 2022 study found that Transformers pretrained only on natural language can be finetuned on only 0.03% of parameters and become competitive with LSTMs on a variety of logical and visual tasks, demonstrating <a href="/facts/Transfer_learning/pNz4P2KP">transfer learning</a>.<a class="footnote-ref" id="fnref:6" href="#fn:6">6</a> The LLaVA was a vision-language model composed of a language model (Vicuna-13B)<a class="footnote-ref" id="fnref:7" href="#fn:7">7</a> and a vision model (<a href="/facts/Vision_transformer/J7VbF9ae">ViT</a>-L/14), connected by a linear layer. Only the linear layer is finetuned.<a class="footnote-ref" id="fnref:8" href="#fn:8">8</a>
<a href="/facts/Vision_transformer/J7VbF9ae">Vision transformers</a><a class="footnote-ref" id="fnref:9" href="#fn:9">9</a> adapt the transformer to computer vision by breaking down input images as a series of patches, turning them into vectors, and treating them like tokens in a standard transformer.
Conformer<a class="footnote-ref" id="fnref:10" href="#fn:10">10</a> and later <a href="/facts/Whisper_(speech_recognition_system)/sRoHwvmv">Whisper</a><a class="footnote-ref" id="fnref:11" href="#fn:11">11</a> follow the same pattern for <a href="/facts/Speech_recognition/z7S7Pgk6">speech recognition</a>, first turning the speech signal into a <a href="/facts/Spectrogram/Jt8zHI0H">spectrogram</a>, which is then treated like an image, i.e. broken down into a series of patches, turned into vectors and treated like tokens in a standard transformer.
<a href="/facts/Perceiver/S94qHUcm">Perceivers</a><a class="footnote-ref" id="fnref:12" href="#fn:12">12</a><a class="footnote-ref" id="fnref:13" href="#fn:13">13</a> are a variant of Transformers designed for multimodality.

For image generation, notable architectures are <a href="/facts/DALL-E/rkWL8P49">DALL-E 1</a> (2021), Parti (2022),<a class="footnote-ref" id="fnref:14" href="#fn:14">14</a> Phenaki (2023),<a class="footnote-ref" id="fnref:15" href="#fn:15">15</a> and Muse (2023).<a class="footnote-ref" id="fnref:16" href="#fn:16">16</a> Unlike later models, DALL-E is not a diffusion model. Instead, it uses a decoder-only Transformer that autoregressively generates a text, followed by the token representation of an image, which is then converted by a <a href="/facts/Variational_autoencoder/I5wBbYpR">variational autoencoder</a> to an image.<a class="footnote-ref" id="fnref:17" href="#fn:17">17</a> Parti is an encoder-decoder Transformer, where the encoder processes a text prompt, and the decoder generates a token representation of an image.<a class="footnote-ref" id="fnref:18" href="#fn:18">18</a> Muse is an encoder-only Transformer that is trained to predict masked image tokens from unmasked image tokens. During generation, all input tokens are masked, and the highest-confidence predictions are included for the next iteration, until all tokens are predicted.<a class="footnote-ref" id="fnref:19" href="#fn:19">19</a> Phenaki is a text-to-video model. It is a bidirectional masked transformer conditioned on pre-computed text tokens. The generated tokens are then decoded to a video.<a class="footnote-ref" id="fnref:20" href="#fn:20">20</a>
<h3>Multimodal large language models</h3>
This section is an excerpt from <a href="/facts/Large_language_model/WnogWVJY">Large language model § Multimodality</a>.[<a href="https://en.wikipedia.org/w/index.php?title=Large_language_model&action=edit">edit</a>]
Multimodality means "having several modalities", and a <a href="/facts/Modality_(human%25E2%2580%2593computer_interaction)/KzL4uqMm">"modality"</a> refers to a type of input or output, such as video, image, audio, text, <a href="/facts/Proprioception/Ik79bA2x">proprioception</a>, etc.<a class="footnote-ref" id="fnref:21" href="#fn:21">21</a> There have been many AI models trained specifically to ingest one modality and output another modality, such as <a href="/facts/AlexNet/FUFE1Xov">AlexNet</a> for image to label,<a class="footnote-ref" id="fnref:22" href="#fn:22">22</a> <a href="/facts/Visual_question_answering/1gplmc6L">visual question answering</a> for image-text to text,<a class="footnote-ref" id="fnref:23" href="#fn:23">23</a> and <a href="/facts/Speech_recognition/z7S7Pgk6">speech recognition</a> for speech to text.
A common method to create multimodal models out of an LLM is to "tokenize" the output of a trained encoder. Concretely, one can construct an LLM that can understand images as follows: take a trained LLM, and take a trained image encoder 
 
