Sony Research Award Program 2026 - Focused Research Award

Registration Deadline:

Event Contact:

  • Swapneeta Date

Explore other Industry Opportunities:

Overview

  • Funding up to $150,000 USD per year for accepted proposals
  • One-year research period with potential renewal for subsequent years
  • Sponsored research agreement required between Sony and PI's institution
  • Direct collaboration with Sony researchers on focused research topics
  • Single all-inclusive payment covering research costs, overhead, and all associated fees
  • Awards paid in USD or respective local currency depending on country
  • Results for noncommercial use by Sony; IP rights specified in research agreement

 

Priority Areas

  • Internal Mechanisms of Multimodal Generative Models for Content Creation: Research on the internal mechanisms of LLMs, MLLMs, speech/audio models, music models, and video generation systems for multimodal content creation. Focus areas include mechanistic interpretability of multimodal generation; causal attribution inside generative models; internal representations of source, authorship, and provenance; knowledge externalization and controllable data usage; representations of style, identity, and modality-specific structure; cross-modal binding, grounding, and inconsistency; intervention, editing, and controllability of internal mechanisms.
  • Anime-Style Expression Exaggeration – Learning Artistic Deformation Priors: Research on learning artistic deformation priors from anime data to automate or assist exaggeration processes for anime and stylized characters. Focus areas include learning artistic exaggeration priors from 2D anime/stylized animation datasets; mapping subtle photoreal performance to exaggerated stylized output; per-style or per-character exaggeration models; integration with 2D and 3D anime character rigs; handling non-physical visual elements; evaluation of artistic faithfulness and animator usability.
  • Knowledge Representation and Reasoning for Graph-Based Modeling: Research on how ontologies and knowledge representation methods support explainable, reusable, and adaptable graph-based modeling for user profile generation, personalized recommendation, micro-region segmentation, and explainability. Focus areas include ontology-based methods to enrich graph representations; formal representation of subjective user signals; graph enhancement through hidden links and semantic structure; ontology-driven reasoning for explainability; knowledge consistency validation; cross-domain mappings; transparency and maintainability improvements.
  • Psychology-Informed User Modeling and Intervention Design for Content Streaming Experiences: Research on methods that infer latent user constructs (values, needs, experiential orientation) from behavioral data to support audience understanding and intervention design beyond recommendation accuracy. Focus areas include psychology-informed user and content modeling grounded in established theory; insight extraction using psychological/sociological knowledge; identification of user attributes governing intervention responsiveness; evaluation methodology beyond offline accuracy metrics including uplift evaluation.
  • Post-Training of SpeechLMs for Anime: Research on adapting speech LMs to capture anime-specific stylistic, emotional, and cultural nuances for localization and character-based voice interaction. Focus areas include improving culturally-aware translation; production-quality localization; controllable conversational systems; automated data annotation pipeline for post-training.
  • Innovative Visual Technology Powered by AI: Research in image and video technologies based on machine learning to advance 2D, 3D, and 4D capabilities. Focus areas include generative AI (image to photo-realistic video, video to video, style transfer, modal transfer, neural rendering, high controllability); multimodal content generation based on VLM; physical AI (VLA) connecting virtual and real world; novel 3D/4D model generation/reconstruction (NeRF, gaussian splatting, world models); perceptual metrics for image quality prediction based on VLM; explainable and controllable AI for visual content processing and generation.
  • AI-Based Digital Human Content Creation for Sports Science and Sports Entertainment: Research on AI and ML-based techniques for data-driven digital human 3D model creation and animation for sports science (biomechanics, physiology, coaching, performance analysis, injury prevention) and sports entertainment (immersive 3D/4D visualization). Focus areas include high-fidelity biomechanical measurement and diagnostics using digital human CG models from 2D images/videos; high-fidelity immersive sports visualization delivering emotional engagement and novel fan experiences.
  • Provenance Proofs for AI-Generated Derivative Content: Research on cryptographic and machine-learning techniques to create tamper-resistant provenance chains for AI-generated content tracking across multiple generations of modification. Focus areas include provenance representation models; cryptographic provenance infrastructure; derivative content attribution; AI-native watermarking and fingerprinting; provenance across multi-step generation pipelines; privacy-preserving verification; scalable provenance storage/retrieval; trust, authenticity, and misinformation detection; copyright and creator compensation frameworks; benchmarking and evaluation.
  • Long-Term Memory for Sequential Decision-Making in Game Video Understanding: Research on scalable memory architectures and long-context learning methods enabling AI systems to retain, retrieve, and reason over information across extended gameplay experiences. Focus areas include long-term memory representations for gameplay video and multimodal data; memory acquisition, storage, compression, retrieval mechanisms; hierarchical/multi-resolution memory; attention-based memory with learned policies; knowledge-grounded and hypothesis-driven gameplay understanding; controllable and contextual memory; RL-style memory management; state-space and linear-attention approaches; action and state-grounded memory; strategies for small model memory.
  • Cloud Creation Everywhere: Research on cloud technology to enable social experiences linking real and virtual worlds, supporting creativity outside the home. Focus areas include increasing engagement with creative experiences outside home; integrating game experiences with social media and social expression; connecting people to online communities asynchronously; transforming real-world spaces into virtual cloud spaces; novel peer-to-peer experiences; cloud tools and services (traditional cloud servers or peer-to-peer clouds); mobile phones, tablets, cloud-connected peripherals, AR/VR; connecting to existing social media services for interoperability.
  • Design and Evaluation of an Intermediate Interaction Layer for Integrating Diverse Input/Output Environments: Research on abstraction-layer technologies inferring user operational intent from multimodal inputs and physiological states, automatically mapping to application commands and adaptive feedback. Focus areas include integrating multimodal inputs (gaze, voice, gesture, biometric data) to infer intent; intermediate abstraction layers translating arbitrary I/O devices to unified control schemes; adaptive feedback mechanisms dynamically generating responses; multimodal communication across heterogeneous devices including nonverbal cue translation; prototyping and evaluation in games, XR, and remote communication platforms.