 
 
 E
 
 
 {\displaystyle E}
 
. Make a small multilayered perceptron 
 
 
 
 f
 
 
 {\displaystyle f}
 
, so that for any image 
 
 
 
 y
 
 
 {\displaystyle y}
 
, the post-processed vector 
 
 
 
 f
 (
 E
 (
 y
 )
 )
 
 
 {\displaystyle f(E(y))}
 
 has the same dimensions as an encoded token. That is an "image token". Then, one can interleave text tokens and image tokens. The compound model is then fine-tuned on an image-text dataset. This basic construction can be applied with more sophistication to improve the model. The image encoder may be frozen to improve stability.<a class="footnote-ref" id="fnref:24" href="#fn:24">24</a>
Flamingo demonstrated the effectiveness of the tokenization method, finetuning a pair of pretrained language model and image encoder to perform better on visual question answering than models trained from scratch.<a class="footnote-ref" id="fnref:25" href="#fn:25">25</a> <a href="/facts/Pathways_Language_Model/wG7N5sz5">Google PaLM</a> model was fine-tuned into a multimodal model PaLM-E using the tokenization method, and applied to robotic control.<a class="footnote-ref" id="fnref:26" href="#fn:26">26</a> <a href="/facts/LLaMA/rmrWj35n">LLaMA</a> models have also been turned multimodal using the tokenization method, to allow image inputs,<a class="footnote-ref" id="fnref:27" href="#fn:27">27</a> and video inputs.<a class="footnote-ref" id="fnref:28" href="#fn:28">28</a>

<a href="/facts/GPT-4/MuXdd4sW">GPT-4</a> can use both text and image as inputs<a class="footnote-ref" id="fnref:29" href="#fn:29">29</a> (although the vision component was not released to the public until GPT-4V<a class="footnote-ref" id="fnref:30" href="#fn:30">30</a>); <a href="/facts/Google_DeepMind/n5ehBwCx">Google DeepMind</a>'s <a href="/facts/Gemini_(language_model)/eD5D8xBo">Gemini</a> is also multimodal.<a class="footnote-ref" id="fnref:31" href="#fn:31">31</a> Mistral introduced its own multimodal Pixtral 12B model in September 2024.<a class="footnote-ref" id="fnref:32" href="#fn:32">32</a>
<h2 id="multimodal-deep-boltzmann-machines">Multimodal deep Boltzmann machines</h2>
A <a href="/facts/Boltzmann_machine/2wyLI0pI">Boltzmann machine</a> is a type of <a href="/facts/Stochastic_neural_network/BKu7GCU3">stochastic neural network</a> invented by <a href="/facts/Geoffrey_Hinton/HJU6lC2H">Geoffrey Hinton</a> and <a href="/facts/Terry_Sejnowski/rckEWrwp">Terry Sejnowski</a> in 1985. Boltzmann machines can be seen as the <a href="/facts/Stochastic_process/Ng1n1dnB">stochastic</a>, <a href="/facts/Generative_model/JUgExNIP">generative</a> counterpart of <a href="/facts/Hopfield_net/xUBr6rpm">Hopfield nets</a>. They are named after the <a href="/facts/Boltzmann_distribution/XBU530uu">Boltzmann distribution</a> in statistical mechanics. The units in Boltzmann machines are divided into two groups: visible units and hidden units. Each unit is like a neuron with a binary output that represents whether it is activated or not.<a class="footnote-ref" id="fnref:33" href="#fn:33">33</a> General Boltzmann machines allow connection between any units. However, learning is impractical using general Boltzmann Machines because the computational time is exponential to the size of the machine. A more efficient architecture is called <a href="/facts/Restricted_Boltzmann_machine/D5afVGCs">restricted Boltzmann machine</a> where connection is only allowed between hidden unit and visible unit, which is described in the next section.
Multimodal deep Boltzmann machines can process and learn from different types of information, such as images and text, simultaneously. This can notably be done by having a separate deep Boltzmann machine for each modality, for example one for images and one for text, joined at an additional top hidden layer.<a class="footnote-ref" id="fnref:34" href="#fn:34">34</a>

<h2 id="applications">Applications</h2>
Multimodal machine learning has numerous applications across various domains:

<h3>Cross-Modal Retrieval</h3>
Cross-modal retrieval allows users to search for data across different modalities (e.g., retrieving images based on text descriptions), improving multimedia search engines and content recommendation systems. Models like <a href="/facts/Contrastive_Language-Image_Pre-training/rkWL8P49">CLIP</a> facilitate efficient, accurate retrieval by embedding data in a shared space, demonstrating strong performance even in zero-shot settings.<a class="footnote-ref" id="fnref:35" href="#fn:35">35</a>

<h3>Classification and Missing Data Retrieval</h3>
Multimodal Deep Boltzmann Machines outperform traditional models like <a href="/facts/Support_vector_machine/XobxpdBG">support vector machines</a> and <a href="/facts/Latent_Dirichlet_allocation/BUWlwKWt">latent Dirichlet allocation</a> in classification tasks and can predict missing data in multimodal datasets, such as images and text.

<h3>Healthcare Diagnostics</h3>
Multimodal models integrate medical imaging, genomic data, and patient records to improve diagnostic accuracy and early disease detection, especially in cancer screening.<a class="footnote-ref" id="fnref:36" href="#fn:36">36</a><a class="footnote-ref" id="fnref:37" href="#fn:37">37</a><a class="footnote-ref" id="fnref:38" href="#fn:38">38</a>

<h3>Content Generation</h3>
Models like DALL·E generate images from textual descriptions, benefiting creative industries, while cross-modal retrieval enables dynamic multimedia searches.<a class="footnote-ref" id="fnref:39" href="#fn:39">39</a>

<h3>Robotics and HCI</h3>
Multimodal learning improves interaction in robotics and AI by integrating sensory inputs like speech, vision, and touch, aiding autonomous systems and human-computer interaction.

<h3>Emotion Recognition</h3>
Combining visual, audio, and text data, multimodal systems enhance sentiment analysis and emotion recognition, applied in customer service, social media, and marketing.

<h2 id="see-also">See also</h2>
<ul><li><a href="/facts/Hopfield_network/xUBr6rpm">Hopfield network</a></li>
<li><a href="/facts/Markov_random_field/DJazwaeP">Markov random field</a></li>
<li><a href="/facts/Markov_chain_Monte_Carlo/eTjk3WQy">Markov chain Monte Carlo</a></li></ul>