 

Eligibility & Requirements

Principal Investigator Requirements:

  • Must be affiliated with eligible university, educational institution, or governmental/non-profit research institute
  • Must be full-time professor (full professor, associate professor, or assistant professor) or researcher
  • Must be eligible to supervise Ph.D. students at their institution
  • For UK: Includes Senior Lecturers, Principal Lecturers, Lecturers, and Readers eligible to supervise Ph.D. students
  • Adjunct professors and adjunct researchers are not eligible
  • Professor Emeritus is not eligible

Geographic Eligibility (PI's institution must be located in):

  • United States, Canada, India
  • European countries: Austria, Belgium, Bulgaria, Croatia, Republic of Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland, United Kingdom

Co-Principal Investigators:

  • Co-PIs allowed but must be from same institution as PI
  • All Co-PIs must meet same eligibility criteria as PI
  • Co-PIs required to sign program documents
  • Only one award made to PI and PI's institution if selected

Multiple Submissions:

  • PI or different PIs from same institution may submit multiple proposals for different research topics
  • Each PI limited to one proposal per Focused Research Theme
  • No identical proposals from same PI
  • Unlimited proposals from same university with different PIs for same topic

Proposal Requirements:

  • Select one Focused Research Theme from provided list
  • Must describe differentiation from current state-of-the-art
  • Proposal limited to 11 pages total: 10-page maximum proposal (including all references) plus 1-page budget summary
  • PDF format, under 16 MB file size
  • Minimum 10-point font size
  • Must include: Title, abstract, methods, goals/milestones, references, selected Focused Research Theme
  • Must include best contact email and phone number (with country code) for PI
  • Separate CV file for PI (and Co-PIs if applicable) required, no page limit but under 16 MB
  • Proposals for one-year period only; no multi-year proposals accepted
  • Budget must be all-inclusive of research costs, overhead, and all fees (maximum $150K USD)

Institutional Requirements:

  • Sponsored research agreement required between Sony and PI's institution before funding available
  • No confidential or proprietary information in submitted proposals (treated as non-confidential and non-proprietary)
  • PI and institution must have full authority to grant noncommercial use rights to Sony for any background IP
  • No third-party involvement allowed in sponsored research agreement

Deliverables:

  • Three quarterly reports minimum
  • Final research summary report
  • Responses to progress questionnaire
  • Other deliverables may be specified based on project nature

Evaluation Criteria:

  • Research innovation and differentiation from state-of-the-art
  • Alignment with selected Focused Research Theme and Sony's collaborative research goals
  • Technical approach and methodology
  • Potential for collaborative research with Sony scientists
  • Feasibility and timeline
  • Qualifications of research team
  • Potential for renewal and extended collaboration

 

Timeline

  • July 15, 2026: Application period opens
  • September 15, 2026, Submission deadline
  • Around March 2027: Award winners notified
  • After March 2027: Sponsored research agreement negotiation and execution with selected proposals
  • Following agreement execution: Funding available via single payment
  • One year from funding start: Initial research period
  • Following year one: Possible extension requiring separate discussion and award for subsequent years

 

How to Submit