<h2 id="references">References</h2>

<ol>
<li id="fn:1">Hendriksen, Mariya; Bleeker, Maurits; Vakulenko, Svitlana; van Noord, Nanne; Kuiper, Ernst; de Rijke, Maarten (2021). "Extending CLIP for Category-to-image Retrieval in E-commerce". arXiv:2112.11294 [cs.CV]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:1" class="footnote-back-ref">↩</a></li>
<li id="fn:2">"Stable Diffusion Repository on GitHub". CompVis - Machine Vision and Learning Research Group, LMU Munich. 17 September 2022. Archived from the original on January 18, 2023. Retrieved 17 September 2022. <a href="https://github.com/CompVis/stable-diffusion" target="_blank">https://github.com/CompVis/stable-diffusion</a> <a href="#fnref:2" class="footnote-back-ref">↩</a></li>
<li id="fn:3">LAION-AI/aesthetic-predictor, LAION AI, 2024-09-06, retrieved 2024-09-08 <a href="https://github.com/LAION-AI/aesthetic-predictor" target="_blank">https://github.com/LAION-AI/aesthetic-predictor</a> <a href="#fnref:3" class="footnote-back-ref">↩</a></li>
<li id="fn:4">Mokady, Ron; Hertz, Amir; Bermano, Amit H. (2021). "ClipCap: CLIP Prefix for Image Captioning". arXiv:2111.09734 [cs.CV]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:4" class="footnote-back-ref">↩</a></li>
<li id="fn:5">Zia, Tehseen (January 8, 2024). "Unveiling of Large Multimodal Models: Shaping the Landscape of Language Models in 2024". Unite.ai. Retrieved 2024-06-01. <a href="https://www.unite.ai/unveiling-of-large-multimodal-models-shaping-the-landscape-of-language-models-in-2024/" target="_blank">https://www.unite.ai/unveiling-of-large-multimodal-models-shaping-the-landscape-of-language-models-in-2024/</a> <a href="#fnref:5" class="footnote-back-ref">↩</a></li>
<li id="fn:6">Lu, Kevin; Grover, Aditya; Abbeel, Pieter; Mordatch, Igor (2022-06-28). "Frozen Pretrained Transformers as Universal Computation Engines". Proceedings of the AAAI Conference on Artificial Intelligence. 36 (7): 7628–7636. doi:10.1609/aaai.v36i7.20729. ISSN 2374-3468. <a href="https://ojs.aaai.org/index.php/AAAI/article/view/20729" target="_blank">https://ojs.aaai.org/index.php/AAAI/article/view/20729</a> <a href="#fnref:6" class="footnote-back-ref">↩</a></li>
<li id="fn:7">"Vicuna: An Open-Source Chatbot Impressing GPT-4 with 90%* ChatGPT Quality | LMSYS Org". lmsys.org. Retrieved 2024-08-11. <a href="https://lmsys.org/blog/2023-03-30-vicuna" target="_blank">https://lmsys.org/blog/2023-03-30-vicuna</a> <a href="#fnref:7" class="footnote-back-ref">↩</a></li>
<li id="fn:8">Liu, Haotian; Li, Chunyuan; Wu, Qingyang; Lee, Yong Jae (2023-12-15). "Visual Instruction Tuning". Advances in Neural Information Processing Systems. 36: 34892–34916. <a href="https://proceedings.neurips.cc/paper_files/paper/2023/hash/6dcf277ea32ce3288914faf369fe6de0-Abstract-Conference.html" target="_blank">https://proceedings.neurips.cc/paper_files/paper/2023/hash/6dcf277ea32ce3288914faf369fe6de0-Abstract-Conference.html</a> <a href="#fnref:8" class="footnote-back-ref">↩</a></li>
<li id="fn:9">Dosovitskiy, Alexey; Beyer, Lucas; Kolesnikov, Alexander; Weissenborn, Dirk; Zhai, Xiaohua; Unterthiner, Thomas; Dehghani, Mostafa; Minderer, Matthias; Heigold, Georg; Gelly, Sylvain; Uszkoreit, Jakob (2021-06-03). "An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale". arXiv:2010.11929 [cs.CV]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:9" class="footnote-back-ref">↩</a></li>
<li id="fn:10">Gulati, Anmol; Qin, James; Chiu, Chung-Cheng; Parmar, Niki; Zhang, Yu; Yu, Jiahui; Han, Wei; Wang, Shibo; Zhang, Zhengdong; Wu, Yonghui; Pang, Ruoming (2020). "Conformer: Convolution-augmented Transformer for Speech Recognition". arXiv:2005.08100 [eess.AS]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:10" class="footnote-back-ref">↩</a></li>
<li id="fn:11">Radford, Alec; Kim, Jong Wook; Xu, Tao; Brockman, Greg; McLeavey, Christine; Sutskever, Ilya (2022). "Robust Speech Recognition via Large-Scale Weak Supervision". arXiv:2212.04356 [eess.AS]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:11" class="footnote-back-ref">↩</a></li>
<li id="fn:12">Jaegle, Andrew; Gimeno, Felix; Brock, Andrew; Zisserman, Andrew; Vinyals, Oriol; Carreira, Joao (2021-06-22). "Perceiver: General Perception with Iterative Attention". arXiv:2103.03206 [cs.CV]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:12" class="footnote-back-ref">↩</a></li>
<li id="fn:13">Jaegle, Andrew; Borgeaud, Sebastian; Alayrac, Jean-Baptiste; Doersch, Carl; Ionescu, Catalin; Ding, David; Koppula, Skanda; Zoran, Daniel; Brock, Andrew; Shelhamer, Evan; Hénaff, Olivier (2021-08-02). "Perceiver IO: A General Architecture for Structured Inputs & Outputs". arXiv:2107.14795 [cs.LG]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:13" class="footnote-back-ref">↩</a></li>
<li id="fn:14">"Parti: Pathways Autoregressive Text-to-Image Model". sites.research.google. Retrieved 2024-08-09. <a href="https://sites.research.google/parti/" target="_blank">https://sites.research.google/parti/</a> <a href="#fnref:14" class="footnote-back-ref">↩</a></li>
<li id="fn:15">Villegas, Ruben; Babaeizadeh, Mohammad; Kindermans, Pieter-Jan; Moraldo, Hernan; Zhang, Han; Saffar, Mohammad Taghi; Castro, Santiago; Kunze, Julius; Erhan, Dumitru (2022-09-29). "Phenaki: Variable Length Video Generation from Open Domain Textual Descriptions". {{cite journal}}: Cite journal requires |journal= (help) <a href="https://openreview.net/forum?id=vOEXS39nOF" target="_blank">https://openreview.net/forum?id=vOEXS39nOF</a> <a href="#fnref:15" class="footnote-back-ref">↩</a></li>
<li id="fn:16">Chang, Huiwen; Zhang, Han; Barber, Jarred; Maschinot, A. J.; Lezama, Jose; Jiang, Lu; Yang, Ming-Hsuan; Murphy, Kevin; Freeman, William T. (2023-01-02). "Muse: Text-To-Image Generation via Masked Generative Transformers". arXiv:2301.00704 [cs.CV]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:16" class="footnote-back-ref">↩</a></li>
<li id="fn:17">Ramesh, Aditya; Pavlov, Mikhail; Goh, Gabriel; Gray, Scott; Voss, Chelsea; Radford, Alec; Chen, Mark; Sutskever, Ilya (2021-02-26), Zero-Shot Text-to-Image Generation, arXiv:2102.12092 <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:17" class="footnote-back-ref">↩</a></li>
<li id="fn:18">Yu, Jiahui; Xu, Yuanzhong; Koh, Jing Yu; Luong, Thang; Baid, Gunjan; Wang, Zirui; Vasudevan, Vijay; Ku, Alexander; Yang, Yinfei (2022-06-21), Scaling Autoregressive Models for Content-Rich Text-to-Image Generation, arXiv:2206.10789 <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:18" class="footnote-back-ref">↩</a></li>
<li id="fn:19">Chang, Huiwen; Zhang, Han; Barber, Jarred; Maschinot, A. J.; Lezama, Jose; Jiang, Lu; Yang, Ming-Hsuan; Murphy, Kevin; Freeman, William T. (2023-01-02). "Muse: Text-To-Image Generation via Masked Generative Transformers". arXiv:2301.00704 [cs.CV]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:19" class="footnote-back-ref">↩</a></li>
<li id="fn:20">Villegas, Ruben; Babaeizadeh, Mohammad; Kindermans, Pieter-Jan; Moraldo, Hernan; Zhang, Han; Saffar, Mohammad Taghi; Castro, Santiago; Kunze, Julius; Erhan, Dumitru (2022-09-29). "Phenaki: Variable Length Video Generation from Open Domain Textual Descriptions". {{cite journal}}: Cite journal requires |journal= (help) <a href="https://openreview.net/forum?id=vOEXS39nOF" target="_blank">https://openreview.net/forum?id=vOEXS39nOF</a> <a href="#fnref:20" class="footnote-back-ref">↩</a></li>
<li id="fn:21">Kiros, Ryan; Salakhutdinov, Ruslan; Zemel, Rich (2014-06-18). "Multimodal Neural Language Models". Proceedings of the 31st International Conference on Machine Learning. PMLR: 595–603. Archived from the original on 2023-07-02. Retrieved 2023-07-02. <a href="https://proceedings.mlr.press/v32/kiros14.html" target="_blank">https://proceedings.mlr.press/v32/kiros14.html</a> <a href="#fnref:21" class="footnote-back-ref">↩</a></li>
<li id="fn:22">Krizhevsky, Alex; Sutskever, Ilya; Hinton, Geoffrey E (2012). "ImageNet Classification with Deep Convolutional Neural Networks". Advances in Neural Information Processing Systems. 25. Curran Associates, Inc. Archived from the original on 2023-07-02. Retrieved 2023-07-02. <a href="https://proceedings.neurips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html" target="_blank">https://proceedings.neurips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html</a> <a href="#fnref:22" class="footnote-back-ref">↩</a></li>
<li id="fn:23">Antol, Stanislaw; Agrawal, Aishwarya; Lu, Jiasen; Mitchell, Margaret; Batra, Dhruv; Zitnick, C. Lawrence; Parikh, Devi (2015). "VQA: Visual Question Answering". ICCV: 2425–2433. Archived from the original on 2023-07-02. Retrieved 2023-07-02. <a href="https://openaccess.thecvf.com/content_iccv_2015/html/Antol_VQA_Visual_Question_ICCV_2015_paper.html" target="_blank">https://openaccess.thecvf.com/content_iccv_2015/html/Antol_VQA_Visual_Question_ICCV_2015_paper.html</a> <a href="#fnref:23" class="footnote-back-ref">↩</a></li>
<li id="fn:24">Li, Junnan; Li, Dongxu; Savarese, Silvio; Hoi, Steven (2023-01-01). "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models". arXiv:2301.12597 [cs.CV]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:24" class="footnote-back-ref">↩</a></li>
<li id="fn:25">Alayrac, Jean-Baptiste; Donahue, Jeff; Luc, Pauline; Miech, Antoine; Barr, Iain; Hasson, Yana; Lenc, Karel; Mensch, Arthur; Millican, Katherine; Reynolds, Malcolm; Ring, Roman; Rutherford, Eliza; Cabi, Serkan; Han, Tengda; Gong, Zhitao (2022-12-06). "Flamingo: a Visual Language Model for Few-Shot Learning". Advances in Neural Information Processing Systems. 35: 23716–23736. arXiv:2204.14198. Archived from the original on 2023-07-02. Retrieved 2023-07-02. <a href="https://proceedings.neurips.cc/paper_files/paper/2022/hash/960a172bc7fbf0177ccccbb411a7d800-Abstract-Conference.html" target="_blank">https://proceedings.neurips.cc/paper_files/paper/2022/hash/960a172bc7fbf0177ccccbb411a7d800-Abstract-Conference.html</a> <a href="#fnref:25" class="footnote-back-ref">↩</a></li>
<li id="fn:26">Driess, Danny; Xia, Fei; Sajjadi, Mehdi S. M.; Lynch, Corey; Chowdhery, Aakanksha; Ichter, Brian; Wahid, Ayzaan; Tompson, Jonathan; Vuong, Quan; Yu, Tianhe; Huang, Wenlong; Chebotar, Yevgen; Sermanet, Pierre; Duckworth, Daniel; Levine, Sergey (2023-03-01). "PaLM-E: An Embodied Multimodal Language Model". arXiv:2303.03378 [cs.LG]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:26" class="footnote-back-ref">↩</a></li>
<li id="fn:27">Liu, Haotian; Li, Chunyuan; Wu, Qingyang; Lee, Yong Jae (2023-04-01). "Visual Instruction Tuning". arXiv:2304.08485 [cs.CV]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:27" class="footnote-back-ref">↩</a></li>
<li id="fn:28">Zhang, Hang; Li, Xin; Bing, Lidong (2023-06-01). "Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understanding". arXiv:2306.02858 [cs.CL]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:28" class="footnote-back-ref">↩</a></li>
<li id="fn:29">OpenAI (2023-03-27). "GPT-4 Technical Report". arXiv:2303.08774 [cs.CL]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:29" class="footnote-back-ref">↩</a></li>
<li id="fn:30">OpenAI (September 25, 2023). "GPT-4V(ision) System Card" (PDF). <a href="https://cdn.openai.com/papers/GPTV_System_Card.pdf" target="_blank">https://cdn.openai.com/papers/GPTV_System_Card.pdf</a> <a href="#fnref:30" class="footnote-back-ref">↩</a></li>
<li id="fn:31">Pichai, Sundar (10 May 2023), Google Keynote (Google I/O '23), timestamp 15:31, retrieved 2023-07-02 <a href="https://www.youtube.com/watch?v=cNfINi5CNbY&t=931s" target="_blank">https://www.youtube.com/watch?v=cNfINi5CNbY&t=931s</a> <a href="#fnref:31" class="footnote-back-ref">↩</a></li>
<li id="fn:32">Wiggers, Kyle (11 September 2024). "Mistral releases Pixtral 12B, its first multimodal model". TechCrunch. Retrieved 14 September 2024. <a href="https://techcrunch.com/2024/09/11/mistral-releases-pixtral-its-first-multimodal-model/?utm_medium=aisecret.us&utm_source=aisecret.us&utm_campaign=aisecret.us" target="_blank">https://techcrunch.com/2024/09/11/mistral-releases-pixtral-its-first-multimodal-model/?utm_medium=aisecret.us&utm_source=aisecret.us&utm_campaign=aisecret.us</a> <a href="#fnref:32" class="footnote-back-ref">↩</a></li>
<li id="fn:33">Dey, Victor (2021-09-03). "Beginners Guide to Boltzmann Machine". Analytics India Magazine. Retrieved 2024-03-02. <a href="https://analyticsindiamag.com/beginners-guide-to-boltzmann-machines/" target="_blank">https://analyticsindiamag.com/beginners-guide-to-boltzmann-machines/</a> <a href="#fnref:33" class="footnote-back-ref">↩</a></li>
<li id="fn:34">"Multimodal Learning with Deep Boltzmann Machine" (PDF). 2014. Archived (PDF) from the original on 2015-06-21. Retrieved 2015-06-14. <a href="http://www.jmlr.org/papers/volume15/srivastava14b/srivastava14b.pdf" target="_blank">http://www.jmlr.org/papers/volume15/srivastava14b/srivastava14b.pdf</a> <a href="#fnref:34" class="footnote-back-ref">↩</a></li>
<li id="fn:35">Hendriksen, Mariya; Vakulenko, Svitlana; Kuiper, Ernst; de Rijke, Maarten (2023). "Scene-centric vs. Object-centric Image-Text Cross-modal Retrieval: A Reproducibility Study". arXiv:2301.05174 [cs.CV]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:35" class="footnote-back-ref">↩</a></li>
<li id="fn:36">Quach, Katyanna. "Harvard boffins build multimodal AI system to predict cancer". The Register. Archived from the original on 20 September 2022. Retrieved 16 September 2022. <a href="https://www.theregister.com/2022/08/09/ai_cancer_multimodal/" target="_blank">https://www.theregister.com/2022/08/09/ai_cancer_multimodal/</a> <a href="#fnref:36" class="footnote-back-ref">↩</a></li>
<li id="fn:37">Chen, Richard J.; Lu, Ming Y.; Williamson, Drew F. K.; Chen, Tiffany Y.; Lipkova, Jana; Noor, Zahra; Shaban, Muhammad; Shady, Maha; Williams, Mane; Joo, Bumjin; Mahmood, Faisal (8 August 2022). "Pan-cancer integrative histology-genomic analysis via multimodal deep learning". Cancer Cell. 40 (8): 865–878.e6. doi:10.1016/j.ccell.2022.07.004. ISSN 1535-6108. PMC 10397370. PMID 35944502. S2CID 251456162.
Teaching hospital press release: "New AI technology integrates multiple data types to predict cancer outcomes". Brigham and Women's Hospital via medicalxpress.com. Archived from the original on 20 September 2022. Retrieved 18 September 2022.
 <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10397370" target="_blank">https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10397370</a> <a href="#fnref:37" class="footnote-back-ref">↩</a></li>
<li id="fn:38">Shi, Yuge; Siddharth, N.; Paige, Brooks; Torr, Philip HS (2019). "Variational Mixture-of-Experts Autoencoders for Multi-Modal Deep Generative Models". arXiv:1911.03393 [cs.LG]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:38" class="footnote-back-ref">↩</a></li>
<li id="fn:39">Shi, Yuge; Siddharth, N.; Paige, Brooks; Torr, Philip HS (2019). "Variational Mixture-of-Experts Autoencoders for Multi-Modal Deep Generative Models". arXiv:1911.03393 [cs.LG]. <a href="/wiki/ArXiv_(identifier)" target="_blank">/wiki/ArXiv_(identifier)</a> <a href="#fnref:39" class="footnote-back-ref">↩</a></li>
</ol>

Multimodal learning open-in-new

Multimodal learning